Though rarely the subject of sustained scholarly attention, the law concerning statutes of limitations fairly bristles with subtle, intricate, often misunderstood issues, as is well illustrated by the appeal in this ease. The appeal comes about as follows. Wolin and Wortman, the trustees of a pension plan governed by the Employee Retirement Income Security Act of 1974, 29 U.S.C. §§ 1001 at seq., brought suit against a broker named Mackevieh and the broker’s employer, E.F. Hutton, succeeded by Smith Barney. The suit charged that Mackevieh, and so by the principle of respondeat superior his employer as well, had violated the fiduciary duties imposed upon him by the Act by advising the trustees to make risky, illiquid investments while assuring them that the investments were liquid and safe. An initial question was whether Mackevieh was a fiduciary. Not everyone who provides investment advice to an ERISA plan is. A broker who merely touts stocks to the plan is not. Farm King Supply, Inc. v. Edward D. Jones & Co., 884 F.2d 288, 292 (7th Cir.1989). The statute, 29 U.S.C. § 1002(21)(A)(ii), as glossed by the Department of Labor’s regulations, 29 C.F.R. § 2510.3-21, and by the cases, such as Farm King Supply and Thomas, Head & Greisen Employees Trust v. Buster, 24 F.3d 1114, 1117-20 (9th Cir.1994), requires that the investment advisor, in order to be deemed a fiduciary, with all that that status implies, be rendering advice pursuant to an agreement, be paid for the advice, and have influence approaching control over the plan’s investment decisions. There is no doubt that Mackevieh satisfied these criteria. He was therefore a fiduciary of the plan, obligated to render prudent advice to the trustees and, even more clearly, to refrain from deliberately misleading them. 29 U.S.C. § 1104(a)(1); In re Unisys Savings Plan Litigation, 74 F.3d 420, 434, 442 (3d Cir.1996); Burdett v. *850Miller, 957 F.2d 1375, 1381 (7th Cir.1992). The district judge nevertheless granted summary judgment for the defendants on the ground that the suit was untimely.
The plaintiff in a suit under ERISA against a fiduciary has six years after the “breach or violation” in which to sue, or three years after the plaintiff “had actual knowledge of the breach or violation,” whichever comes first, “except that in the case of fraud or concealment, [the suit] may be commenced not later than six years after the date of discovery of such breach or violation.” 29 U.S.C. § 1113(a). (There is no subsection (b), so from here on we shall drop the reference to (a).) The meaning of the “actual knowledge” clause is reasonably clear, as we shall see; the clause modifies and supplants the judge-made doctrine of equitable tolling. The “except” clause, however, is deeply puzzling. The purpose could be to codify another judge-made doctrine, that of equitable estoppel in its aspect as a defense to a plea of the statute of limitations — whether the whole of the doctrine, or the part of it that goes by the name “fraudulent concealment,” or perhaps a concept of fraudulent concealment that crosses the boundary that separates equitable estoppel from equitable tolling. E.g., Martin v. Consultants & Administrators, Inc., 966 F.2d 1078, 1093-94 (7th Cir.1992); Radiology Center, S.C. v. Stifel, Nicolaus & Co., 919 F.2d 1216, 1220 (7th Cir.1990); J. Geils Band Employee Benefit Plan v. Smith Barney Shearson, Inc., 76 F.3d 1245, 1252 (1st Cir.1996). To establish equitable estoppel a plaintiff must show that the defendant had taken steps to prevent the plaintiff from filing suit within the statutory period. Cada v. Baxter Healthcare Corp., 920 F.2d 446, 450-51 (7th Cir.1990). The defendant might have promised the plaintiff not to plead the statute of limitations, or might — this would be fraudulent concealment — have concealed his identity or other facts that the plaintiff needed in order to be able to file suit.
Among the puzzles of the “except” clause, besides its use of the term “fraud or concealment” rather than “fraudulent concealment” or “equitable estoppel,” is that the six-year deadline for suing that it establishes seems both too short and too long. Too short because after the prospective plaintiff discovers the wrongful act the defendant may take steps to prevent him from suing — for example by falsely promising not to plead the statute of limitations — that may be effective for more than six years. Too long because after the plaintiff discovers the wrongful act, and assuming the defendant immediately desists from any efforts to prevent the plaintiff from suing, the plaintiff has a full six years to sue even though the preceding clause of the statute gives a plaintiff only three years to sue after he obtains actual knowledge of the wrongful act. Once the defendant has stopped trying to obstruct the bringing of the suit and his previous obstructive efforts have dissipated, the plaintiff will be in exactly the same position that he would have occupied had he, at just the same time, obtained actual knowledge of the wrongful act.
