Patricia Burdett sued Robert Miller, charging a violation of the RICO statute (Racketeer Influenced and Corrupt Organizations, 18 U.S.C. § 1962) and, in a pendent count, a breach of fiduciary duty under Illinois law. After a bench trial, the district judge awarded Burdett damages in excess of $600,000 (after trebling), and later an attorney's fee of some $125,000. Miller challenges both awards.

Burdett is a sales representative for a typography firm and a highly successful salesperson, but she is unsophisticated about investing and her modest stock portfolio is managed by her stock broker. Miller is a certified public accountant, a professor of accounting at Northwestern University, and the owner of his own accounting firm. Burdett and Miller met and became friends in 1979 when Burdett enrolled in a course that Miller taught. She hired him to prepare her income tax returns. They would have lunch occasionally and discuss both business and personal matters. In 1983 Burdett's income soared to $200,000 and she asked Miller for advice on how she might minimize the tax bite. He made a number of suggestions, including that she invest in tax shelters. In the ensuing two years he steered her to a series of shelters sponsored by corporations controlled by three acquaintances of his, Mark George, Tim McDonald, and Tom Fox. When advising her to invest in the first of these ventures he did not tell her that it was the group's first venture, that investment units in it would be unmarketable, and that the two units he was urging her to buy (at $10,000 apiece) represented a third of the total investment in the project. She not only bought the two units but at his further urging executed a promissory note for $20,000 to the tax shelter, secured by a letter of credit that she obtained from a bank; he assured her that she would never be asked to make good on the letter of credit. She received no prospectus or other written information about the project. The other ventures were broadly similar, though in one Miller sold Burdett his own shares without disclosing that he was the owner. The house of cards collapsed in 1986. George, McDonald, and Fox fled to Canada. Burdett's investment in all the ventures was wiped out, and in addition she was forced to make good on the letter of credit. She lost a total of $200,000 before subtracting the tax benefits that the tax shelters generated for her and the tax savings that she obtained by being able to write off the losses from the fraud against her income.

The RICO statute forbids persons associated with an enterprise to conduct, or conspire to conduct, that enterprise's affairs through a pattern of "racketeering activity," defined as the commission of two or more violations of any of a number of specified statutes, including federal securities statutes. 18 U.S.C. § 1962. Burdett's complaint charged Miller with having conducted the affairs of an enterprise defined as Miller plus his accounting firm through a pattern of racketeering activity consisting of misleading statements and omissions that violated federal securities laws. The district judge found, however, that while Miller and his three associates, George, McDonald, and Fox, had indeed committed numerous violations of those laws, the accounting firm had not been involved in the shenanigans at all, so that there was no "enterprise" consisting of Miller and his firm of which it could be said that Miller had conducted the enterprise's activities through a pattern of racketeering activity. Burdett does not challenge this finding. The judge went on, however, to find that Miller plus his three associates in the fraud constituted a RICO enterprise the affairs of which Miller had conspired to conduct through a pattern of racketeering activity consisting of the securities violations.

Miller argues that there is insufficient evidence that the four individuals were leagued in an "enterprise" within the meaning of the statute. They may have conspired to defraud Burdett, but not every conspiracy is an enterprise. That is true, United States v. Neapolitan, 791 F.2d 489, 499-500 (7th Cir.1986); Hartz v. Friedman, 919 F.2d 469, 471 (7th Cir.1990), but there was enough evidence of structure to justify a finding of an informal enterprise. The statute is aimed not only at formal enterprises such as corporations, labor unions, and government departments controlled by racketeers (in the special sense that the statute gives this term) but also at criminal gangs, which have a less formal, a less reticulated and differentiated structure. There must be some structure, to distinguish an enterprise from a mere conspiracy, but there need not be much. Here that structure is supplied by the continuity of the informal enterprise--it was not an ad hoc affair but persisted as an identifiable entity through time--and by the differentiation of roles within it. Miller acted as the respectable front man to enlist the mark, Patricia Burdett, while the other three created and operated over a period of two years a series of tax shelters designed to separate the mark from her money. The enterprise so constituted resembled the informal enterprise between a criminal defense lawyer and local police officials that we held sufficient for RICO liability in United States v. Masters, 924 F.2d 1362, 1367 (7th Cir.1991).

