Unpublished opinions are not binding precedent in this circuit.

PER CURIAM:

Brian Conner appeals a district court order denying his motion for relief under 28 U.S.C. § 2255 based on his claim that he received ineffective assistance of counsel at sentencing. Finding no reversible error, we affirm.

I.

In 1990, Conner, a certified emergency medical technician, became the owner, operator, and president of Convalescent Transports, Inc. ("CTI"). The North Carolina corporation was in the business of providing ambulance and wheelchair transportation services for, among others, medical patients covered by Medicare and Medicaid. Both Medicare and Medicaid have explicit regulations concerning the conditions under which they will provide reimbursement for ambulance transportation services. The regulations essentially require a showing of medical necessity.

Sometime after October 1991, Conner began submitting false claims to Medicare and Medicaid for ambulance transportation services that CTI rendered. Employees were told to transport all dialysis patients by ambulance, and employees were instructed to falsify the ambulance call reports to make it appear that transportation by ambulance was medically necessary.

On the basis of this conduct, Conner was charged by superseding indictment with 350 counts of health care fraud, conspiracy to commit health care fraud, and obstruction of the criminal investigation of health care fraud. See 18 U.S.C. §§ 371, 1347, 1518. The district court dismissed four counts on the government's motion, and the case proceeded to trial. A jury found Conner guilty of all remaining counts.

After the convictions, a presentence report ("PSR") was prepared. It included a recommendation for a two-level enhancement for abuse of a position of trust, see U.S. Sentencing Guidelines Manual § 3B1.3 (2005), and a loss determination of more than $2,500,000 but not more than $7,000,000, which would have resulted in an 18-level enhancement to the applicable base offense level, see U.S.S.G. § 2B1.1(b)(1)(J). Conner objected to both of these enhancements.

At Conner's sentencing hearing, the government presented detailed testimony concerning the loss amount. The government showed that CTI had received $6,822,690.54 between 1997 and 2002 in reimbursements on 35,328 claims for non-emergency dialysis transports. The government determined how much of these payments constituted the government's loss by sampling and extrapolation. In this regard, "RAT-STATS," a computer program developed by the United States Health and Human Services' Office of Inspector General, was used to perform three different steps: (1) determining the sample size needed to represent the data; (2) randomly generating the list of particular claims to review as part of the sample; and (3) extrapolating from the reviewed claims. The first two steps produced a sample of 230 of the claims paid to CTI for transportation of patients in connection with dialysis. Government agents then attempted to retrieve the records corresponding to these claims but were only able to find documentation for 165 of them. These records were, in turn, reviewed by a medical fraud investigator.

Treating the 65 claims with no documentation ("the missing-records claims") as invalid, the investigator testified that of the 230 claims, only 14 were justified by medical necessity, and the average overpayment was $188.03 per claim. Multiplying the per-claim average by the total number of claims (35,328) yielded a total overpayment of $6,642,582 for the 230 claims. Based on this amount, the government's statistics expert, Suzanne Moody, testified that RAT-STATS indicated that, with a 90% confidence interval, the range of overpayment was between $6,330,298 and $6,954.866. Moody also testified alternatively that if the 65 missing-records claims were treated as fully valid, the lower end of the overpayment range would drop to $3,738,866.

G. Christopher Kelly, who represented Conner at trial and at sentencing, raised several objections to the government's loss amount, including arguments that the calculations were partly based on claims that were not part of the scheme and that the loss amount included all payments made for non-medically-necessary services rather than only those payments that had been procured by fraud. Kelly also maintained that the government's extrapolation methodology was not reliable and specifically focused on the missing-records claims. Kelly questioned Moody regarding how changing the overpayment amounts of only roughly 28% of the claims (65 out of 230) could reduce the estimated loss amount by about 41%. Kelly later argued to the district court that Moody had not provided a satisfactory explanation.

Kelly also questioned the government's witnesses concerning the government's inability to locate the documentation regarding the missing-records claims. Kelly subsequently asserted that the government had not exercised due diligence in trying to locate the 65 missing-records claims and that this was an additional reason that the government's methodology was flawed. Kelly added that the government had the opportunity to do more sampling and make its estimates much more reliable and precise but had failed to do so. Kelly contended that the appropriate loss amount would be the amount proven by the evidence presented at trial, which he claimed would have been less than $30,000.