The explanation may be that Congress decided to substitute a fixed period of years for the open-ended judge-made doctrines of fraudulent concealment and equitable estop-pel, thus promoting (as in statutes of repose) certainty of legal obligation. Larson v. Northrop Corp., 21 F.3d 1164, 1172 (D.C.Cir.1994); Martin v. Consultants & Administrators, Inc., supra, 966 F.2d at 1103-04 (concurring opinion). If there is fraudulent concealment the plaintiff gets a generous six years from the date of discovery of the violation in which to sue, but no more, even if the defendant somehow succeeds in preventing him from suing within that period.
The facts of the present case are simpler than their interpretation. In 1984 Wolin and Wortman, financial naifs who owned a modest business selling office supplies and furniture, retained Mackevieh to advise them on investing the assets of their ERISA plan, which amounted to some $650,000. They told Mackevieh that they wanted the assets placed in safe and liquid investments. Mackevieh persuaded them to place $200,000 in a pair of real estate limited partnerships. (The rest of the assets presumably he invested properly.) He assured Wolin and Wort-man that these would be safe and liquid investments. They were not. His assurances to them are the alleged wrongdoing, which the plaintiffs describe as fraud.
*851A remarkable feature of the case is that the prospectuses and other documents that Mackevich furnished the trustees stated in simple, lucid, prominent, and unmistakable language that investments in these real estate limited partnerships were risky and illiquid. Wolin and Wortman can credibly claim that they were defrauded by Mackevich only by confessing their own breach of fiduciary obligation to the plan of which they were the trustees and hence the primary fiduciaries. See 29 U.S.C. §§ 1102(a), 1103(a); Central States, Southeast & Southwest Areas Pension Fund v. Central Transport, Inc., 472 U.S. 559, 572, 105 S.Ct. 2883, 2841, 86 L.Ed.2d 447 (1985); In re Unisys Savings Plan Litigation, supra, 74 F.3d at 435-36; Plumbers & Steamfitters Local 150 Pension Fund v. Vertex Construction Co., 932 F.2d 1443, 1450 (11th Cir.1991). They failed to take the most elementary precautions to safeguard the plan’s assets, violating “ERISA’s duty to investigate,” which “requires fiduciaries to review the data a consultant gathers, to assess its significance, and to supplement it where necessary.” In re Unisys Savings Plan Litigation, supra, 74 F.3d at 435. One might expect Wolin and Wort-man to be defendants along with Mackevich and his employer in a suit for breach of fiduciary obligation brought by the beneficiaries of the plan, rather than plaintiffs. Cf. SEC v. Lauer, 52 F.3d 667 (7th Cir.1995). A further obstacle to this suit is that no suit for securities fraud can be maintained on the basis of oral representations plainly inconsistent with written ones, e.g., Associates in Adolescent Psychiatry, S.C. v. Home Life Ins. Co., 941 F.2d 561, 571 (7th Cir.1991); see also Brown v. E.F. Hutton Group, Inc., 991 F.2d 1020, 1033 (2d Cir.1993), and this principle should apply to a suit under ERISA that charges securities fraud rather than anything special to ERISA.
