Registration Form S-3 under the Securities Act of 1933 is reserved for firms with a substantial following among analysts and professional investors. The Securities and Exchange Commission believes that markets correctly value the securities of well-followed firms, so that new sales may rely on information that has been digested and expressed in the security’s price. Securities Act Release No. 6383, 47 Fed.Reg. 11380 (1982). Registration on Form S-3 principally entails incorporation by reference of the firm’s other filings, such as its comprehensive annual Form 10-K and its quarterly supplements. Firms eligible to use Form S-3 also may register equity securities “for the shelf” under Rule 415(a)(l)(x), 17 C.F.R. § 230.415(a)(l)(x). Shelf registration allows the firm to hold stock for deferred sale. Information in the registration statement will be dated by the time of sale, but again the SEC believes that the market price of large firms accurately reflects current information despite the gap between registration and selling dates. Securities Act Release No. 6499, 48 Fed.Reg. 52889 (1983), accompanying the permanent adoption of Rule 415. See also Basic, Inc. v. Levinson, 485 U.S. 224, 241-45, 108 S.Ct. 978, 988-90, 99 L.Ed.2d 194 (1988), adopting the fraud-on-the-market approach to damages because capital markets efficiently establish prices that embed available information.
I
Commonwealth Edison Company, an electric utility in Illinois, is eligible to register its securities on Form S-3. In September 1983 the firm put three million shares of common stock on the shelf, using Rule 415. The succinct registration statement incorporated 176 pages of other filings, as Form S-3 permits. Commonwealth Edison sold the shares to the public on December 5, 1983, for the market price of $27,625. Stanley C. Wielgos bought 500 of these shares.
Commonwealth Edison operates several nuclear reactors, and at the time of the shelf registration it had five more under construction — LaSalle 2, Byron 1 and 2, and Braidwood 1 and 2. None could operate without a license, which the Nuclear Regulatory Commission does not issue unless satisfied that the reactor is safe. Problems at Three Mile Island and Chernobyl, coupled with increasing sophistication in reactor engineering and testing that has revealed shortcomings in old designs, have led the NRC to be more and more demanding in recent years, which postpones the operation of reactors under construction and increases their cost. Delay alone substantially increases cost, because utilities must pay for the capital they have used; re-inspections and additional work come on top of the expenses of delay. Like other firms’ reactors, Commonwealth Edison’s have been afflicted with delay and cost overruns — some attributable to the benefit of hindsight (leading to more demanding regulations), some to bureaucratic error and delay, and some to shortcomings by Commonwealth Edison and its contractors in finishing the work to meet regulatory requirements.
Of the five reactors under construction in December 1983, Byron 1 was the closest to receiving an operating license. An arm of the NRC, the Atomic Safety and Licensing Board, was considering Commonwealth Edison’s request for a license. On January 13, 1984, the ASLB did something it had never done before (and has not done since): it denied the application outright, implying that Byron 1 must be dismantled. Commonwealth Edison Co., 19 N.R.C. 36 (1984). The next market day Commonwealth Edison’s stock dropped to $21.50, a loss to equity investors of about $1 billion — which reflected not only the write-off of Byron 1 (discounted by the probability that the NRC would affirm the ASLB’s decision) but also the likely increase in the costs of completing the other four reactors. The NRC’s Atomic Safety and Licensing Appeal Board reversed the ASLB in May 1984, Commonwealth Edison Co., 19 N.R.C. 1163 (1984), and five months later the ASLB recommended that the NRC issue a license for Byron 1, Commonwealth Edison Co., 20 N.R.C. 1203 (1984), which it did. Stock prices rebounded. Delay in *511starting Byron 1, plus the expense of re-inspections during that period, cost Commonwealth Edison more than $200 million. The Illinois Commerce Commission allowed Commonwealth Edison to add the outlay to its rate base, but the Supreme Court of Illinois disagreed, People ex rel. Hartigan v. Illinois Commerce Commission, 117 Ill.2d 120, 109 Ill.Dec. 797, 510 N.E.2d 865 (1987). State officials later excluded the costs, and Commonwealth Edison is in the process of refunding about $200 million to its customers.