In the end, the district court accepted the reliability of the sampling process. However, the court also accepted Kelly's argument that the government had failed to show that it exercised sufficient diligence in searching for the documentation related to the 65 missing-records claims or alternatively in reviewing substitute claims. In an effort to ensure that Conner was not penalized by the government's lack of diligence, the court calculated the loss based on the assumption that the 65 missing-records claims were entirely valid. The district court also agreed with Kelly that trips transporting patients to and from hospitals were not properly included and thus counted such claims as valid as well. With those two assumptions, the court found a loss amount of $3,613,165.00, nearly $3 million less than the government's proposed amount. Unfortunately for Conner, this quite substantial reduction still left him the same loss range of more than $2.5 million and not more than $7 million. See U.S.S.G. § 2B1.1(b)(1)(J). Thus, the associated 18-level enhancement and the two-level abuse-of-position-of-trust enhancement, which the court also applied over Kelly's objection, left Conner with a total offense level of 32. This level, in conjunction with Conner's Criminal History Category of I, yielded a guideline range of 121 to 151 months' imprisonment. The district court sentenced Conner at the highest point in that range.

We affirmed Conner's sentence on appeal, holding, as is relevant here, that the government's extrapolation provided adequate support for the district court's loss determination. See United States v. Conner, 262 F. App'x 515, 518-19 & n.5 (4th Cir. 2008). In so doing, we specifically rejected an argument by Conner that the sampling process was not adequately shown to be random. See id. at 519 n.5.

Conner subsequently filed a motion to vacate his judgment or set aside his sentence pursuant to 28 U.S.C. § 2255, alleging that Kelly was ineffective at trial and at sentencing. As is relevant here, Conner alleged that Kelly was constitutionally ineffective at sentencing in failing to offer expert testimony to challenge the government's statistical evidence and in failing to argue for a reduction in his loss amount for the value received by the government from CTI's provision of services.

In support of his motion, Conner offered evidence from two expert witnesses who challenged the reliability of the government's loss calculation methodology on a number of bases. They contended that the sample size was too small, and that "deleting" the 65 missing-records claims called into question the randomness of the sample. J.A. 96. Conner's experts also pointed out additional flaws in planning, sample design, conduct of the actual sample, data analysis, and the presentation of the final results.

Conner also presented an affidavit and testimony from Joseph B. Cheshire V, an attorney who represented Conner in his direct appeal. Cheshire opined that Kelly should have hired an expert to study the government's theory of how to calculate the loss and that Kelly should have tried to minimize the loss amount by identifying benefits that the government received from CTI's provision of services.

Additionally, Conner presented an affidavit and testimony from Keith A. Williams, an attorney who represented one of Conner's co-defendants. As is relevant here, Williams opined that Kelly was constitutionally deficient in not arguing for a reduction in loss amount based on the value to the government of services CTI rendered. He also asserted that an expert "could have provided some assistance in preparing for and presenting" arguments at sentencing. J.A. 278.

The government also presented an affidavit and testimony from Kelly. He noted that he "had had civil cases that dealt with [RAT-STATS] and that type of thing so [he] knew some of the issues that could occur with those." J.A. 331. He also stated that he researched the validity of the use of RAT-STATS in federal jurisdictions. He admitted, however, that "it would have been helpful to have had an expert at sentencing." J.A. 332. He explained that Conner's inability to pay for an expert figured in to his decision not to seek one, but he conceded that he did not seek court appointment of an expert.

After considering the evidence before the court, a federal magistrate judge recommended denying each of Conner's claims. Conner subsequently filed objections with the district court, but the district court overruled the objections, adopted the magistrate's findings and analysis, and denied the motion to vacate. See Conner v. United States, Nos. 4:04-CR-27-FL-2, 4:09-CV-96-FL, 2010 WL 4484397, at *8 (E.D.N.C. Nov. 1, 2010). The court also granted Conner a certificate of appealability. See id.

II.