We need not decide just how thin the trustees’ suit is, however, because we agree with the district judge that even if it is meritorious it was filed too late. The alleged fraud occurred in 1984. The suit was not filed until 1995, eleven years later. No problem, say the plaintiffs; the defendants concealed the fraud until 1990. Every year, the plaintiffs’ ERISA plan was required to file an information return with the Internal Revenue Service listing the plan’s assets and liabilities. Every year, Wolin would call Mackevich and ask him what the limited partnerships were worth and Mackevich would tell him they were worth substantially more than their purchase price but that to be conservative Wolin should list as their value on the report to the IRS the purchase price. All this time Wolin and Wortman were receiving annual reports from each limited partnership that stated in language that could not have been much plainer that the value of the partnership shares had declined because of the decline in the real estate market. In 1989 Wolin and Wortman decided to terminate the ERISA plan and distribute its assets. The following year they sold the plan’s investments in the real estate limited partnerships for $73,000 and $38,000, respectively. They had received capital distributions before, so it is unclear what exactly they lost from their investment, compared to the safe and liquid alternative investments that they say they would have preferred; but it is plain that they lost something. When they found out the going price for the limited partnership shares, the scales fell from their eyes; they had actual knowledge at last, in 1990, of the fraud that had been perpetrated against them in 1984. They had, they claim, by virtue of the fraudulent concealment provision of the statute, six years from 1990 within which to sue. They sued within six years.
Fraudulent concealment, however, is distinct from a fraud that is concealed. E.g., Cada v. Baxter Healthcare Corp., supra, 920 F.2d at 451; Radiology Center, S.C. v. Stifel, Nicolaus & Co., supra, 919 F.2d at 1221. Otherwise the statute of limitations would not start to run in a fraud case until the fraud was exposed, even if the defendant had made no effort to conceal it beyond what is implicit in committing fraud and the plaintiff should have discovered the fraud long before its public exposure. But just how distinct from the original fraud the fraudulent concealment must be has troubled the courts. Some cases attempt a threefold distinction: between the original fraud and fraudulent concealment and within the cate*852gory of fraudulent concealment between “self-concealing” acts and acts of “active concealment.” E.g., Martin v. Consultants & Administrators, Inc., supra, 966 F.2d at 1093-95. A self-concealing act is an act committed during the course of the original fraud that has the effect of concealing the fraud from its victims. Active concealment refers to acts intended to conceal the original fraud that are distinct from that original fraud. The distinction corresponds to the more conventional, but also, perhaps, more perspicuous, distinction between equitable tolling and equitable estoppel as grounds for extending the running of a statute of limitations. Id. at 1101-03 (concurring opinion); Cada v. Baxter Healthcare Corp., supra, 920 F.2d at 450-51. Equitable tolling is invoked when the prospective plaintiff simply does not have and cannot with due diligence obtain information essential to bringing a suit. It is easy to see how this might happen in a fraud case, even if the prospective defendant made no special effort to cover up the fraud; unlike being hit over the head by a baseball bat wielded by a known assailant, fraud is secretive by nature. Equitable estoppel, as we have seen, is invoked when the prospective defendant does make such a special effort to cover up the fraud or does something else to prevent the. prospective plaintiff from suing in time, such as promising not to plead the statute of limitations as a defense.
Both grounds, equitable tolling corresponding to self-concealing acts, and equitable estoppel corresponding to active concealment, must be distinguished from another judge-made doctrine of statute of limitations law generally applicable in federal cases (unless of course modified by statute— a highly pertinent qualification, as we shall shortly see): the discovery rule. Under that rule, the statute of limitations does not even begin to run until the prospective plaintiff learns or should learn that he has been injured — not that he has been wronged, just that he has been injured. Id. at 450. After the statute of limitations starts running, it still may be arrested. This is the domain of equitable tolling and equitable estoppel. They differ critically in scope in the following respect: when the plea is equitable tolling rather than equitable estoppel, the defendant is innocent of the delay (though not of course of the original wrong), so the plaintiff must use due diligence to be allowed to toll the statute of limitations; if he does not, he has no equitable claim to avoid the time bar. In the case of equitable estoppel, which requires active misconduct by the defendant, the plaintiff is not required to be diligent. Martin v. Consultants & Administrators, Inc., supra, 966 F.2d at 1094 n. 17; id. at 1102-03 (concurring opinion). At least that is the rule in this circuit. Some of the other circuits disagree. See, e.g., J. Geils Band Employee Benefit Plan v. Smith Barney Shearson, Inc., supra, 76 F.3d at 1258; Golden Budha Corp. v. Canadian Land Co., 931 F.2d 196, 201 (2d Cir.1991). The disagreement, however, may be apparent rather than real, because the cases often run equitable tolling and equitable estoppel together. (We said that statute of limitations law is confusing, and now you’ll believe us! And we’re not through.)