But we agree with Miller that the enterprise was injected into the litigation too late. The complaint specified an enterprise consisting of Miller and his accounting firm. So did the pretrial order and the pretrial briefs. Of course much evidence came in during the trial concerning the dealings between Miller and his three associates, because Miller was charged with conspiring with them to conduct the enterprise consisting of himself and his firm through a pattern of racketeering activity. But there was no mention of an enterprise consisting of the four conspirators themselves. And no motion to amend the pleadings. Burdett moved for a directed finding of liability on the RICO count; the motion did not mention the new enterprise. The district judge did not permit closing argument or post-trial briefs, so the issue couldn't come in at that stage. In fact the enterprise was not changed until the district judge, after the trial, on her own initiative dropped a footnote to that effect in the opinion announcing her findings of fact and conclusions of law.

That was too late. It is true that when "issues not raised by the pleadings are tried by express or implied consent of the parties, they shall be treated in all respects as if they had been raised in the pleadings." Fed.R.Civ.P. 15(b). This is so even if there is no motion to amend the pleadings; indeed, that's the point of Rule 15(b). The key word, however, is "consent." 6A Charles Alan Wright, Arthur R. Miller & Mary Kay Kane, Federal Practice and Procedure § 1493, at pp. 19-20 (2d ed. 1990). The district judge inferred the parties' consent from the fact that both had presented extensive evidence concerning Miller's connection with George, McDonald, and Fox. But that presentation was not a manifestation of consent to a charge that the four were leagued in an enterprise. Miller had no warning that evidence manifestly admissible because relevant to the conspiracy charge would also be used to establish the existence of an enterprise to which no one in the course of this litigation had alluded.

Ours is an adversarial system; the judge looks to the parties to frame the issues for trial and judgment. Our busy district judges do not have the time to play the "proactive" role of a Continental European judge. True, they want to do justice as well as merely umpire disputes, and they should not be criticized when they point out to counsel a line of argument or inquiry that he has overlooked, W.T. Rogers Co. v. Keene, 778 F.2d 334, 345 (7th Cir.1985); Morrison v. Murray Biscuit Co., 797 F.2d 1430, 1433 (7th Cir.1986), although they are not obliged to do so and (with immaterial exceptions) they may not do so when an issue has been waived. Thomas v. Indiana, 910 F.2d 1413, 1415 (7th Cir.1990). When the unfolding evidence persuaded the district judge that the plaintiff's counsel had misidentified the RICO enterprise, she could without impropriety have invited him to shift the line of his attack, and if he had taken up the invitation the defendant could have sought to parry the attack. The judge did not do this but instead changed the plaintiff's theory of the case after the time had passed for the defendant to present contrary evidence.

Burdett argues that the error was harmless because the proof of the four-man enterprise is conclusive. It is not. So informal an enterprise as the judge found here is at the limit of RICO's reach, and Miller might have presented convincing evidence that the alleged "enterprise" lacked even the minimum structure that the judge inferred from evidence not directed to the issue because it wasn't an issue.

Burdett quite properly does not argue that if the judge erred in changing the enterprise the remedy should be a new trial rather than judgment for the defendant. Save within the dispensation of Rule 15(b) and other provisions of law not cited to us, a plaintiff who fails at trial to prove an essential element of his case may not retry the case on a different theory. Gilbert v. Frank, 949 F.2d 637, 640 (2d Cir.1991); Midamar Corp. v. National-Ben Franklin Ins. Co., 898 F.2d 1333, 1338 (8th Cir.1990); Cowgill v. Raymark Industries, Inc., 832 F.2d 798, 804 (3d Cir.1987); cf. Johnson v. Allyn & Bacon, Inc., 731 F.2d 64, 73 (1st Cir.1984). Burdett does not argue that the judge was wrong to reject an enterprise consisting of Miller and his accounting firm, so that route to establishing a RICO violation is closed too.