Between the ASLB’s decision and its reversal, Wielgos filed this suit on behalf of all who purchased in the shelf offering, naming as defendants the issuer and its underwriters. The complaint demanded $6,125 per share, the amount equity securities declined between purchase and suit. Judge McMillan certified the case as a class action; after his resignation the case was transferred to Judge Shadur, who granted summary judgment for the defendants. 688 F.Supp. 331 (N.D.Ill.1988). Wielgos’s initial complaint, and its first amended version, presented simple claims under § 11 of the ’33 Act, 15 U.S.C. § 77k. Ten months after starting the action, having had the opportunity for discovery, Wiel-gos filed a second amended complaint adding numerous claims under other portions of the securities laws. Commonwealth Edison then filed a counterclaim, contending that the additions were frivolous and that plaintiffs counsel, A. Denison Weaver, had acted without either legal or factual investigation. Weaver retreated to a third amended complaint, streamlining the case once again. In the interim, however, defendants had incurred substantial legal fees. Judge Shadur held that a counterclaim is the wrong way to seek sanctions but invited the defendants to file appropriate motions, 688 F.Supp. at 344-45. They did, and the judge found that the claims in the second amended complaint violated both 28 U.S.C. § 1927 and Fed.R.Civ.P. 11, leading to an award of approximately $300,000 to compensate the defendants for 40% of the incremental legal fees and 75% of the incremental costs they bore responding to the second amended complaint, plus interest. 123 F.R.D. 299 (1988); 127 F.R.D. 135 (1989); 1989 WL 105311, 1989 U.S. Dist. Lexis 10455. We shall hear the appeal from this award presently; for now, however, we have two appeals: one from summary judgment on the third amended complaint, and the other from the award of costs.
II
Costs need not detain us because Wielgos did not file a timely notice of appeal. The district court quantified the costs on July 8, 1988, but did not enter an order to that effect. On August 4 Wielgos filed a notice of appeal directed to the award. On September 19, 1988, the clerk entered a judgment specifying the award of costs. Wielgos promptly served a motion to reconsider the award, which the judge denied on October 4. Wielgos did not file another notice of appeal. Although the notice of August 4 would have sufficed under Fed.R.App.P. 4(a)(2) as one filed after announcement of the judgment but before its entry, a motion to reconsider nullifies all earlier notices of appeal and requires a fresh notice after the denial. Fed. R.App.P. 4(a)(4); Griggs v. Provident Consumer Discount Co., 459 U.S. 56, 103 S.Ct. 400, 74 L.Ed.2d 225 (1982).
Weaver insists that the notice of August 4 is effective because the award of costs merges with the original Rule 58 judgment on the merits. If that were so, however, then the motion for reconsideration filed in September would have suspended the entire judgment (including the decision on the merits), and Rule 4(a)(4) would require us to dismiss both notices of appeal. Although the clerk must enter a Rule 58 judgment specifying costs, the decision on the merits is independently appealable, Budinich v. Becton Dickinson & Co., 486 U.S. 196, 108 S.Ct. 1717, 1720, 100 L.Ed.2d 178 (1988); Brown Shoe Co. v. United States, 370 U.S. 294, 308-09, 82 S.Ct. 1502, 1514-15, 8 L.Ed.2d 510 (1962); Dickinson v. Petroleum Conversion Corp., 338 U.S. 507, 513-16, 70 S.Ct. 322, 325-26, 94 L.Ed. 299 (1950), and a separate notice of appeal is necessary if any party wishes to contest *512an award of costs made later. So we have jurisdiction of the appeal on the merits, but appeal No. 88-2527, challenging the quantification of costs, must be dismissed for want of jurisdiction.
Ill
In its final form the complaint contained two claims. Wielgos contended that the issuer and its underwriters violated § 11 first by underestimating the completion costs of the reactors and second by failing to reveal that the application for Byron l’s license was before the ASLB. Judge Sha-dur held the first claim precluded by Rule 175, 17 C.F.R. § 230.175. The second failed because the chance of a flat turn-down was so slim that the proceeding was not “material”.
A
Documents incorporated by reference into the ■ registration statement estimated the total costs of building Byron and Braid-wood and the years these reactors would begin making power. Commonwealth Edison gave 1984 as the service date for Byron 1, 1985 for Byron 2 and Braidwood 1, and 1986 for Braidwood 2. It estimated the total costs of the two Byron reactors at $3.34 billion and of the two Braidwood reactors at $3.1 billion. Each projection came with a caution such as this:
[T]he Company has under construction the additional generating units set forth below, the completion and operation of which will be subject to various regulatory approvals. These approvals may be subject to delay because of the opposition of parties who have intervened or may intervene in proceedings with respect thereto, changes in regulatory requirements or changes in design and construction of these units.
Anyone who followed Commonwealth Edison’s filings would have seen that each year the firm increased the estimated costs and delayed the estimated startup date of one or more reactors. No one had to read the fine print of a Form 10-K to recognize that higher costs and deferred completion are facts of life in the nuclear power industry; newspapers report this regularly, and the analysts who specialize in utility stocks know the story in detail.