An analogy can be drawn to the distinction between accidental and deliberate torts. There is no defense, partial or complete, of contributory negligence to the latter, but there is to the former. Deliberate wrongdoing is not to be blamed on the victim, just as the victim’s provocation or other fault is not a defense to criminal liability. But suppose there is no deliberate misconduct by the defendant to throw the victim off the scent, to convince him that the defendant has not made any misrepresentations or misleading omissions, or otherwise to dissuade him from suing; yet the victim, while aware that he has been hurt, so that the statute of limitations is running, lacks information that he must have in order to bring suit. Then he must sue as soon after the statute of limitations has expired as he obtains the information or would have done so had he been reasonably diligent. Cada v. Baxter Healthcare Corp., supra, 920 F.2d at 452-53. This may not be until the fraud is publicly exposed, but then again it may be sooner, depending on what the plaintiff knew or should have known. So while in a case of equitable estoppel the plaintiff has the full statutory period in which to sue after the *853estoppel has ceased to impede him, even if he did not attempt diligently to pursue his rights, in a case of equitable tolling the plaintiff must be continuously diligent and sue (if he is beyond the statutory period) as soon as it is practicable for him to do so.
We have set forth these general principles governing statutes of limitations in eases brought under federal law because of the continuing uncertainty about them that is evident in the case law and the arguments of counsel and because these principles are essential background for understanding the present case. But unfortunately for the cause of simplicity in law (perhaps a lost cause in the United States), 29 U.S.C. § 1113 deviates from the standard tripartite pattern of discovery — equitable tolling- — equitable estoppel. ERISA’s statute of limitations for suits against fiduciaries does not begin to run at the time of injury, but instead (unless the “fraud or concealment” provision is invoked, giving the plaintiff additional time) at the time that the prospective plaintiff acquires actual knowledge of the breach of fiduciary obligation, which may precede any injury. Larson v. Northrop Corp., supra, 21 F.3d at 1174; Ziegler v. Connecticut General Life Ins. Co., 916 F.2d 548, 550 (9th Cir.1990); cf. International Union v. Murata Erie North America, Inc., 980 F.2d 889, 898, 900 (3d Cir.1992). The First Circuit thus erred when it said in J. Geils Band Employee Benefit Plan v. Smith Barney Shearson, Inc., supra, 76 F.3d at 1252, that section 1113 codifies the federal discovery rule. That rule, as we noted earlier, keys the accrual of the claim to the discovery of injury rather than to the discovery of the wrongful act that caused the injury.
We have held, moreover, that “actual knowledge” in section 1113 means actual knowledge; it does not also mean constructive knowledge, or inquiry notice. Radiology Center, S.C. n Stifel, Nicolaus & Co., supra, 919 F.2d at 1222. (A person has “constructive knowledge” of something when he does not know it but should.) It is not enough that the victim of the fraud should have discovered the fraud; he must actually have discovered it. Section 1113 thus not only alters the standard federal discovery rule but also alters and supplants the defense of equitable tolling. Because the statute of limitations does not begin to run until the plaintiff knows everything that he needs to know in order to sue, there is no room for equitable tolling.
Had we held in the Radiology case that “actual knowledge” means “actual or constructive knowledge,” then the actual-knowledge clause of section 1113 would simply be a codification of equitable tolling. But the doctrine of equitable tolling requires that the plaintiff lack constructive as well as actual knowledge in order to be permitted to sue after the deadline in the statute of limitations has expired. The plaintiffs in this case, had they been diligent, would have discovered the fraud long before 1990 — -indeed, before the fraud was even committed. For if diligent they would have discovered it when they received, and before they signed, the subscription agreement for the shares in the two limited partnerships. A written statement available to the victims of fraud that reveals that a fraud has been committed furnishes constructive or inquiry notice of the fraud, and constructive notice creates a duty of diligent inquiry. Eckstein v. Balcor Film Investors, 58 F.3d 1162, 1168 (7th Cir.1995).