This does not end the appeal, because the district judge also found that Miller had violated a fiduciary duty to Burdett imposed on him by Illinois law by giving her deliberately misleading investment advice. A fiduciary duty is the duty of an agent to treat his principal with the utmost candor, rectitude, care, loyalty, and good faith--in fact to treat the principal as well as the agent would treat himself. Pohl v. National Benefits Consultants, Inc., 956 F.2d 126, 128 (7th Cir.1992); Market Street Associates Limited Partnership v. Frey, 941 F.2d 588, 593 (7th Cir.1991), and cases cited there; see also Restatement (Second) of Trusts § 2, comment b (1959); id., § 170(1); Restatement (Second) of Agency § 13, comment a (1958). The common law imposes that duty when the disparity between the parties in knowledge or power relevant to the performance of an undertaking is so vast that it is a reasonable inference that had the parties in advance negotiated expressly over the issue they would have agreed that the agent owed the principal the high duty that we have described, because otherwise the principal would be placing himself at the agent's mercy. An example is the relation between a guardian and his minor ward, or a lawyer and his client. The ward, the client, is in no position to supervise or control the actions of his principal on his behalf; he must take those actions on trust; the fiduciary principle is designed to prevent that trust from being misplaced.

We have given two examples of categories of relations in which fiduciary duties are imposed (lawyer-client, guardian-ward), and the relation between an investment advisor and the people he advises is not a third. But fiduciary duties are sometimes imposed on an ad hoc basis. Klass v. Hallas, 16 Ill. 2d 161, 164, 157 N.E.2d 261, 263 (1959); Carey Electric Contracting, Inc. v. First National Bank, 74 Ill.App.3d 233, 236, 30 Ill. Dec. 104, 108, 392 N.E.2d 759, 763 (1979). If a person solicits another to trust him in matters in which he represents himself to be expert as well as trustworthy and the other is not expert and accepts the offer and reposes complete trust in him, a fiduciary relation is established.

We have emphasized knowledge and expertise but we do not mean to suggest that every expert is automatically a fiduciary. That is not the law in Illinois or anywhere else. A fiduciary relation arises only if "one person has reposed trust and confidence in another who thereby gains influence and superiority over the other." Amendola v. Bayer, 907 F.2d 760, 763 (7th Cir.1990); see also Ray v. Winter, 67 Ill. 2d 296, 304, 10 Ill. Dec. 225, 229, 367 N.E.2d 678, 682 (1977). One source of such an ascendancy, however--a source conspicuous in the lawyer-client relation--is that the agent has (or claims to have) expert knowledge the deployment of which the principal cannot monitor. Miller cultivated a relation of trust with Burdett over a period of years, holding himself out as an expert in a field (investments) in which she was inexperienced and unsophisticated. He knew that she took his advice uncritically and unquestioningly and that she sought no "second opinion" or even--until the end, when at last her suspicions were aroused--any documentary confirmation of the investments to which he steered her.

Miller could have protected himself from being deemed a fiduciary by explaining the character and circumstances of the investments to Burdett, by disclosing his stake in them to her, by seeing to it that she received prospectuses and other documents describing the risks of the investments, and if need be by advising her to seek additional investment counsel before staking large sums on these risky ventures. He did none of these things. It is true that he did not ask her simply to sign over all her wealth to him to be invested in his sole discretion. But the district court was entitled to find that he did the next best thing from his standpoint (and in doing so brought himself within the orbit of fiduciary duty). That was to invite her to accept his advice with no questions asked or answered, in reliance on his professional and professorial status, on his insight into the arcana of tax shelter investments--a technical area about which she was ignorant--and on a continuing business relationship shading into a social friendship.

We cannot say that in finding a fiduciary relation the district judge committed a clear error. Actually she did commit one clear error on the fiduciary count, and that was to apply the normal civil standard of preponderance of the evidence, rather than the higher standard of proof--proof by clear and convincing evidence--that Illinois requires to establish the existence of a fiduciary duty outside of the per se categories such as lawyer-client and guardian-ward. Perry v. Wyeth, 25 Ill. 2d 250, 253, 184 N.E.2d 861, 863 (1962); Bosak v. McDonough, 192 Ill.App.3d 799, 805, 139 Ill. Dec. 917, 921, 549 N.E.2d 643, 647 (1989); Carey Electric Contracting, Inc. v. First National Bank, supra, 74 Ill.App.3d at 236, 30 Ill. Dec. at 108, 392 N.E.2d at 763; Amendola v. Bayer, supra, 907 F.2d at 764. But Miller waived the error in the district court by failing to ask the judge for the higher standard. The normal standard of proof in a civil case is, of course, proof by a preponderance of the evidence, not proof by clear and convincing evidence. (Even many fraud cases are governed by the normal standard. Spitz v. Commissioner, 954 F.2d 1382, 1383 (7th Cir.1992).) A federal district judge cannot be presumed to carry around in his head every esoteric rule of the law of the state in which he happens to sit. If the parties do not mention the standard of proof in a civil case, the district judge is bound to apply the normal civil standard, just as he will apply the substantive law of the forum state if the case is a diversity case and neither party argues choice of law. Wood v. Mid-Valley Inc., 942 F.2d 425, 426-27 (7th Cir.1991). The preponderance standard, and the forum state's substantive law, are the default rules to be applied in such situations in the absence of objection. And that is what happened here. The parties did not mention the burden of proof in any of their filings in the district court; the pretrial order didn't mention it; so naturally the judge applied the preponderance standard.