Conner first argues that Kelly was ineffective in failing to offer expert testimony challenging the statistical and random sampling methodology used by the government at sentencing ("the expert claim"). We disagree.

In considering the denial of a § 2255 motion, we review a district court's factual findings from an evidentiary hearing for clear error, and we review de novo mixed issues of law and fact, such as whether established facts demonstrate a deficient performance by counsel. See United States v. Roane, 378 F.3d 382, 395 (4th Cir. 2004).

Claims of ineffective assistance of counsel are reviewed under the standards of Strickland v. Washington, 466 U.S. 668 (1984), and its progeny. To be entitled to relief, Conner must demonstrate "that counsel's performance was deficient" and that "the deficient performance prejudiced the defense." Id. at 687. To demonstrate inadequate or deficient performance, Conner "must show that counsel's representation fell below an objective standard of reasonableness" measured by "prevailing professional norms." Id. at 688. Our application of this standard "must be highly deferential," and we "must indulge a strong presumption that counsel's conduct falls within the wide range of reasonable professional assistance." Id. at 689. Moreover, counsel's performance must not be judged with the benefit of hindsight; rather, we consider "counsel's perspective at the time" of the representation in question. See id. To demonstrate prejudice, Conner "must show that there is a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different." Id. at 694.

While Strickland applies in both capital and noncapital sentencing proceedings, see, e.g., Glover v. United States, 531 U.S. 198, 202-04 (2001) (applying Strickland test in noncapital case), what constitutes deficient performance can differ depending on the type of proceeding. The ABA's Criminal Justice Standards, which can serve as a tool for evaluating the reasonableness of counsel's representation, see Wiggins v. Smith, 539 U.S. 510, 524 (2003), indicate that defense counsel in a noncapital sentencing proceeding should (1) promptly investigate the circumstances and facts relevant to sentencing, (2) present the court with any basis that will help achieve an outcome favorable to the defense, and (3) supplement or challenge information provided in any presentence report. See ABA Criminal Justice Standards 4-4.1(a) and 4-8.1(b).

In this case, to fully understand the district court's findings, we also must consider the magistrate's analysis of Conner's claims. The magistrate rejected Conner's assertion that Kelly's representation at sentencing was constitutionally deficient and concluded that, even assuming that Kelly was deficient in the ways Conner alleged, Conner could not establish a reasonable probability he would have received a more lenient sentence had Kelly taken the steps Conner now says he should have taken. As relates to the expert claim, Conner objected to the magistrate's conclusions regarding both Strickland prongs. Regarding the second prong, Conner maintained that if Kelly had been able to convince the district court that the government's extrapolation was invalid, then the extrapolation "would have been required to be re-done" and there was "more than a reasonable likelihood that a different sentence would have" resulted. J.A. 727.

The district court rejected both of these arguments. Regarding the second prong, the court reasoned that even if Kelly's presentation of expert testimony would have convinced the district court to require the government to take another sample and redo its extrapolation, Conner made no showing that he would have ended up with a smaller loss amount than he did having the court assume that the 65 missing-records claims were actually completely legitimate. See Conner, 2010 WL 4484397, at *5.

In his initial brief to us, Conner challenged the district court's conclusion that Conner failed to establish that Kelly's representation was rendered deficient by his decision not to present expert testimony challenging the government's extrapolation methodology. However, he did not address the district court's determination that he could not show a reasonable probability that redoing the government's analysis would have led to a more lenient sentence. Conner addressed this issue for the first time in his reply brief, claiming he demonstrated prejudice because an expert could have shown that the government's extrapolation was flawed, and thus caused the district court to reject it and determine the loss amount by considering only the trial evidence, which would have supported a loss finding of less than $30,000.