But that is not important here, because the doctrine of equitable tolling with its duty of reasonable inquiry is not applicable to this case. Here the statute of limitations did not start to run on these plaintiffs until 1990, when they acquired by their own admission actual knowledge of the fraud. The only question is whether the three-year or the six-year statute of limitations began to run then. That depends on whether there was fraudulent concealment of the original fraud. The original fraud that is alleged consists of the representations concerning safety and liquidity that Mackevieh made at the time of the original sale of the limited partnership shares to the trustees. The fraudulent concealment that is alleged consists of the representations that he made later concerning the value of the shares. Being distinct from the original fraud — a single, simultaneous barrage of representations concerning the safety and liquidity of the real *854estate limited partnerships, rather than a continuing fraud — these representations fall into the class of active concealment rather than into that of self-concealing acts. This is essential for the plaintiffs, since fraudulent concealment in its sense of self-concealment is subject to a duty of due diligence (Martin v. Consultants & Administrators, Inc., supra, 966 F.2d at 1095 n. 17) that the plaintiffs obviously failed to comply with. In either form, however, fraudulent concealment is a species of fraud (and so for example must be pleaded with particularity, Larson v. Northrop Corp., supra, 21 F.3d at 1173), in this case a species of securities fraud. And this dooms the plaintiffs, because the principle that oral misrepresentations will not override explicit written representations is, in our view, applicable to this as to any securities fraud. Cf. J. Geils Band Employee Benefit Plan v. Smith Barney Shearson, Inc., supra, 76 F.3d at 1258-60. Had the trustees read the annual reports of the real estate limited partnerships — short documents written in clear, everyday English — they would have known that the value of their shares was falling, not rising, because the real estate market was in decline.
So there was no fraudulent concealment. But suppose we are wrong in thinking that oral representations contradicting clear written statements cannot constitute fraudulent concealment. The question would then be whether Mackevieh, having steered the all-unknowing plaintiffs to unsuitable investments in 1984, afterward said or did anything that might have prevented the plaintiffs from tumbling to the fact that they had been conned. Did he conceal the riskiness or illi-quidity of the investments from the plaintiffs? He did not. He told Wolin that the investments were profitable, not that they were safe and liquid. (The complaint contains an allegation that he did tell Wolin they were safe and liquid, but it was dropped in the plaintiffs’ statement of facts in the district court.) Many risky and illiquid investments are profitable. Indeed, a rational risk-averse investor — and most investors are both rational and risk-averse — will not assume risk, or surrender liquidity, without being compensated in the form of a higher expected return. So telling an investor that he is making money on his investment is not telling him' anything about the risk or liquidity of the investment — except possibly, ‘Watch out!”
The alert reader will wonder, though, whether we can end the analysis here. Shouldn’t we consider loss as well as knowledge? Wolin and Wortman were unlikely to bring a lawsuit against Mackevieh and his employer as long as they thought he was making money for them. What would they be suing for? Tregenza v. Great American Communications Co., 12 F.3d 717, 720-21 (7th Cir.1993), suggests that maybe soothing noises calculated to lull the naive or greedy investor could be worked up into a ground for invoking equitable estoppel. Eckstein v. Balcor Film Investors, supra, 58 F.3d at 1169, may have seemed to close this particular door, saying that “whether a given investment makes or loses money is by and large unrelated to the question whether the prospectus was actionable.” See also Caviness v. Derand Resources Corp., 983 F.2d 1295, 1302 (4th Cir.1993). But the cases can be reconciled by noting that, if the victim of the fraud is lulled into believing that he has not been hurt, not only will he have little incentive to sue; he may not be able to sue, because he may not be able to prove injury, the normal sine qua non of liability for tort. E.g., Rozenfeld v. Medical Protective Co., 73 F.3d 154, 157 (7th Cir.1996).
The plaintiffs in our case claim, however implausibly, that they did not learn they’d lost money on their investments in the real estate limited partnerships until 1990. The normal statute of limitations does not begin to run until the prospective plaintiff has or should have discovered that he has been injured. Soothing noises that convince the fraud victim that he is making money might therefore be thought to postpone the expiration of the period in which the victim must sue, even if a more alert victim would have seen through the smokescreen. (If even an alert victim would not have discovered that he had been injured, the discovery rule, without any help from the doctrine of equitable estoppel, would postpone the running of the statute of limitations.) ERISA is special, though. Its statute of limitations for *855suits against fiduciaries begins to run when the breach of fiduciary duty is discovered, irrespective of any injury. That occurred in 1990, and the plaintiffs had three years to sue unless, before then, the defendants had fraudulently concealed the original fraud.