Even in Miller's brief in this court there is no suggestion that the judge applied the wrong standard. The brief does remark in passing that the evidence for a fiduciary relation was not clear and convincing, but does not acknowledge that the judge applied a different standard. Miller's lawyer may not have believed that there was a practical difference between preponderance and clear and convincing in the circumstances of this case, although in many cases the difference is critical. See, e.g., Spitz v. Commissioner, supra. Nothing in his brief save the bare mention of the clear and convincing standard indicates such a belief, and we do not think that that glancing reference was enough to preserve the issue, Freeman United Coal Co. v. Office of Workers' Compensation Programs, 957 F.2d 302, 304-05 (7th Cir.1992); United States v. Berkowitz, 927 F.2d 1376, 1384 (7th Cir.1991); Zelazny v. Lyng, 853 F.2d 540, 542 n. 1 (7th Cir.1988), or to alert the other side that it must argue that the issue had indeed been waived in the district court in order to avoid being found to have waived the waiver issue, as in Thomas v. Indiana, 910 F.2d 1413, 1415 (7th Cir.1990). We think the parties must be taken to have consented to having the trial conducted and findings made under the usual standard. It would have been better had the judge allowed closing argument or post-trial briefs, for then Miller would have had another opportunity to point out that the judge should apply the higher standard of proof in evaluating the evidence. Yet he does not argue that he would have mentioned it. In fact it seems that the parties were oblivious to the issue throughout the trial.

The remaining question is the measure of damages for the breach of fiduciary duty. Burdett argues and the district judge agreed that the damages should not be reduced by the amount of money that Burdett was able to save by deducting the loss of her investment from her income on her tax returns. This is the correct general rule. Cereal Byproducts Co. v. Hall, 16 Ill.App.2d 79, 81-82, 147 N.E.2d 383, 384, aff'd 15 Ill. 2d 313, 155 N.E.2d 14 (1958); Fullmer v. Wohlfeiler & Beck, 905 F.2d 1394, 1402 (10th Cir.1990); Danzig v. Grynberg & Associates, 161 Cal. App. 3d 1128, 1139-40, 208 Cal. Rptr. 336, 343 (1984); Western-Realco Limited Partnership 1983-A v. Harrison, 791 P.2d 1139, 1147 (Colo.App.1989); Coty v. Ramsey Associates, Inc., 149 Vt. 451, 546 A.2d 196, 204 (1988); cf. Randall v. Loftsgaarden, 478 U.S. 647, 658-60, 106 S. Ct. 3143, 3150-51, 92 L. Ed. 2d 525 (1986). Suppose, to take a simpler case, that Miller had tortiously destroyed Burdett's Ming vase worth $10,000 and Burdett had deducted this amount as a casualty loss on her federal income tax return, garnering a tax saving of $3,000. Miller could not in the ensuing tort suit deduct the $3,000 from the damages due Burdett. The tort caused a harm of $10,000, and the fact that the plaintiff was able to lay off a part of the harm on someone else--the taxpayer--is not a good reason to cut down the tortfeasor's damages. It is true that the result is a windfall to the plaintiff, but this is better than an equivalent windfall to the tortfeasor, and the plaintiff's windfall could be eliminated by a long-overdue change in the tax law to make all damages awards that replace taxable income themselves taxable. Actually we don't know how the Internal Revenue Service will treat Burdett's award of damages, because while damages in personal injury cases are not taxable even when they replace lost income (as in a case where the injury disables the plaintiff from working), 26 U.S.C. § 104(a)(2), we do not know whether the breach of a fiduciary relation would be deemed a personal injury in the sense of the statute (see generally Boris I. Bittker & Martin J. McMahon, Federal Income Taxation of Individuals p 7.1, at pp. 7-5 to 7-7 (1988))--although it is a fair guess that the Internal Revenue Service will take the position that it is not when the only loss is financial. But the only important point here is that the tax treatment of the damages award is irrelevant to the defendant's liability; it is a matter between the plaintiff and the government.

