OPINION

BYBEE, Circuit Judge:

Section 1 of the Sherman Antitrust Act of 1890, 15 U.S.C. § 1, prohibits "[e]very contract, combination ..., or conspiracy, in restraint of trade or commerce." For more than a century, the National Collegiate Athletic Association (NCAA) has prescribed rules governing the eligibility of athletes at its more than 1,000 member colleges and universities. Those rules prohibit student-athletes from being paid for the use of their names, images, and likenesses (NILs). The question presented in this momentous case is whether the NCAA's rules are subject to the antitrust laws and, if so, whether they are an unlawful restraint of trade.

After a bench trial and in a thorough opinion, the district court concluded that the NCAA's compensation rules were an unlawful restraint of trade. It then enjoined the NCAA from prohibiting its member schools from giving student-athletes scholarships up to the full cost of attendance at their respective schools and up to $5,000 per year in deferred compensation, to be held in trust for student-athletes until after they leave college. As far as we are aware, the district court's decision is the first by any federal court to hold that any aspect of the NCAA's amateurism rules violate the antitrust laws, let alone to mandate by injunction that the NCAA change its practices.

We conclude that the district court's decision was largely correct. Although we agree with the Supreme Court and our sister circuits that many of the NCAA's amateurism rules are likely to be procompetitive, we hold that those rules are not exempt from antitrust scrutiny; rather, they must be analyzed under the Rule of Reason. Applying the Rule of Reason, we conclude that the district court correctly identified one proper alternative to the current NCAA compensation rules — i.e., allowing NCAA members to give scholarships up to the full cost of attendance — but that the district court's other remedy, allowing students to be paid cash compensation of up to $5,000 per year, was erroneous. We therefore affirm in part and reverse in part.

I

A. The NCAA

American colleges and universities have been competing in sports for nearly 150 years: the era of intercollegiate athletics began, by most accounts, on November 6, 1869, when Rutgers and Princeton met in the first college football game in American history — a game more akin to soccer than to modern American football, played with "25 men to a side." Joseph N. Crowley, In the Arena: The NCAA's First Century 2 (2006), available at https://www.ncaapublications.com/p-4039-in-the-arena-the-ncaas-first-century.aspx. College football quickly grew in popularity over the next few decades.

Fin de siècle college football was a rough game. Serious injuries were common, and it was not unheard of for players to be killed during games. Schools were also free to hire nonstudent ringers to compete on their teams or to purchase players away from other schools. By 1905, these and other problems had brought college football to a moment of crisis, and President Theodore Roosevelt convened a conference at the White House to address the issue of injuries in college football. Later that year, the presidents of 62 colleges and universities founded the Intercollegiate Athletic Association to create uniform rules for college football. In 1910, the IAA changed its name to the National Collegiate Athletic Association (NCAA), and it has kept that name to this day.

The NCAA has grown to include some 1,100 member schools, organized into three divisions: Division I, Division II, and Division III. Division I schools are those with the largest athletic programs — schools must sponsor at least fourteen varsity sports teams to qualify for Division I — and they provide the most financial aid to student-athletes. Division I has about 350 members.

For football competition only, Division I's membership is divided into two subdivisions: the Football Bowl Subdivision (FBS) and the Football Championship Subdivision (FCS). FBS schools are permitted to offer more full scholarships to their football players and, as a result, the level of competition is generally higher in FBS than in FCS. FBS consists of about 120 of the nation's premier college football schools.

B. The Amateurism Rules

One of the NCAA's earliest reforms of intercollegiate sports was a requirement that the participants be amateurs. President C.A. Richmond of Union College commented in 1921 that the competition among colleges to acquire the best players had come to resemble "the contest in dreadnoughts" that had led to World War I,1 and the NCAA sought to curb this problem by restricting eligibility for college sports to athletes who received no compensation whatsoever.2 But the NCAA, still a voluntary organization, lacked the ability to enforce this requirement effectively, and schools continued to pay their athletes under the table in a variety of creative ways; a 1929 study found that 81 out of 112 schools surveyed provided some sort of improper inducement to their athletes.

The NCAA began to strengthen its enforcement capabilities in 1948, when it adopted what became known as the "Sanity Code" — a set of rules that prohibited schools from giving athletes financial aid that was based on athletic ability and not available to ordinary students. See Daniel E. Lazaroff, The NCAA in Its Second Century: Defender of Amateurism or Antitrust Recidivist?, 86 Or. L.Rev. 329, 333 (2007). The Sanity Code also created a new "compliance mechanism" to enforce the NCAA's rules — "a Compliance Committee that could terminate an institution's NCAA membership." Id.

In 1956, the NCAA departed from the Sanity Code's approach to financial aid by changing its rules to permit its members, for the first time, to give student-athletes scholarships based on athletic ability. These scholarships were capped at the amount of a full "grant in aid," defined as the total cost of "tuition and fees, room and board, and required course-related books." Student-athletes were prohibited from receiving any "financial aid based on athletics ability" in excess of the value of a grant-in-aid, on pain of losing their eligibility for collegiate athletics. Student-athletes could seek additional financial aid not related to their athletic skills; if they chose to do this, the total amount of athletic and nonathletic financial aid they received could not exceed the "cost of attendance" at their respective schools.3

In August 2014, the NCAA announced it would allow athletic conferences to authorize their member schools to increase scholarships up to the full cost of attendance. The 80 member schools of the five largest athletic conferences in the country voted in January 2015 to take that step, and the scholarship cap at those schools is now at the full cost of attendance. Marc Tracy, Top Conferences to Allow Aid for Athletes' Full Bills, N.Y. Times, Jan. 18, 2015, at SP8.

In addition to its financial aid rules, the NCAA has adopted numerous other amateurism rules that limit student-athletes' compensation and their interactions with professional sports leagues. An athlete can lose his amateur status, for example, if he signs a contract with a professional team, enters a professional league's player draft, or hires an agent. And, most importantly, an athlete is prohibited — with few exceptions — from receiving any "pay" based on his athletic ability, whether from boosters, companies seeking endorsements, or would-be licensors of the athlete's name, image, and likeness (NIL).

C. The O'Bannon and Keller Litigation

In 2008, Ed O'Bannon, a former All-American basketball player at UCLA, visited a friend's house, where his friend's son told O'Bannon that he was depicted in a college basketball video game produced by Electronic Arts (EA), a software company that produced video games based on college football and men's basketball from the late 1990s until around 2013. The friend's son turned on the video game, and O'Bannon saw an avatar of himself — a virtual player who visually resembled O'Bannon, played for UCLA, and wore O'Bannon's jersey number, 31. O'Bannon had never consented to the use of his likeness in the video game, and he had not been compensated for it.

In 2009, O'Bannon sued the NCAA and the Collegiate Licensing Company (CLC), the entity which licenses the trademarks of the NCAA and a number of its member schools for commercial use, in federal court. The gravamen of O'Bannon's complaint was that the NCAA's amateurism rules, insofar as they prevented student-athletes from being compensated for the use of their NILs, were an illegal restraint of trade under Section 1 of the Sherman Act, 15 U.S.C. § 1.

Around the same time, Sam Keller, the former starting quarterback for the Arizona State University and University of Nebraska football teams, separately brought suit against the NCAA, CLC, and EA. Keller alleged that EA had impermissibly used student-athletes' NILs in its video games and that the NCAA and CLC had wrongfully turned a blind eye to EA's misappropriation of these NILs. The complaint stated a claim under Indiana's and California's right of publicity statutes, as well as a number of common-law claims.

The two cases were consolidated during pretrial proceedings. The defendants moved to dismiss Keller's right-of-publicity claims on First Amendment grounds. The district court denied the motion to dismiss, and we affirmed that decision, holding that "[u]nder California's transformative use defense, EA's use of the likenesses of college athletes like Samuel Keller in its video games is not, as a matter of law, protected by the First Amendment." In re NCAA Student-Athlete Name & Likeness Licensing Litig. ("Keller"), 724 F.3d 1268, 1284 (9th Cir.2013).

In November 2013, the district court granted the plaintiffs' motion for class certification. The court held that certification of a damages class under Rule 23(b)(3) was inappropriate, but it certified the following class under Rule 23(b)(2) for injunctive and declaratory relief:

After class certification was granted, the plaintiffs voluntarily dismissed their damages claims with prejudice. The plaintiffs also settled their claims against EA and CLC, and the district court preliminarily approved the settlement. O'Bannon and Keller were deconsolidated, and in June 2014, the antitrust claims against the NCAA at issue in O'Bannon went to a bench trial before the district court.

D. The District Court's Decision

After a fourteen-day bench trial, the district court entered judgment for the plaintiffs, concluding that the NCAA's rules prohibiting student-athletes from receiving compensation for their NILs violate Section 1 of the Sherman Act. O'Bannon v. NCAA, 7 F.Supp.3d 955 (N.D.Cal.2014).