But wait a minute. Can it really be correct that ERISA’s statute of limitations begins to run when the breach of duty is discovered, irrespective of injury? What if the breach is discovered before there is any injury? A number of cases, illustrated by Larson v. Northrop Corp., supra, 21 F.3d at 1171, hold that the victim of an ERISA fraud by a fiduciary can sue when he discovers the breach, even if he hasn’t yet been injured. This makes sense. The breach of fiduciary duty — here alleged to be Mackevich’s steering the plaintiffs into risky, illiquid investments without disclosing the risk and illiquidity — creates a high probability of future loss, and the victim can obtain injunctive relief, to head off that future loss, before it occurs. He should be encouraged to do that, rather than sit back and play heads I win tails you lose — suing the fiduciary if the investment goes sour, pocketing the profits if the investment is a success. If the victim of a breach of fiduciary duty can and should sue when he discovers the breach rather than later, when he is injured by the breach, then soothing noises probably are not a basis for equitable estoppel in these cases because the noises do not interfere with the fraud victim’s ability to sue. So once again the plaintiffs’ argument for a longer period in which to sue fails.
But suppose that this is wrong too and that the normal discovery rule — the statute of limitations begins to run when the plaintiff discovers or should discover that he is injured — coexists with the special discovery rule of section 1113, as it can be made to do as follows: The victim of a breach of fiduciary duty has three years to sue after the discovery that he has been injured, and after the discovery of the breach, whichever comes later. If he discovers his injury in 1990 and the breach in 1992, he must sue by 1995. If he discovers the breach in 1991 and the injury in 1993, he must sue by 1996. This is the approximate pattern that would exist if suits under ERISA against fiduciaries were governed by a normal statute of limitations rather than by the “actual knowledge” provision of section 1113. The statute of limitations would begin to run when the prospective plaintiff learned that he had been injured. But if he couldn’t obtain enough (other) information to bring suit until after the statutory period had run, he would have extra time in which to sue.
Our plaintiffs, if they can be believed, discovered both the breach, and that they had been injured by it, in 1990. Mackevich’s soothing noises may have prevented them from discovering their injury until then, and if that were relevant (it isn’t) the question would be whether those soothing noises could be a ground for asserting fraudulent concealment. We suspect (no stronger word would be appropriate) not. It would be a case not of concealing fraudulent or otherwise unlawful activity, but, like promising not to plead the statute of limitations, of trying to prevent or delay the victim of the wrong from taking action. The defendant would not be hiding; he would merely be telling the plaintiff you’ve got nothing to complain about because you’re making money. If this is a ground for equitable estoppel, it seems to be one distinct from fraudulent concealment.
This brings into view the question whether a defense of equitable estoppel not limited to fraudulent concealment is either comprehended within section 1113, despite the absence of a textual handle, or survives as an independent nonstatutory defense to ERISA’s statute of limitations. We need not decide, because the plaintiffs have not argued equitable estoppel. They have argued fraudulent concealment, which is not a synonym for equitable estoppel, though it overlaps with it. There was no fraudulent concealment here — and there was none even if the term can, contrary to what we just suggested, be stretched, even in an ERISA case with its peculiar rule governing the accrual of the claim against the fiduciary, to cover efforts to make the victim of fraud think he has lost no money rather than that there has been no fraud. To recur to an earlier point, the plaintiffs received statements from the real estate limited partnerships that showed they were losing money, and there is no securities *856fraud ease against an ERISA trustee who is given unequivocal written notice of facts inconsistent with the fraud — the fraud under present discussion being Mackevich’s efforts to reassure the plaintiffs that the investments were doing all right.
We would not be inclined to relieve the plaintiffs from their waiver in failing to argue equitable estoppel even if a defense of equitable estoppel distinct from fraudulent concealment might have merit here. This appears to be a threadbare suit by persons who have not exercised due care in the management of other people’s money and see an opportunity to shift the cost of their mismanagement to other, broader shoulders. The bringers of suit based on technicalities will not be heard to complain about defenses based on technicalities. Cowen v. Bank United, 70 F.3d 937, 944 (7th Cir.1995).
Affirmed.