This prejudice argument is not properly before us for two reasons. First, inherent in Conner's new argument is a contention that the district court erred in assuming that the government would have been allowed to "redo" its statistical analysis if the district court agreed with Kelly's expert. Not only did Conner not make this argument to the district court, he affirmatively argued the opposite — that had Kelly employed an expert at sentencing, the government would have been required to redo its statistical analysis. Accordingly, the argument is waived. See First Va. Banks, Inc. v. BP Exploration & Oil, Inc., 206 F.3d 404, 407 n.1 (4th Cir. 2000) ("Because neither of these arguments were raised below, we decline to consider them on appeal."). Additionally, even had Conner raised this argument in the district court, he would have waived it by failing to raise it in his initial brief. See Cavallo v. Star Enter., 100 F.3d 1150, 1152 n.2 (4th Cir. 1996) ("[A]n issue first argued in a reply brief is not properly before a court of appeals.").

In any event, even assuming arguendo that Conner properly preserved a challenge to the district court's ruling that he could not establish prejudice from his asserted deficiency, we agree with the district court that Kelly's performance was constitutionally adequate.

As the district court noted, Kelly was quite active during sentencing:

Conner, 2010 WL 4484397, at *4. As the court explained,

Id. at *4 n.7 (citations omitted).

Kelly was well aware that extrapolations similar to the government's in this case had "been upheld numerous times in the federal courts." J.A. 115. In light of that fact, it was his judgment "that under the facts of this case, and the time and financial limitations that the family placed on [him], that calling an expert at sentencing would [not] have been possible or beneficial." J.A. 115.

Based on all of these facts, we simply cannot conclude that Conner has rebutted the "strong presumption" that Kelly's performance was constitutionally reasonable. Strickland, 466 U.S. at 689. By vigorously exploiting the government's lack of diligence in searching for the documentation for the missing-records claims (and by challenging the relevance of claims relating to transportation to and from hospitals), Kelly obtained a loss amount for his client that was millions of dollars less than the government sought. Although his efforts did not yield a guideline range lower than the government had proposed, we conclude that his performance was within prevailing professional norms.

For similar reasons, we do not believe that Conner was prejudiced by Kelly's decision not to call an expert. Even had Kelly been able to use an expert to persuade the district court to reject the government's methodology, it is sheer speculation to conclude that the ultimate result would have been a loss determination of less than $2,500,000.

III.

Conner also argues that the district court erred in rejecting his claim that Kelly was constitutionally ineffective at sentencing because he did not argue that the loss amount should be reduced by the value of benefits that the government received by virtue of the provision of CTI's services ("the benefits received claim"). See U.S.S.G. § 2B1.1 cmt. n.3(F)(ii) ("In a case involving government benefits (e.g., grants, loans, entitlement program payments), loss shall be considered to be not less than the value of the benefits obtained by unintended recipients or diverted to unintended uses, as the case may be."); United States v. Dawkins, 202 F.3d 711, 715 (4th Cir. 2000) ("[W]e advise the district court to consider loss as the difference between the amount of benefits [the defendant] actually received and the amount he would have received had he truthfully and accurately completed the . . . forms."). For example, he maintains that even with regard to claims in which ambulance transportation was not medically necessary, non-emergency transport may have been necessary, and Kelly should have argued that Conner was entitled to a credit for the value of such transport.

We cannot conclude that counsel's representation was constitutionally deficient. As noted, Kelly made numerous arguments disputing the government's loss calculation, and he indeed achieved a significant victory in establishing the government's lack of diligence in searching for the missing-records documentation. As a result of his challenge to the quality of the government's investigation, and his arguments as to how the inclusion of those claims would undercut the reliability of the government's extrapolation, the loss amount was based on the assumption that the missing-records claims, which made up more than a quarter of the entire sample, were 100% valid. One need not be an expert in statistics to recognize that that change resulted in a significant reduction of Kelly's loss amount even if it was not enough to reduce his guideline range.

Our review of the record makes clear that even if there were good arguments that Kelly did not make, there were many good ones that he did make and indeed made effectively. See Mickens v. Taylor, 240 F.3d 348, 363 (4th Cir. 2001) (en banc) ("The Sixth Amendment guarantee of counsel does not guarantee an ideal or perfect representation."). Given the "highly deferential" standard by which we judge counsel's performance, we cannot conclude that Kelly's representation at sentencing was constitutionally deficient. Strickland, 466 U.S. at 689.

IV.

In sum, for the foregoing reasons, we affirm the district court's order denying Conner's § 2255 motion.

AFFIRMED.