1. The Markets

The court began by identifying the markets in which the NCAA allegedly restrained trade. It identified two markets that were potentially affected by the challenged NCAA rules.

a. The college education market

First, the court found that there is a "college education market" in which FBS football and Division I basketball schools compete to recruit the best high school players by offering them "unique bundles of goods and services" that include not only scholarships but also coaching, athletic facilities, and the opportunity to face high-quality athletic competition. Id. at 965-66. The court found that very few athletes talented enough to play FBS football or Division I basketball opt not to attend an FBS/Division I school; hardly any choose to attend an FCS, Division II, or Division III school or to compete in minor or foreign professional sports leagues, and athletes are not allowed to join either the NFL or the NBA directly from high school.5 Id. at 966. Thus, the court concluded, the market specifically for FBS football and Division I basketball scholarships is cognizable under the antitrust laws because "there are no professional [or college] football or basketball leagues capable of supplying a substitute for the bundle of goods and services that FBS football and Division I basketball schools provide." Id. at 968.

b. The group licensing market

The second market identified by the district court was a "group licensing market" in which, but for the NCAA's compensation rules, college football and basketball athletes would be able to sell group licenses for the use of their NILs. Id. The court broke this "group licensing market" down into three submarkets in which players' NILs could be profitably licensed: (1) live game telecasts, (2) sports video games, and (3) game rebroadcasts, advertisements, and other archival footage.6 Id. With respect to live game telecasts, the court noted that the TV networks that broadcast live college football and basketball games "often seek to acquire the rights to use" the players' NILs, which the court concluded "demonstrate[s] that there is a demand for these rights" on the networks' part. Id. at 968-69. With respect to video games, the court found that the use of NILs increased the attractiveness of college sports video games to consumers, creating a demand for players' NILs.7 Id. at 970. And with respect to archival footage, the court noted that the NCAA had licensed footage of student-athletes — including current athletes — to a third-party licensing company, T3Media, proving that there is demand for such footage. Id. at 970-71.

2. The Rule of Reason

Having concluded that the NCAA's compensation rules potentially restrained competition in these two markets, the court proceeded to analyze the legality of the challenged NCAA rules with respect to those markets, applying the Rule of Reason. Id. at 984-1009. The district court found that the NCAA's rules have an anticompetitive effect in the college education market but not in the group licensing market. It then concluded that the rules serve procompetitive purposes. Finally, it determined that the procompetitive purposes of the rules could be achieved by less restrictive alternative restraints and that the current rules were therefore unlawful.

a. Anticompetitive effects

At the first step of the Rule of Reason, the court found that the NCAA's rules have an anticompetitive effect on the college education market. Were it not for those rules, the court explained, schools would compete with each other by offering recruits compensation exceeding the cost of attendance, which would "effectively lower the price that the recruits must pay for the combination of educational and athletic opportunities that the schools provide." Id. at 972. The rules prohibiting compensation for the use of student-athletes' NILs are thus a price-fixing agreement: recruits pay for the bundles of services provided by colleges with their labor and their NILs, but the "sellers" of these bundles — the colleges — collectively "agree to value [NILs] at zero." Id. at 973. Under this theory, colleges and universities behave as a cartel — a group of sellers who have colluded to fix the price of their product.

The court found in the alternative that the college education market can be thought of as a market in which student-athletes are sellers rather than buyers and the schools are purchasers of athletic services. In the court's alternative view, the college education market is a monopsony — a market in which there is only one buyer (the NCAA schools, acting collectively) for a particular good or service (the labor and NIL rights of student-athletes), and the colleges' agreement not to pay anything to purchase recruits' NILs causes harm to competition. Id. at 973, 991.

By contrast, the court found that the NCAA's rules do not have an anticompetitive effect on any of the submarkets of the group licensing market. The court explained that although these submarkets exist, there would be no competition in any of them if the challenged NCAA rules were abolished. The court reasoned that the value of an NIL license to a live game broadcaster or a video game company would depend on the licensee's acquiring every other NIL license that was available. A live game broadcaster, for example, would need to acquire a license from every team or player whose games it might telecast. Similarly, a video game producer would want to acquire NIL rights for all of the teams it needed to include in the game. Given these requirements, the court deemed it highly unlikely that groups of student-athletes would compete with each other to sell their NIL rights; on the contrary, they would have an incentive to cooperate to make sure that the package of NIL rights sold to buyers was as complete as possible. Id. at 993-98. With respect to archival footage, meanwhile, the court found that the NCAA's licensing arrangement with T3Media did not deprive student-athletes of any compensation they might otherwise receive because T3Media is prohibited from licensing footage of current athletes and must obtain the consent of any former athlete whose NIL appears in its footage. Id. at 998-99.

b. Procompetitive purposes

At the second step of the Rule of Reason, the NCAA proffered four procompetitive purposes for its rules prohibiting student-athletes from receiving compensation for the use of their NILs: (1) preserving "amateurism" in college sports; (2) promoting competitive balance in FBS football and Division I basketball; (3) integrating academics and athletics; and (4) increasing output in the college education market. Id. at 999. The court accepted the first and third justifications in part while rejecting the others.

(1) Amateurism. The NCAA argued to the district court that restrictions on student-athlete compensation are "necessary to preserve the amateur tradition and identity of college sports." Id. It contended that amateurism had been one of the NCAA's core principles since its founding and that amateurism is a key driver of college sports' popularity with consumers and fans. Id. at 999-1000.

The district court rejected the NCAA's contention that it had a "longstanding commitment to amateurism," concluding instead that the NCAA's definition of amateurism was "malleable," changing frequently over time in "significant and contradictory ways." Id. at 1000. The court suggested that, even today, the NCAA's definition of amateurism is inconsistent: although players generally cannot receive compensation other than scholarships, tennis players are permitted to accept up to $10,000 in prize money before enrolling in college, and student-athletes are permitted to accept Pell grants even when those grants raise their total financial aid package above their cost of attendance. Id. It thus concluded that amateurism was not, in fact, a "core principle[]" of the NCAA. Id.

The district court was not persuaded that amateurism is the primary driver of consumer demand for college sports — but it did find that amateurism serves some procompetitive purposes. The court first concluded that consumers are primarily attracted to college sports for reasons unrelated to amateurism, such as loyalty to their alma mater or affinity for the school in their region of the country. Id. at 977-78. It also found much of the NCAA's evidence about amateurism unreliable. For example, the NCAA provided a survey conducted by Dr. J. Michael Dennis, a "survey research expert," which purported to show that Americans "generally oppose[] the idea of paying college football and basketball players." Id. at 975. The court deemed the Dennis survey "unpersuasive" for a couple reasons, one of which was that it believed the survey's initial question skewed the results by priming respondents to think about illicit payments to student-athletes rather than the possibility of allowing athletes to be paid. Id.

But the district court ultimately found that the NCAA's "current understanding of amateurism" plays some role in preserving "the popularity of the NCAA's product." Id. at 1005. It found that the NCAA's current rules serve a procompetitive benefit by promoting this understanding of amateurism, which in turn helps preserve consumer demand for college sports.

(2) Competitive Balance. The NCAA argued before the district court that restricting compensation to student-athletes helps level the playing field between FBS and Division I schools in recruiting, thereby maintaining competitive balance among those schools' football and basketball teams. Id. at 1001-02.

The district court acknowledged that promoting competitive balance could be a valid procompetitive purpose under the antitrust laws, but it concluded that the challenged NCAA rules do not promote competitive balance. The court noted that numerous economists have studied the NCAA over the years and that "nearly all" of them have concluded that the NCAA's compensation rules do not promote competitive balance. Id. at 978. The court also explained that although the NCAA forbids its member schools to pay student-athletes anything beyond a fixed scholarship, it allows schools to spend as much as they like on other aspects of their athletic programs, such as coaching, facilities, and the like, which "negate[s] whatever equalizing effect the NCAA's restraints on student-athlete compensation might have once had." Id. at 1002. The court concluded that competitive balance was thus not a viable justification for restricting student-athlete compensation.

(3) Integrating Academics and Athletics. The NCAA's third procompetitive justification for its restraints on studentathlete compensation was that these restraints integrate academics and athletics and thereby "improve the quality of educational services provided to student-athletes." Id. According to the NCAA, student-athletes derive long-term benefits from participating fully in academic life at their schools, which the compensation rules encourage them to do. Id. at 979-80.

The district court allowed that this was a viable procompetitive justification for the NCAA's regulating the college education market, but it concluded that most of the benefits of academic and athletic "integration" are not the result of the NCAA's rules restricting compensation. Rather, these benefits are achieved by other NCAA rules — such as those requiring student-athletes to attend class, prohibiting athletes-only dorms, and forbidding student-athletes to practice more than a certain number of hours per week. Id. at 980. The court explained that the only way in which the compensation rules might facilitate the integration of athletics and academics is that, by prohibiting student-athletes from being paid large sums of money not available to ordinary students, the rules prevent the creation of a social "wedge" between student-athletes and the rest of the student body. Id. at 980, 1003. It held, however, that even though the avoidance of such a "wedge" is a legitimate procompetitive goal, it does not justify a total, "sweeping prohibition" on paying student-athletes for the use of their NILs. Id. at 1003.

(4) Increasing Output. The fourth and final procompetitive justification alleged by the NCAA was that the restraints on student-athlete compensation "increase output" in the college education market by increasing the available opportunities for students to play FBS football or Division I basketball. Id. at 1003-04. The NCAA contended that its rules accomplish this goal by attracting schools with a philosophical commitment to amateurism to compete in Division I and by enabling schools to compete in Division I that otherwise could not afford to do so. Id. at 1004.

The district court rejected this justification. The court found the idea that schools join Division I because of a philosophical commitment to amateurism "implausible," noting that some major-conference schools had lobbied to change the NCAA's scholarship rules to raise compensation limits. Id. at 981. The court also explained that schools in FCS, Division II, and Division III are subject to the same amateurism rules as Division I schools, making it unlikely that schools choose to join Division I because of the amateurism rules. Id.