This case, however, involves two types of tax benefit and they have different implications for liability. One is simply the benefit that Burdett obtained by being able to take a deduction for the loss from the fraud that wiped out her extensive investment in the tax shelters. That is the identical benefit as in our hypothetical case of conversion. The other is the anticipated tax benefits of the investments. The investments were, after all, tax shelters. They were intended to generate tax benefits. To the extent they did, this was a gain to Burdett not offset by a loss to the taxpayer, because the taxpayer intends as it were to incur a loss on a lawful tax shelter deal. Suppose that in exchange for an investment of $25,000 in a nonfraudulent tax shelter Burdett would have obtained $20,000 in tax savings and $10,000 in return of principal (plus interest), for a net profit of $5,000. As a result of fraud, the tax shelter generates the anticipated $20,000 in tax savings but her principal is wiped out and she gets no interest. Her loss is $5,000 (remember that she invested $25,000)--not $30,000--just as if, in an investment of $25,000 that involved no tax angle, she had been repaid $20,000 before the defendant absconded with the balance of the investment. (It is not, however, $5,000 minus the tax savings she obtained from deducting the $5,000 loss on her income tax return.) As a matter of fact, one of Burdett's investments was a $100,000 loan, half of which was repaid. She does not claim that half as damages. It is equally unreasonable for her to claim as losses the anticipated, and realized, tax benefits of the tax shelters.

The loan example shows that the tax character of those benefits is actually irrelevant. The essential point is that with regard to benefits of whatever sort that are promised and received, there is no injury, so no basis for a claim of damages. The fiduciary count must therefore be remanded to enable a partitioning of the two tax benefits that Burdett received.

We emphasize that in speaking in the tax-shelter context of "benefits ... promised and received" we mean the present value of the net anticipated tax benefits, not the tax writeoff as such. Tax shelters generally defer rather than eliminate tax liability, so that the net benefit to the taxpayer is not the entire writeoff but rather the time value of the deferred tax and the possibility of lower tax rates when the tax eventually comes due. These may be substantial benefits but they cannot be automatically equated to the amount of tax writeoff enabled by the shelter.

Although the attorney's fee award falls with the RICO count (the only count in the case for which attorney's fees could be awarded), for the guidance of the bench and bar of this circuit we point out that the judge erred in multiplying the plaintiff's attorneys' hourly billing rates by two, or any other number larger than one, in deciding how large a fee to award. "Multipliers" are for classes of case where in their absence plaintiffs might not be able to induce competent counsel to take their case. Usually these are small-money or no-money (i.e., equitable) cases that are costly to try and likely to fail even when meritorious, or class-action cases, in which a multiplier may take the place of the contingent-fee contract that the lawyers for the class cannot negotiate because they are not actually retained by--they do not make a contract with--the members of the class. Civil litigation under RICO is ordinary commercial litigation and more--from a plaintiff's standpoint--since a winning plaintiff in ordinary commercial litigation does not get treble damages plus a court-awarded attorney's fee. There is no reason to offer plaintiffs extraordinary incentives. Risk of loss is not alone enough, given the possibility of contingent-fee arrangements. Kirchoff v. Flynn, 786 F.2d 320, 323-26 (7th Cir.1986); McKinnon v. City of Berwyn, 750 F.2d 1383, 1392 (7th Cir.1984); and see the plurality and concurring opinions in Pennsylvania v. Delaware Valley Citizens' Council for Clean Air, 483 U.S. 711, 107 S. Ct. 3078, 97 L. Ed. 2d 585 (1987); also King v. Palmer, 950 F.2d 771 (D.C.Cir.1991) (en banc). We hold that ordinary billing rates are plenty for fee awards in RICO cases; the multiplication of those rates should not be permitted.

The judgment is reversed with directions to enter judgment for the defendant on the RICO count, vacate the award of attorneys' fees, and conduct further proceedings consistent with this opinion on the damages to which the plaintiff is entitled for the defendant's breach of his fiduciary duty to her.