The court likewise found no support in the record for the notion that the NCAA's compensation rules enable more schools to compete in Division I. The court found that, because Division I schools do not share revenue, there is no reason to believe that the cost savings from not paying student-athletes are being used to fund additional scholarships at low-revenue schools or to enable those schools to join Division I. Id. at 1004. The court also noted that the plaintiffs were not seeking to require that all schools pay their student-athletes; rather, they sought an injunction permitting schools to do so. Schools that could not afford to pay their student-athletes would not need to do so if the plaintiffs prevailed and would therefore not be driven out of Division I by a ruling in the plaintiffs' favor. Id.

c. Less restrictive alternatives

Having found that the NCAA had presented two procompetitive justifications for "circumscribed" limits on student-athlete compensation — i.e., increasing consumer demand for college sports and preventing the formation of a "wedge" between student-athletes and other students — the court proceeded to the third and final step of the Rule of Reason, where it considered whether there were means of achieving the NCAA's procompetitive purposes that were "substantially less restrictive" than a total ban on compensating student-athletes for use of their NILs. Id. at 1004-05.

The court held that the plaintiffs had identified two legitimate, less restrictive alternatives to the current NCAA rules: (1) allowing schools to award stipends to student-athletes up to the full cost of attendance, thereby making up for any "shortfall" in their grants-in-aid, and (2) permitting schools to hold a portion of their licensing revenues in trust, to be distributed to student-athletes in equal shares after they leave college.8 Id. at 1005-06. The court determined that neither of these alternatives to the total ban on NIL compensation would undermine the NCAA's procompetitive purposes. The court also held that it would be permissible for the NCAA to prohibit schools from funding these stipends or trusts with anything other than revenue derived from the use of players' NILs. Id. at 1005.

After entering judgment for the plaintiffs on their antitrust claims, the district court permanently enjoined the NCAA from prohibiting its member schools from (1) compensating FBS football and Division I men's basketball players for the use of their NILs by awarding them grants-in-aid up to the full cost of attendance at their respective schools, or (2) paying up to $5,000 per year in deferred compensation to FBS football and Division I men's basketball players for the use of their NILs, through trust funds distributable after they leave school. The NCAA timely appealed, and we have jurisdiction under 28 U.S.C. § 1291.

II

We review the district court's findings of fact after the bench trial for clear error and review the district court's conclusions of law de novo. FTC v. BurnLounge, Inc., 753 F.3d 878, 883 (9th Cir. 2014). Our clear-error review of the district court's findings of fact is "deferential"; "we will accept the district court's findings of fact unless we are left with the definite and firm conviction that a mistake has been committed." Id. (alteration and internal quotation marks omitted).

III

On appeal, the NCAA contends that the plaintiffs' Sherman Act claim fails on the merits, but it also argues that we are precluded altogether from reaching the merits, for three independent reasons: (1) The Supreme Court held in NCAA v. Board of Regents of the University of Oklahoma, 468 U.S. 85, 104 S.Ct. 2948, 82 L.Ed.2d 70 (1984), that the NCAA's amateurism rules are "valid as a matter of law"; (2) the compensation rules at issue here are not covered by the Sherman Act at all because they do not regulate commercial activity; and (3) the plaintiffs have no standing to sue under the Sherman Act because they have not suffered "antitrust injury." We find none of these three arguments persuasive.

A. Board of Regents Did Not Declare the NCAA's Amateurism Rules "Valid as a Matter of Law"

We consider, first, the NCAA's claim that, under Board of Regents, all NCAA amateurism rules are "valid as a matter of law."

Board of Regents concerned the NCAA's then-prevailing rules for televising college football games. The rules allowed television networks to negotiate directly with schools and conferences for the right to televise games, but they imposed caps on the total number of games that could be broadcast on television each year and the number of games that any particular school could televise. Id. at 91-94, 104 S.Ct. 2948. The University of Oklahoma and the University of Georgia challenged this regime as an illegal restraint of trade under Section 1.

The Court observed that the television rules resembled two kinds of agreements that are ordinarily considered per se unlawful when made among horizontal competitors in the same market: a price-fixing agreement (in that the rules set a minimum aggregate price that the television networks were required to pay the NCAA's members) and an output-restriction agreement (in that the rules artificially capped the number of televised game licenses for sale). Id. at 99-100, 104 S.Ct. 2948. But it concluded that applying a per se rule of invalidity to the NCAA's television rules would be "inappropriate" because college football is "an industry in which horizontal restraints on competition are essential if the product is to be available at all." Id. at 100-01, 104 S.Ct. 2948. The Court elaborated:

Id. at 101-02, 104 S.Ct. 2948 (emphasis added). The Court held that the NCAA's rules should therefore be analyzed under the Rule of Reason.

Applying the Rule of Reason, the Court struck down the television rules on the ground that they did not serve any legitimate procompetitive purpose. Id. at 113-20, 104 S.Ct. 2948. It then concluded its opinion by stating:

Id. at 120, 104 S.Ct. 2948 (emphasis added).

Quoting heavily from the language in Board of Regents that we have emphasized, the NCAA contends that any Section 1 challenge to its amateurism rules must fail as a matter of law because the Board of Regents Court held that those rules are presumptively valid. We disagree.

The Board of Regents Court certainly discussed the NCAA's amateurism rules at great length, but it did not do so in order to pass upon the rules' merits, given that they were not before the Court. Rather, the Court discussed the amateurism rules for a different and particular purpose: to explain why NCAA rules should be analyzed under the Rule of Reason, rather than held to be illegal per se. The point was a significant one. Naked horizontal agreements among competitors to fix the price of a good or service, or to restrict their output, are usually condemned as per se unlawful. See, e.g., United States v. Trenton Potteries Co., 273 U.S. 392, 398, 47 S.Ct. 377, 71 L.Ed. 700 (1927); see also, e.g., Broad. Music, Inc. v. CBS, Inc., 441 U.S. 1, 19-20, 99 S.Ct. 1551, 60 L.Ed.2d 1 (1979) (arrangements that "almost always tend to restrict competition and decrease output" are usually per se illegal). The Board of Regents Court decided, however, that because college sports could not exist without certain horizontal agreements, NCAA rules should not be held per se unlawful even when — like the television rules in Board of Regents — they appear to be pure "restraints on the ability of member institutions to compete in terms of price and output." Bd. of Regents, 468 U.S. at 103, 104 S.Ct. 2948.

Board of Regents, in other words, did not approve the NCAA's amateurism rules as categorically consistent with the Sherman Act. Rather, it held that, because many NCAA rules (among them, the amateurism rules)9 are part of the "character and quality of the [NCAA's] `product,'" id. at 102, 104 S.Ct. 2948, no NCAA rule should be invalidated without a Rule of Reason analysis. The Court's long encomium to amateurism, though impressive-sounding, was therefore dicta. To be sure, "[w]e do not treat considered dicta from the Supreme Court lightly"; such dicta should be accorded "appropriate deference." United States v. Augustine, 712 F.3d 1290, 1295 (9th Cir.2013). Where applicable, we will give the quoted passages from Board of Regents that deference. But we are not bound by Board of Regents to conclude that every NCAA rule that somehow relates to amateurism is automatically valid.

What is more, even if the language in Board of Regents addressing amateurism were not dicta, it would not support the tremendous weight that the NCAA seeks to place upon it. The Court's opinion supports the proposition that the preservation of amateurism is a legitimate procompetitive purpose for the NCAA to pursue, but the NCAA is not asking us to find merely that its amateurism rules are procompetitive; rather, it asks us to hold that those rules are essentially exempt from antitrust scrutiny.10 Nothing in Board of Regents supports such an exemption. To say that the NCAA's amateurism rules are procompetitive, as Board of Regents did, is not to say that they are automatically lawful; a restraint that serves a procompetitive purpose can still be invalid under the Rule of Reason if a substantially less restrictive rule would further the same objectives equally well. See Bd. of Regents, 468 U.S. at 101 n. 23, 104 S.Ct. 2948 ("While as the guardian of an important American tradition, the NCAA's motives must be accorded a respectful presumption of validity, it is nevertheless well settled that good motives will not validate an otherwise anticompetitive practice.").

The NCAA cites decisions of three of our sister circuits, claiming that each adopted its view of Board of Regents. Two of these three cases, however, ultimately subjected the NCAA's rules to Rule of Reason scrutiny — the very approach we adopt today. See Smith v. NCAA, 139 F.3d 180, 186 (3d Cir.1998), vacated on other grounds by NCAA v. Smith, 525 U.S. 459, 119 S.Ct. 924, 142 L.Ed.2d 929 (1999); McCormack v. NCAA, 845 F.2d 1338, 1344-45 (5th Cir.1988). Only one — the Seventh Circuit's decision in Agnew v. NCAA, 683 F.3d 328 (7th Cir.2012) — comes close to agreeing with the NCAA's interpretation of Board of Regents, and we find it unpersuasive.

In Agnew, two former college football players who lost their scholarships challenged certain NCAA rules that prohibited schools from offering multi-year scholarships and capped the number of football scholarships each school could offer. Id. at 332-33. The Agnew court read Board of Regents broadly and concluded that, "when an NCAA bylaw is clearly meant to help maintain the `revered tradition of amateurism in college sports' or the `preservation of the student-athlete in higher education,' the bylaw [should] be presumed procompetitive." Id. at 342-43 (quoting Bd. of Regents, 468 U.S. at 120, 104 S.Ct. 2948). The court concluded, however, that the scholarship limitations that were before it did not "implicate the preservation of amateurism," since awarding more or longer scholarships to college athletes would not change their status as amateurs. Id. at 344. Thus, no "procompetitive presumption" applied to the scholarship rules. Id. at 345. Instead of dismissing the plaintiffs' antitrust claims on the merits, the court dismissed them on the unrelated ground that the plaintiffs had failed to plead the existence of a cognizable market. Id.

Like the amateurism language in Board of Regents, Agnew's "procompetitive presumption" was dicta that was ultimately unnecessary to the court's resolution of that case. But we would not adopt the Agnew presumption even if it were not dicta. Agnew's analysis rested on the dubious proposition that in Board of Regents, the Supreme Court "blessed" NCAA rules that were not before it, and did so to a sufficient degree to virtually exempt those rules from antitrust scrutiny. Id. at 341. We doubt that was the Court's intent, and we will not give such an aggressive construction to its words.

In sum, we accept Board of Regents' guidance as informative with respect to the procompetitive purposes served by the NCAA's amateurism rules, but we will go no further than that. The amateurism rules' validity must be proved, not presumed.

B. The Compensation Rules Regulate "Commercial Activity"

The NCAA next argues that we cannot reach the merits of the plaintiffs' Sherman Act claim because the compensation rules are not subject to the Sherman Act at all. The NCAA points out that Section 1 of the Sherman Act applies only to "restraint[s] of trade or commerce," 15 U.S.C. § 1, and claims that its compensation rules are mere "eligibility rules" that do not regulate any "commercial activity."

This argument is not credible. Although restraints that have no effect on commerce are indeed exempt from Section 1, the modern legal understanding of "commerce" is broad, "including almost every activity from which the actor anticipates economic gain." Phillip Areeda & Herbert Hovenkamp, Antitrust Law: An Analysis of Antitrust Principles and Their Application, ¶ 260b (4th ed.2013). That definition surely encompasses the transaction in which an athletic recruit exchanges his labor and NIL rights for a scholarship at a Division I school because it is undeniable that both parties to that exchange anticipate economic gain from it. See, e.g., Agnew, 683 F.3d at 340 ("No knowledgeable observer could earnestly assert that big-time college football programs competing for highly sought-after high school football players do not anticipate economic gain from a successful recruiting program."). Moreover, Board of Regents' discussion of the procompetitive justifications for NCAA amateurism rules shows that the Court "presume[d] the applicability of the Sherman Act to NCAA bylaws, since no procompetitive justifications would be necessary for noncommercial activity to which the Sherman Act does not apply." Id. at 339.

It is no answer to these observations to say, as the NCAA does in its briefs, that the compensation rules are "eligibility rules" rather than direct restraints on the terms of agreements between schools and recruits. True enough, the compensation rules are written in the form of eligibility rules; they provide that an athlete who receives compensation other than the scholarships specifically permitted by the NCAA loses his eligibility for collegiate sports. The mere fact that a rule can be characterized as an "eligibility rule," however, does not mean the rule is not a restraint of trade; were the law otherwise, the NCAA could insulate its member schools' relationships with student-athletes from antitrust scrutiny by renaming every rule governing student-athletes an "eligibility rule." The antitrust laws are not to be avoided by such "clever manipulation of words." Simpson v. Union Oil Co. of Cal., 377 U.S. 13, 21-22, 84 S.Ct. 1051, 12 L.Ed.2d 98 (1964).

In other words, the substance of the compensation rules matters far more than how they are styled. And in substance, the rules clearly regulate the terms of commercial transactions between athletic recruits and their chosen schools: a school may not give a recruit compensation beyond a grant-in-aid, and the recruit may not accept compensation beyond that limit, lest the recruit be disqualified and the transaction vitiated. The NCAA's argument that its compensation rules are "eligibility" restrictions, rather than substantive restrictions on the price terms of recruiting agreements, is but a sleight of hand. There is real money at issue here.

As the NCAA points out, two circuits have held that certain NCAA rules are noncommercial in nature. In Smith v. NCAA, the Third Circuit dismissed a student-athlete's challenge to the NCAA's "Postbaccalaureate Bylaw," which prohibited athletes from participating in athletics at postgraduate schools other than their undergraduate schools, on the grounds that the Sherman Act did not apply to that Bylaw. The Smith court held that eligibility rules such as the Postbaccalaureate Bylaw "are not related to the NCAA's commercial or business activities. Rather than intending to provide the NCAA with a commercial advantage, the eligibility rules primarily seek to ensure fair competition in intercollegiate athletics." Smith, 139 F.3d at 185.

The Sixth Circuit, meanwhile, held in Bassett v. NCAA, 528 F.3d 426, 430, 433 (6th Cir.2008), that the NCAA's rules against giving recruits "improper inducements" were "explicitly noncommercial." The court explained:

Id. at 433.

Neither Smith nor Bassett convinces us that the NCAA's Postbaccalaureate Bylaw challenged in Smith was a true "eligibility" rule, akin to the rules limiting the number of years that student-athletes may play collegiate sports or requiring student-athletes to complete a certain number of credit hours each semester. As the Smith court expressly noted, the Postbaccalaureate Bylaw was "not related to the NCAA's commercial or business activities." Smith, 139 F.3d at 185. By contrast, the rules here — which regulate what compensation NCAA schools may give student-athletes, and how much — do relate to the NCAA's business activities: the labor of student-athletes is an integral and essential component of the NCAA's "product," and a rule setting the price of that labor goes to the heart of the NCAA's business. Thus, the rules at issue here are more like rules affecting the NCAA's dealings with its coaches or with corporate business partners — which courts have held to be commercial — than they are like the Bylaw challenged in Smith. See Bd. of Regents, 468 U.S. at 104-13, 104 S.Ct. 2948 (applying Sherman Act to rules governing NCAA members' contracts with television networks); Law v. NCAA, 134 F.3d 1010, 1024 (10th Cir.1998) (applying Sherman Act to NCAA rules limiting compensation of basketball coaches).

Bassett cannot be distinguished here in the way that Smith can since it involved an NCAA rule relating to payments to athletic recruits, but we believe Bassett was simply wrong on this point. Bassett's reasoning, in fine, is that rules that seek to combat commercialism in college sports by preventing schools from competing to pay student-athletes cannot be considered restraints on "commerce." We simply cannot understand this logic. Rules that are "anti-commercial and designed to promote and ensure competitiveness," Bassett, 528 F.3d at 433, surely affect commerce just as much as rules promoting commercialism. The intent behind the NCAA's compensation rules does not change the fact that the exchange they regulate — labor for in-kind compensation — is a quintessentially commercial transaction.

We therefore conclude that the NCAA's compensation rules are within the ambit of the Sherman Act.

C. The Plaintiffs Demonstrated that the Compensation Rules Cause Them Injury in Fact

The NCAA's last argument antecedent to the merits is that the plaintiffs' Section 1 claim fails at the threshold because the plaintiffs have failed to show that they have suffered "antitrust injury." Antitrust injury is a heightened standing requirement that applies to private parties suing to enforce the antitrust laws. To satisfy the antitrust-injury requirement, a plaintiff must show "injury of the type the antitrust laws were intended to prevent and that flows from that which makes defendants' acts unlawful." Glen Holly Entm't, Inc. v. Tektronix Inc., 343 F.3d 1000, 1007-08 (9th Cir.2003) (quoting Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 489, 97 S.Ct. 690, 50 L.Ed.2d 701 (1977)) (internal quotation marks omitted).

Although the NCAA purports to be making an antitrust-injury argument, it is mistaken. The NCAA has not contended that the plaintiffs' injuries are not "of the type the antitrust laws were intended to prevent." Rather, the NCAA has made a garden-variety standing argument: it alleges that the plaintiffs have not been injured in fact by the compensation rules because those rules do not deprive them of any NIL compensation they would otherwise receive. Addressing each of the potential markets for NIL rights that the district court identified, the NCAA argues that (1) there are no legally-recognized NIL rights for participants in live game broadcasts; (2) the NCAA's compensation rules do not deprive the plaintiffs of compensation for use of their NILs in video games because the NCAA no longer permits college sports video games to be made and has a separate policy forbidding the use of student-athletes' NILs in video games; and (3) the NCAA's licensing agreement for archival footage with T3Media does not deprive athletes of NIL compensation for archival footage because it prevents T3Media from licensing student-athletes' NILs while they are in school and requires the company to obtain consent once student-athletes have left school.

We conclude that the plaintiffs have shown that they are injured in fact as a result of the NCAA's rules having foreclosed the market for their NILs in video games. We therefore do not reach the thornier questions of whether participants in live TV broadcasts of college sporting events have enforceable rights of publicity or whether the plaintiffs are injured by the NCAA's current licensing arrangement for archival footage.

1. Absent the NCAA's compensation rules, video game makers would negotiate with student-athletes for the right to use their NILs

As we have explained, the district court found that, if permitted to do so, video game makers such as EA would negotiate with college athletes for the right to use their NILs in video games because these companies want to make games that are as realistic as possible. O'Bannon, 7 F.Supp.3d at 970. The district court noted that EA currently negotiates with the NFL and NBA players' unions for the right to use their members' NILs in pro sports video games. Id. The plaintiffs also put into evidence a copy of a 2005 presentation by EA representatives to the NCAA, which stated that EA's inability to use college athletes' NILs was the "number one factor holding back NCAA video game growth."

The NCAA argues, however, that we cannot find that the plaintiffs have suffered an injury in fact based on lost compensation from video game companies because the NCAA has terminated its relationship with EA and is not currently working with any other video game maker.11 We disagree. The district court found that it is entirely possible that the NCAA will resume its support for college sports video games at some point in the future, given that the NCAA found such games to be profitable in the past, id., and that finding of fact was not clearly erroneous. Given the NCAA's previous, lengthy relationship with EA and the other evidence presented, it was reasonable for the district court to conclude that the NCAA may well begin working with EA or another video game company in the future.12

Our conclusion is unaffected by the NCAA's claim that other rules and policies, not directly at issue here, would forbid video game makers from using student-athletes' NILs in their games if such games were to be made again. The NCAA did, after all, permit EA to continue making NCAA video games for some time after EA began incorporating recognizable player avatars into the games. Moreover, Joel Linzner, a EA executive, testified at trial that EA "made a long-sustained effort to work with the NCAA" to change the policy against using student-athletes' NILs, and that NCAA executives were "supportive" of the idea. It was not clearly erroneous for the district court to conclude on the basis of this evidence that the NCAA might well either change its policy barring the use of athletes' NILs in video games or decline to enforce it.

2. Whether the Copyright Act preempts right-of-publicity claims based on sports video games is tangential to this case and irrelevant to the plaintiffs' standing

In addition to arguing that its current policies against college sports video games defeat the plaintiffs' claims to standing, the NCAA also contends that there are legal barriers that would prevent the plaintiffs from being compensated by a video game maker. Specifically, the NCAA argues that the Copyright Act would preempt any right-of-publicity claim arising out of the use of those NILs in sports video games.13 Thus, the NCAA maintains, if it were to resume its support for college sports video games and permit video game companies to use student-athletes' NILs, the video game makers would not pay student-athletes for their NILs; rather, they could use the NILs for free.

We decline to consider this argument, for two reasons. First, it is convoluted and far afield from the main issues in this case. The NCAA asks us to decide whether, assuming that EA or some other video game company were to make a college sports video game that incorporated student-athletes' NILs and then refuse to pay student-athletes for those NILs, the game maker would have a viable Copyright Act defense to a right-of-publicity lawsuit brought by the athletes. That question is a complex one, implicating both Section 301 of the Copyright Act, 17 U.S.C. § 301, which expressly preempts certain common-law claims, and a murky body of case law holding that, in some circumstances, the Act impliedly preempts claims that fall outside of Section 301's scope. See, e.g., Facenda v. NFL Films, Inc., 542 F.3d 1007, 1028-32 (3d Cir.2008) (suggesting, on the basis of a conflict preemption analysis, that federal copyright law can "impliedly preempt[]" right-of-publicity claims). It is scarcely fit for resolution within the confines of a standing inquiry in an antitrust suit between the NCAA and its studentathletes that involves neither EA nor any other video game company as a party. Should a college sports video game be made in the future and the right-of-publicity suit envisioned by the NCAA come to pass, the court hearing that suit will be in a far better position to resolve the question of Copyright Act preemption than we are.

Second and more importantly, the NCAA's argument about the Copyright Act, even if correct, is irrelevant to whether the plaintiffs lack standing. On the NCAA's interpretation of the Copyright Act, professional football and basketball players have no enforceable right-of-publicity claims against video game makers either — yet EA currently pays NFL and NBA players for the right to use their NILs in its video games. O'Bannon, 7 F.Supp.3d at 970. Thus, there is every reason to believe that, if permitted to do so, EA or another video game company would pay NCAA athletes for their NIL rights rather than test the enforceability of those rights in court. That the NCAA's rules deny the plaintiffs all opportunity to receive this compensation is sufficient to endow them with standing to bring this lawsuit. See 13A Charles Alan Wright & Arthur R. Miller, Federal Practice and Procedure § 3531.4 (3d ed. 1998) ("[L]oss of an opportunity may constitute injury, even though it is not certain that any benefit would have been realized if the opportunity had been accorded." (collecting cases)); cf., e.g., United States v. Students Challenging Regulatory Agency Procedures (SCRAP), 412 U.S. 669, 689 n. 14, 93 S.Ct. 2405, 37 L.Ed.2d 254 (1973) (rejecting the government's argument that standing should be limited "to those who have been `significantly' affected by agency action"); Preminger v. Peake, 552 F.3d 757, 763 (9th Cir.2008) ("The injury may be minimal.").

* * *

Because the plaintiffs have shown that, absent the NCAA's compensation rules, video game makers would likely pay them for the right to use their NILs in college sports video games, the plaintiffs have satisfied the requirement of injury in fact and, by extension, the requirement of antitrust injury.

IV

Having rejected all of the NCAA's preliminary legal arguments, we proceed to review the plaintiffs' Section 1 claim on the merits. Although in another context the NCAA's decision to value student-athletes' NIL at zero might be per se illegal price fixing, we are persuaded — as was the Supreme Court in Board of Regents and the district court here — that the appropriate rule is the Rule of Reason. As the Supreme Court observed, the NCAA "market[s] a particular brand ... [that] makes it more popular than professional sports to which it might otherwise be comparable." Board of Regents, 468 U.S. at 101-02, 104 S.Ct. 2948. Because the "integrity of the `product' cannot be preserved except by mutual agreement," "restraints on competition are essential if the product is to be available at all." Id. at 101, 102, 104 S.Ct. 2948; see also id. at 117, 104 S.Ct. 2948 ("Our decision not to apply a per se rule to this case rests in large part on our recognition that a certain degree of cooperation is necessary if the type of competition that [the NCAA] and its member institutions seek to market is to be preserved." (footnote omitted)).

Like the district court, we follow the three-step framework of the Rule of Reason: "[1] The plaintiff bears the initial burden of showing that the restraint produces significant anticompetitive effects within a relevant market. [2] If the plaintiff meets this burden, the defendant must come forward with evidence of the restraint's procompetitive effects. [3] The plaintiff must then show that any legitimate objectives can be achieved in a substantially less restrictive manner." Tanaka v. Univ. of S. Cal., 252 F.3d 1059, 1063 (9th Cir.2001) (citations and internal quotation marks omitted).

A. Significant Anticompetitive Effects Within a Relevant Market

As we have recounted, the district court made the following factual findings: (1) that a cognizable "college education market" exists, wherein colleges compete for the services of athletic recruits by offering them scholarships and various amenities, such as coaching and facilities; (2) that if the NCAA's compensation rules did not exist, member schools would compete to offer recruits compensation for their NILs; and (3) that the compensation rules therefore have a significant anticompetitive effect on the college education market, in that they fix an aspect of the "price" that recruits pay to attend college (or, alternatively, an aspect of the price that schools pay to secure recruits' services). These findings have substantial support in the record.

By and large, the NCAA does not challenge the district court's findings. It does not take issue with the way that the district court defined the college education market. Nor does it appear to dispute the district court's conclusion that the compensation rules restrain the NCAA's member schools from competing with each other within that market, at least to a certain degree. Instead, the NCAA makes three modest arguments about why the compensation rules do not have a significant anticompetitive effect. First, it argues that because the plaintiffs never showed that the rules reduce output in the college education market, the plaintiffs did not meet their burden of showing a significant anticompetitive effect. Second, it argues that the rules have no anticompetitive effect because schools would not pay studentathletes anything for their NIL rights in any event, given that those rights are worth nothing. And finally, the NCAA argues that even if the district court was right that schools would pay student-athletes for their NIL rights, any such payments would be small, which means that the compensation rules' anticompetitive effects cannot be considered significant.

We can dispose of the first two arguments quickly. First, the NCAA's contention that the plaintiffs' claim fails because they did not show a decrease in output in the college education market is simply incorrect. Here, the NCAA argues that output in the college education market "consists of opportunities for student-athletes to participate in FBS football or Division I men's basketball," and it quotes the district court's finding that these opportunities have "increased steadily over time." See O'Bannon, 7 F.Supp.3d at 981. But this argument misses the mark. Although output reductions are one common kind of anticompetitive effect in antitrust cases, a "reduction in output is not the only measure of anticompetitive effect." Areeda & Hovenkamp ¶ 1503b(1) (emphasis added).

The "combination[s] condemned by the [Sherman] Act" also include "price-fixing... by purchasers" even though "the persons specially injured ... are sellers, not customers or consumers." Mandeville Island Farms, Inc. v. Am. Crystal Sugar Co., 334 U.S. 219, 235, 68 S.Ct. 996, 92 L.Ed. 1328 (1948). At trial, the plaintiffs demonstrated that the NCAA's compensation rules have just this kind of anticompetitive effect: they fix the price of one component of the exchange between school and recruit, thereby precluding competition among schools with respect to that component. The district court found that although consumers of NCAA football and basketball may not be harmed directly by this price-fixing, the "student-athletes themselves are harmed by the price-fixing agreement among FBS football and Division I basketball schools." O'Bannon, 7 F.Supp.3d at 972-73. The athletes accept grants-in-aid, and no more, in exchange for their athletic performance, because the NCAA schools have agreed to value the athletes' NILs at zero, "an anticompetitive effect."14 Id. at 973. This anticompetitive effect satisfied the plaintiffs' initial burden under the Rule of Reason. Cf. Cal. Dental Ass'n v. FTC, 526 U.S. 756, 777, 119 S.Ct. 1604, 143 L.Ed.2d 935 (1999) ("[R]aising price, reducing output, and dividing markets have the same anticompetitive effects." (quoting Gen. Leaseways, Inc. v. Nat'l Truck Leasing Ass'n, 744 F.2d 588, 594-95 (7th Cir. 1984))).

Second, the NCAA's argument that student-athletes' NILs are, in fact, worth nothing is simply a repackaged version of its arguments about injury in fact, which we have rejected.

Finally, we reject the NCAA's contention that any NIL compensation that student-athletes might receive in the absence of its compensation rules would be de minimis and that the rules therefore do not significantly affect competition in the college education market. This "too small to matter" argument is incompatible with the Supreme Court's holding in Catalano, Inc. v. Target Sales, Inc., 446 U.S. 643, 100 S.Ct. 1925, 64 L.Ed.2d 580 (1980) (per curiam). In Catalano, a group of beer retailers sued a group of beer wholesalers, alleging that the wholesalers had secretly agreed to end their customary practice of extending the retailers interest-free credit for roughly a month after the delivery of beer. Id. at 644, 100 S.Ct. 1925. The Court unanimously held that this agreement was unlawful per se. It reasoned that the agreement was clearly a means of "extinguishing one form of [price] competition among the sellers," given that credit terms were part of the price of the beer, and that the agreement was therefore tantamount to price-fixing. Id. at 649, 100 S.Ct. 1925. The Court was not concerned with whether the agreement affected the market adversely: "It is no excuse that the prices fixed are themselves reasonable." Id. at 647, 100 S.Ct. 1925.

The NCAA's compensation rules function in much the same way as the agreement at issue in Catalano: they "extinguish[] one form of competition" among schools seeking to land recruits. We acknowledge that Catalano was a per se case in which the Court did not analyze the anticompetitive effect of the wholesalers' agreement in detail,15 but the decision nonetheless indicates that an antitrust court should not dismiss an anticompetitive price-fixing agreement as benign simply because the agreement relates only to one component of an overall price. That proposition finds further support in Board of Regents: in Board of Regents, a Rule of Reason case, the Court held that the NCAA's television plan had "a significant potential for anticompetitive effects" without delving into the details of exactly how much the plan restricted output of televised games or how much it fixed the price of TV contracts. 468 U.S. at 104-05, 104 S.Ct. 2948. While the precise value of NIL compensation is uncertain, at this point in the analysis and in light of Catalano and Board of Regents, we conclude that the plaintiffs have met their burden at the first step of the Rule of Reason by showing that the NCAA's compensation rules fix the price of one component (NIL rights) of the bundle that schools provide to recruits.

Because we agree with the district court that the compensation rules have a significant anticompetitive effect on the college education market, we proceed to consider the procompetitive justifications the NCAA proffers for those rules.

B. Procompetitive Effects

As discussed above, the NCAA offered the district court four procompetitive justifications for the compensation rules: (1) promoting amateurism, (2) promoting competitive balance among NCAA schools, (3) integrating student-athletes with their schools' academic community, and (4) increasing output in the college education market. The district court accepted the first and third and rejected the other two.

Although the NCAA's briefs state in passing that the district court erred in failing to "credit all four justifications fully," the NCAA focuses its arguments to this court entirely on the first proffered justification — the promotion of amateurism. We therefore accept the district court's factual findings that the compensation rules do not promote competitive balance, that they do not increase output in the college education market, and that they play a limited role in integrating student-athletes with their schools' academic communities, since we have been offered no meaningful argument that those findings were clearly erroneous. See, e.g., Md. Cas. Co. v. Knight, 96 F.3d 1284, 1291 (9th Cir.1996).

The district court acknowledged that the NCAA's current rules promote amateurism, which in turn plays a role in increasing consumer demand for college sports. O'Bannon, 7 F.Supp.3d at 978. The NCAA does not challenge that specific determination, but it argues to us that the district court gave the amateurism justification short shrift, in two respects. First, it claims that the district court erred by focusing solely on the question of whether amateurism increases consumers' (i.e., fans') demand for college sports and ignoring the fact that amateurism also increases choice for student-athletes by giving them "the only opportunity [they will] have to obtain a college education while playing competitive sports as students." Second, it faults the district court for being inappropriately skeptical of the NCAA's historical commitment to amateurism. Although we might have credited the depth of the NCAA's devotion to amateurism differently, these arguments do not persuade us that the district court clearly erred.

The NCAA is correct that a restraint that broadens choices can be procompetitive. The Court in Board of Regents observed that the difference between college and professional sports "widen[s]" the choices "available to athletes." Bd. of Regents, 468 U.S. at 102, 104 S.Ct. 2948. But we fail to see how the restraint at issue in this particular case — i.e., the NCAA's limits on student-athlete compensation — makes college sports more attractive to recruits, or widens recruits' spectrum of choices in the sense that Board of Regents suggested. As the district court found, it is primarily "the opportunity to earn a higher education" that attracts athletes to college sports rather than professional sports, O'Bannon, 7 F.Supp.3d at 986, and that opportunity would still be available to student-athletes if they were paid some compensation in addition to their athletic scholarships. Nothing in the plaintiffs' prayer for compensation would make student-athletes something other than students and thereby impair their ability to become student-athletes.

Indeed, if anything, loosening or abandoning the compensation rules might be the best way to "widen" recruits' range of choices; athletes might well be more likely to attend college, and stay there longer, if they knew that they were earning some amount of NIL income while they were in school. See Jeffrey L. Harrison & Casey C. Harrison, The Law and Economics of the NCAA's Claim to Monopsony Rights, 54 Antitrust Bull. 923, 948 (2009). We therefore reject the NCAA's claim that, by denying student-athletes compensation apart from scholarships, the NCAA increases the "choices" available to them.16

The NCAA's second point has more force — the district court probably underestimated the NCAA's commitment to amateurism. See Bd. of Regents, 468 U.S. at 120, 104 S.Ct. 2948 (referring to the NCAA's "revered tradition of amateurism in college sports"). But the point is ultimately irrelevant. Even if the NCAA's concept of amateurism had been perfectly coherent and consistent, the NCAA would still need to show that amateurism brings about some procompetitive effect in order to justify it under the antitrust laws. See id. at 101-02 & n. 23, 104 S.Ct. 2948. The NCAA cannot fully answer the district court's finding that the compensation rules have significant anticompetitive effects simply by pointing out that it has adhered to those rules for a long time. Nevertheless, the district court found, and the record supports that there is a concrete procompetitive effect in the NCAA's commitment to amateurism: namely, that the amateur nature of collegiate sports increases their appeal to consumers. We therefore conclude that the NCAA's compensation rules serve the two procompetitive purposes identified by the district court: integrating academics with athletics, and "preserving the popularity of the NCAA's product by promoting its current understanding of amateurism." O'Bannon, 7 F.Supp.3d at 1005.17

We note that the district court's findings are largely consistent with the Supreme Court's own description of the college football market as "a particular brand of football" that draws from "an academic tradition [that] differentiates [it] from and makes it more popular than professional sports to which it might otherwise be comparable, such as, for example, minor league baseball." Bd. of Regents, 468 U.S. at 101-02, 104 S.Ct. 2948. "Thus, the NCAA plays a vital role in enabling college football to preserve its character, and as a result enables a product to be marketed which might otherwise be unavailable." Id. at 102, 104 S.Ct. 2948. But, as Board of Regents demonstrates, not every rule adopted by the NCAA that restricts the market is necessary to preserving the "character" of college sports. We thus turn to the final inquiry — whether there are reasonable alternatives to the NCAA's current compensation restrictions.

C. Substantially Less Restrictive Alternatives

The third step in the Rule of Reason analysis is whether there are substantially less restrictive alternatives to the NCAA's current rules. We bear in mind that — to be viable under the Rule of Reason — an alternative must be "virtually as effective" in serving the procompetitive purposes of the NCAA's current rules, and "without significantly increased cost." Cnty. of Tuolumne v. Sonora Cmty. Hosp., 236 F.3d 1148, 1159 (9th Cir.2001) (internal quotation marks omitted). We think that plaintiffs must make a strong evidentiary showing that its alternatives are viable here. Not only do plaintiffs bear the burden at this step, but the Supreme Court has admonished that we must generally afford the NCAA "ample latitude" to superintend college athletics. Bd. of Regents, 468 U.S. at 120, 104 S.Ct. 2948; see also Law v. Nat'l Collegiate Athletic Ass'n, 134 F.3d 1010, 1022 (10th Cir.1998) ("[C]ourts should afford the NCAA plenty of room under the antitrust laws to preserve the amateur character of intercollegiate athletics."); Race Tires Am., Inc. v. Hoosier Racing Tire Corp., 614 F.3d 57, 83 (3d Cir.2010) (noting that, generally, "sports-related organizations should have the right to determine for themselves the set of rules that they believe best advance their respective sport").

The district court identified two substantially less restrictive alternatives: (1) allowing NCAA member schools to give student-athletes grants-in-aid that cover the full cost of attendance; and (2) allowing member schools to pay student-athletes small amounts of deferred cash compensation for use of their NILs.18 O'Bannon, 7 F.Supp.3d at 1005-07. We hold that the district court did not clearly err in finding that raising the grant-in-aid cap would be a substantially less restrictive alternative, but that it clearly erred when it found that allowing students to be paid compensation for their NILs is virtually as effective as the NCAA's current amateur-status rule.

1. Capping the permissible amount of scholarships at the cost of attendance

The district court did not clearly err in finding that allowing NCAA member schools to award grants-in-aid up to their full cost of attendance would be a substantially less restrictive alternative to the current compensation rules. All of the evidence before the district court indicated that raising the grant-in-aid cap to the cost of attendance would have virtually no impact on amateurism: Dr. Mark Emmert, the president of the NCAA, testified at trial that giving student-athletes scholarships up to their full costs of attendance would not violate the NCAA's principles of amateurism because all the money given to students would be going to cover their "legitimate costs" to attend school. Other NCAA witnesses agreed with that assessment. Id. at 983. Nothing in the record, moreover, suggested that consumers of college sports would become less interested in those sports if athletes' scholarships covered their full cost of attendance, or that an increase in the grant-in-aid cap would impede the integration of student-athletes into their academic communities. Id.

The NCAA, along with fifteen scholars of antitrust law appearing as amici curiae, warns us that if we affirm even this more modest of the two less restrictive alternative restraints identified by the district court, we will open the floodgates to new lawsuits demanding all manner of incremental changes in the NCAA's and other organizations' rules. The NCAA and these amici admonish us that as long as a restraint (such as a price cap) is "reasonably necessary to a valid business purpose," it should be upheld; it is not an antitrust court's function to tweak every market restraint that the court believes could be improved.

We agree with the NCAA and the amici that, as a general matter, courts should not use antitrust law to make marginal adjustments to broadly reasonable market restraints. See, e.g., Bruce Drug, Inc. v. Hollister, Inc., 688 F.2d 853, 860 (1st Cir. 1982) (noting that defendants are "not required to adopt the least restrictive" alternative); Am. Motor Inns, Inc. v. Holiday Inns, Inc., 521 F.2d 1230, 1249 (3d Cir. 1975) (denying that "the availability of an alternative means of achieving the asserted business purpose renders the existing arrangement unlawful if that alternative would be less restrictive of competition no matter to how small a degree"). The particular restraint at issue here, however — the grant-in-aid cap that the NCAA set below the cost of attendance — is not such a restraint. To the contrary, the evidence at trial showed that the grant-in-aid cap has no relation whatsoever to the procompetitive purposes of the NCAA: by the NCAA's own standards, student-athletes remain amateurs as long as any money paid to them goes to cover legitimate educational expenses.

Thus, in holding that setting the grant-in-aid cap at student-athletes' full cost of attendance is a substantially less restrictive alternative under the Rule of Reason, we are not declaring that courts are free to micromanage organizational rules or to strike down largely beneficial market restraints with impunity. Rather, our affirmance of this aspect of the district court's decision should be taken to establish only that where, as here, a restraint is patently and inexplicably stricter than is necessary to accomplish all of its procompetitive objectives, an antitrust court can and should invalidate it and order it replaced with a less restrictive alternative.

A compensation cap set at student-athletes' full cost of attendance is a substantially less restrictive alternative means of accomplishing the NCAA's legitimate procompetitive purposes. And there is no evidence that this cap will significantly increase costs; indeed, the NCAA already permits schools to fund student-athletes' full cost of attendance. The district court's determination that the existing compensation rules violate Section 1 of the Sherman Act was correct and its injunction requiring the NCAA to permit schools to provide compensation up to the full cost of attendance was proper.

2. Allowing students to receive cash compensation for their NILs

In our judgment, however, the district court clearly erred in finding it a viable alterative to allow students to receive NIL cash payments untethered to their education expenses. Again, the district court identified two procompetitive purposes served by the NCAA's current rules: "preserving the popularity of the NCAA's product by promoting its current understanding of amateurism" and "integrating academics and athletics." O'Bannon, 7 F.Supp.3d at 1005; see also Board of Regents, 468 U.S. at 117, 104 S.Ct. 2948 ("It is reasonable to assume that most of the regulatory controls of the NCAA are justifiable means of fostering competition among amateur athletic teams and therefore procompetitive because they enhance public interest in intercollegiate athletics."). The question is whether the alternative of allowing students to be paid NIL compensation unrelated to their education expenses, is "virtually as effective" in preserving amateurism as not allowing compensation. Cnty. of Tuolumne, 236 F.3d at 1159 (internal quotation marks omitted).

We cannot agree that a rule permitting schools to pay students pure cash compensation and a rule forbidding them from paying NIL compensation are both equally effective in promoting amateurism and preserving consumer demand.19 Both we and the district court agree that the NCAA's amateurism rule has procompetitive benefits. But in finding that paying students cash compensation would promote amateurism as effectively as not paying them, the district court ignored that not paying student-athletes is precisely what makes them amateurs.20

Having found that amateurism is integral to the NCAA's market, the district court cannot plausibly conclude that being a poorly-paid professional collegiate athlete is "virtually as effective" for that market as being as amateur. Or, to borrow the Supreme Court's analogy, the market for college football is distinct from other sports markets and must be "differentiate[d]" from professional sports lest it become "minor league [football]." Bd. of Regents, 468 U.S. at 102, 104 S.Ct. 2948.

Aside from the self-evident fact that paying students for their NIL rights will vitiate their amateur status as collegiate athletes, the court relied on threadbare evidence in finding that small payments of cash compensation will preserve amateurism as well the NCAA's rule forbidding such payments. Most of the evidence elicited merely indicates that paying students large compensation payments would harm consumer demand more than smaller payments would — not that small cash payments will preserve amateurism. Thus, the evidence was addressed to the wrong question. Instead of asking whether making small payments to student-athletes served the same procompetitive purposes as making no payments, the evidence before the district court went to a different question: Would the collegiate sports market be better off if the NCAA made small payments or big payments? For example, the district court noted that a witness called by the NCAA, Bernard Muir, the athletic director at Stanford University, testified that paying student-athletes modest sums raises less concern than paying them large sums. The district court also relied on Dr. Dennis's opinion survey, which the court read to indicate that in the absence of the NCAA's compensation rules, "the popularity of college sports would likely depend on the size of payments awarded to student-athletes." O'Bannon, 7 F.Supp.3d at 983. Dr. Dennis had found that payments of $200,000 per year to each athlete would alienate the public more than would payments of $20,000 per year. Id. at 975-76, 983. At best, these pieces of evidence indicate that small payments to players will impact consumer demand less than larger payments. But there is a stark difference between finding that small payments are less harmful to the market than large payments — and finding that paying students small sums is virtually as effective in promoting amateurism as not paying them.

The other evidence cited by the district court is even less probative of whether paying these student-athletes will preserve amateurism and consumer demand. The district court adverted to testimony from a sports management expert, Daniel Rascher, who explained that although opinion surveys had shown the public was opposed to rising baseball salaries during the 1970s, and to the decision of the International Olympic Committee to allow professional athletes to compete in the Olympics, the public had continued to watch baseball and the Olympics at the same rate after those changes. Id. at 976-77. But professional baseball and the Olympics are not fit analogues to college sports.21 The Olympics have not been nearly as transformed by the introduction of professionalism as college sports would be.

Finally, the district court, and the dissent, place particular weight on a brief interchange during plaintiffs' cross-examination of one of the NCAA's witnesses, Neal Pilson, a television sports consultant formerly employed at CBS. Pilson testified that "if you're paid for your performance, you're not an amateur," and explained at length why paying students would harm the student-athlete market. Plaintiffs then asked Pilson whether his opinions about amateurism "depend on the level of the money" paid to players, and he acknowledged that his opinion was "impacted by the level." When asked whether there was a line that "should not be crossed" in paying players, Pilson responded "that's a difficult question. I haven't thought about the line. And I haven't been asked to render an opinion on that." When pressed to come up with a figure, Pilson repeated that he was "not sure." He eventually commented that "I tell you that a million dollars would trouble me and $5,000 wouldn't, but that's a pretty good range." When asked whether deferred compensation to students would concern him, Pilson said that while he would not be as concerned by deferred payments, he would still be "troubled by it."22

So far as we can determine, Pilson's offhand comment under cross-examination is the sole support for the district court's $5,000 figure. But even taking Pilson's comments at face value, as the dissent urges, his testimony cannot support the finding that paying student-athletes small sums will be virtually as effective in preserving amateurism as not paying them. Pilson made clear that he was not prepared to opine on whether pure cash compensation, of any amount, would affect amateurism. Indeed, he was never asked about the impact of giving student-athletes small cash payments; instead, like other witnesses, he was asked only whether big payments would be worse than small payments. Pilson's casual comment — "[I] haven't been asked to render an opinion on that. It's not in my report" — that he would not be troubled by $5,000 payments is simply not enough to support the district court's far-reaching conclusion that paying students $5,000 per year will be as effective in preserving amateurism as the NCAA's current policy.23

The difference between offering student-athletes education-related compensation and offering them cash sums untethered to educational expenses is not minor; it is a quantum leap.24 Once that line is crossed, we see no basis for returning to a rule of amateurism and no defined stopping point; we have little doubt that plaintiffs will continue to challenge the arbitrary limit imposed by the district court until they have captured the full value of their NIL. At that point the NCAA will have surrendered its amateurism principles entirely and transitioned from its "particular brand of football" to minor league status. Bd. of Regents, 468 U.S. at 101-02, 104 S.Ct. 2948. In light of that, the meager evidence in the record, and the Supreme Court's admonition that we must afford the NCAA "ample latitude" to superintend college athletics, Bd. of Regents, 468 U.S. at 120, 104 S.Ct. 2948, we think it is clear the district court erred in concluding that small payments in deferred compensation are a substantially less restrictive alternative restraint.25 We thus vacate that portion of the district court's decision and the portion of its injunction requiring the NCAA to allow its member schools to pay this deferred compensation.

V

By way of summation, we wish to emphasize the limited scope of the decision we have reached and the remedy we have approved. Today, we reaffirm that NCAA regulations are subject to antitrust scrutiny and must be tested in the crucible of the Rule of Reason. When those regulations truly serve procompetitive purposes, courts should not hesitate to uphold them. But the NCAA is not above the antitrust laws, and courts cannot and must not shy away from requiring the NCAA to play by the Sherman Act's rules. In this case, the NCAA's rules have been more restrictive than necessary to maintain its tradition of amateurism in support of the college sports market. The Rule of Reason requires that the NCAA permit its schools to provide up to the cost of attendance to their student athletes. It does not require more.

We vacate the district court's judgment and permanent injunction insofar as they require the NCAA to allow its member schools to pay student-athletes up to $5,000 per year in deferred compensation. We otherwise affirm. The parties shall bear their own costs on appeal.

AFFIRMED IN PART and VACATED IN PART.

THOMAS, Chief Judge, concurring in part and dissenting in part:

I largely agree with all but one of the majority's conclusions.1 I respectfully disagree with the majority's conclusion that the district court clearly erred in ordering the NCAA to permit up to $5,000 in deferred compensation above student-athletes' full cost of attendance.

I

We review the district court's determinations of fact for clear error. We are not permitted to "review the evidence de novo and freely substitute our judgment for that of the trial judge." United States v. Ironworkers Local 86, 443 F.2d 544, 549 (9th Cir.1971). Rather, the clear error standard "is significantly deferential, and we will accept the lower court's findings of fact unless we are left with the definite and firm conviction that a mistake has been committed." Lentini v. Cal. Ctr. for the Arts, Escondido, 370 F.3d 837, 848-49 (9th Cir.2004).

There was sufficient evidence in the record to support the award. The district court's conclusion that the proposed alternative restraint satisfied the Rule of Reason was based on testimony from at least four experts-including three experts presented by the NCAA-that providing student-athletes with small amounts of compensation above their cost of attendance most likely would not have a significant impact on consumer interest in college sports. O'Bannon, 7 F.Supp.3d at 976-77, 983-84, 1000-01. It was also based on the fact that FBS football players are currently permitted to accept Pell grants in excess of their cost of attendance, and the fact that Division I tennis recruits are permitted to earn up to $10,000 per year in prize money from athletic events before they enroll in college. Id. at 974, 1000. The majority characterizes the weight of this evidence as "threadbare." Op. at 1077. I respectfully disagree.

The NCAA's own expert witness, Neal Pilson, testified that the level of deferred compensation would have an effect on consumer demand for college athletics, but that paying student-athletes $5,000 per year in trust most likely would not have a significant impact on such demand. He also testified that any negative impact that paying student-athletes might have on consumer demand could be partially mitigated by placing the compensation in a trust fund to be paid out after graduation.

The majority dismisses this testimony because it was made in a very "offhand" manner, and because Pilson proffered the $5,000 amount on cross-examination "[w]hen pressed." Op. at 1078. However, the NCAA presented this witness as an expert on the issue of whether paying college athletes will negatively impact consumer demand for college sports.2 Pilson testified at length on the topic, and his qualifications were not challenged. It is not appropriate for us on appeal to assess demeanor we did not see. As a result, I would take the testimony at face value, and the district court did not clearly err in crediting it.

The majority also dismisses the testimony given by expert witness Dr. Daniel Rascher demonstrating that consumer interest in major league baseball and the Olympics increased after baseball players' salaries rose and professional athletes were allowed to compete in the Olympics. The majority reasons that major league baseball and the Olympics are "not fit analogues to college sports," speculating that college sports would be more significantly transformed by professionalism than have the Olympics. Op. at 1077. However, the majority does not offer any evidentiary support for the distinction, nor explain how or why the district court clearly erred in crediting this testimony.

Moreover, Rascher also testified that consumer demand in sports such as tennis and rugby increased after the sports' governing boards permitted athletes to receive payment. O'Bannon, 7 F.Supp.3d at 977. In my view, the majority errs in dismissing this testimony. The import of Rascher's testimony was that consumer demand typically does not decrease when athletes are permitted to receive payment, and that this general principle holds true across a wide variety of sports and competitive formats. The district court did not clearly err in crediting it.

The district court accepted the testimony of multiple experts that small amounts of compensation would not affect consumer demand, and then used the lowest amount suggested by one of the NCAA's experts. The district court was within its right to do so.3

II

The disagreement between my view and the majority view largely boils down to a difference in opinion as to the procompetitive interests at stake. The majority characterizes our task at step three of the Rule of Reason as determining "whether the alternative of allowing students to be paid NIL compensation unrelated to their education expenses is `virtually as effective' in preserving amateurism as not allowing compensation." Op. at 1076 (emphasis added). This conclusion misstates our inquiry. Rather, we must determine whether allowing student-athletes to be compensated for their NILs is `virtually as effective' in preserving popular demand for college sports as not allowing compensation. In terms of antitrust analysis, the concept of amateurism is relevant only insofar as it relates to consumer interest.

The district court found that there are two, limited procompetitive benefits to the current rule. It found that limits on large amounts of student-athlete compensation preserve the popularity of the NCAA's product, and that limits on large amounts of student-athlete compensation promote the integration of academics and athletics. O'Bannon, 7 F.Supp.3d at 1004-05. In reaching these conclusions, the district court explained:

The district court recounted the testimony of NCAA expert witness Dr. J. Michael Dennis, who conducted a survey of consumer attitudes concerning college sports in 2013. The court found that "[w]hat Dr. Dennis's survey does suggest is that the public's attitudes toward student-athlete compensation depend heavily on the level of compensation that student-athletes would receive." Id. at 1000-01. It noted that this conclusion "is consistent with the testimony of the NCAA's own witnesses, including [Stanford athletic director Bernard] Muir and Mr. Pilson, who both indicated that smaller payments to student-athletes would bother them less than larger payments." Id. at 1001.

The district court determined that "the evidence presented at trial suggests that consumer demand for FBS football and Division I basketball-related products is not driven by the restrictions on student-athlete compensation but instead by other factors, such as school loyalty and geography." Id. The court therefore concluded that:

Id.

The district court's findings of fact provide that one procompetitive benefit of the current rule is that restricting large payments to student-athletes plays a limited role in preserving the popularity of the NCAA's products. In the context of this antitrust suit, the concept of "amateurism" is useful only to the extent that it furthers this goal. In terms of antitrust analysis, amateurism is relevant only insofar as popular demand for college sports is increased by consumer perceptions of and desire for amateurism. Viewed through the antitrust lens, it is consumer desire that we must credit; not the NCAA's preferred articulation of the term.4

Plaintiffs are not required, as the majority suggests, to show that the proposed alternatives are "virtually as effective" at preserving the concept of amateurism as the NCAA chooses to define it. Indeed, this would be a difficult task, given that "amateurism" has proven a nebulous concept prone to ever-changing definition. See O'Bannon, 7 F.Supp.3d at 973-75 (describing the ways that the NCAA's definition of amateurism has changed over time). Even today, the NCAA's conception of amateurism does not fall easily into a bright line rule between paying student-athletes and not paying them. Tennis players are permitted to receive payment of up to $10,000 per year for playing their sport. A tennis player who begins competing at a young age could presumably earn upwards of $50,000 for playing his sport and still be considered an amateur athlete by the NCAA.5

The NCAA insists that consumers will flee if student-athletes are paid even a small sum of money for colleges' use of their NILs. This assertion is contradicted by the district court record and by the NCAA's own rules regarding amateurism. The district court was well within its right to reject it. Division I schools have spent $5 billion on athletic facilities over the past 15 years. The NCAA sold the television rights to broadcast the NCAA men's basketball championship tournament for 12 years to CBS for $10.8 billion dollars. The NCAA insists that this multi-billion dollar industry would be lost if the teenagers and young adults who play for these college teams earn one dollar above their cost of school attendance. That is a difficult argument to swallow. Given the trial evidence, the district court was well within its rights to reject it.

III

The national debate about amateurism in college sports is important. But our task as appellate judges is not to resolve it. Nor could we. Our task is simply to review the district court judgment through the appropriate lens of antitrust law and under the appropriate standard of review. In the end, my disagreement with the majority is founded on the appropriate standard of review. After an extensive bench trial, the district court made a factual finding that payment of $5,000 in deferred compensation would not significantly reduce consumer demand for college sports. This finding was supported by extensive testimony from at least four expert witnesses. There was no evidence to the contrary. Therefore, on this record, I cannot agree with the majority that the district court clearly erred when it determined that paying student-athletes up to $5,000 per year would be "virtually as effective" at preserving the pro-competitive benefits of the current rule. Therefore, I would affirm the district court in all respects.

For these reasons, I concur in part and dissent in part.