Estimates incorporated into the September 1983 prospectus were calculated in December 1982. Commonwealth Edison updates its projections on an annual cycle; by September the December 1982 figures were stale, and Commonwealth Edison said so. Its latest quarterly report, also incorporated by reference into the registration statement, said that .
[t]he Company is in the process of conducting its annual review of the status of its construction program. While that review has not been completed, it appears likely that at the conclusion of the review the Company will announce a delay of approximately three months in the service date and a resultant cost increase for its Byron Unit 1. Any change in the service dates of the remaining generating units under construction will not be ascertained until the completion of the review.
Review began early in September 1983; in late December 1983 (after the securities had been sold) the Manager of Projects submitted his revised estimates to the firm’s Expenditure Control Committee. On January 10 that Committee approved a projection increasing the costs of Byron 1 and 2 by $330 million. When the ASLB denied the application for a license at Byron 1, the firm immediately added another $100 million to the estimate, which it disclosed in a Form 8-K filed on January 17, 1984.
The statements incorporated into the prospectus were erroneous — not only in the sense that they turned out to be inaccurate but also in the sense that by early December 1983, when it sold the stock, Commonwealth Edison’s internal cost estimates exceeded those in the documents on file. A material error in a prospectus usually is enough for liability. The district court granted summary judgment for the defendants, however, on the basis of Rule 175, one of many “safe harbors” established by the SEC on the authority of *513§ 19(a) of the ’33 Act, 15 U.S.C. § 77s(a). Rule 175 provides in part:
(a) A statement within the coverage of paragraph (b) ... which is made by or on behalf of an issuer ... shall be deemed not to be a fraudulent statement (as defined in paragraph (d) of this section), unless it is shown that such statement was made or reaffirmed without a reasonable basis or was disclosed other than in good faith.
(b) This rule applies to the following statements:
(1) A forward-looking statement (as defined in paragraph (c) of this section) made in a document filed with the Commission ...
(c) For the purpose of this rule the term “forward-looking statement” shall mean and shall be limited to:
(1) A statement containing a projection of revenues, income (loss), earnings (loss) per share, capital expenditures, dividends, capital structure or other financial items; ...
All agree that the statements in question estimate “capital expenditures” within the meaning of Rule 175(c)(1). Because prospectuses and the 10-K and 8-K reports are documents “filed with the Commission” under Rule 175(b), they qualify for the safe harbor of Rule 175(a) “unless it is shown that such statement was made or reaffirmed without a reasonable basis or was disclosed other than in good faith.” Rule 175(a) implies that once the issuer shows it has made a “forward-looking statement”, the burden of persuasion concerning “reasonable basis” and “good faith” rests with the plaintiff. As the district judge observed, 688 F.Supp. at 337-38, Wielgos has not tried to establish that Commonwealth Edison and the underwriters acted “other than in good faith”. That leaves the question whether the statements had a “reasonable basis”; the district judge thought they did.
Wielgos tries to escape this by focusing on the first part of Rule 175(a), which says that covered statements “shall be deemed not to be ... fraudulent”. Liability under § 11 does not depend on “fraud”, a term implying a mental element, Ernst & Ernst v. Hochfelder, 425 U.S. 185, 96 S.Ct. 1375, 47 L.Ed.2d 668 (1976); any “untrue statement of a material fact” leads to liability for the issuer, § 11(a), and for every other signer who cannot make out a due diligence or expertise defense, § 11(b). See Herman & MacLean v. Huddleston, 459 U.S. 375, 383, 103 S.Ct. 683, 687, 74 L.Ed.2d 548 (1983). This argument slights the language of Rule 175(a). A qualifying forward-looking statement is “deemed not to be a fraudulent statement {as defined in paragraph (d) of this section)” (emphasis added), and subparagraph (d) states:
For the purpose of this rule the term “fraudulent statement” shall mean a statement which is an untrue statement of a material fact, a statement false or misleading with respect to any material fact, an omission to state a material fact necessary to make a statement not misleading, or which constitutes the employment of a manipulative, deceptive, or fraudulent device, contrivance, scheme, transaction, act, practice, course of business, or an artifice to defraud, as those terms are used in the Securities Act of 1933 or the rules or regulations promulgated thereunder.
“[Fjraudulent statement” in Rule 175(a) turns out to be shorthand for all of the bases of liability in the ’33 Act and its implementing rules. Rule 175 is a safe harbor after all; qualifying statements may not be the basis of liability.
Forward-looking statements need not be correct; it is enough that they have a reasonable basis. In December 1982, when Commonwealth Edison made the estimates that Wielgos challenges, it used the best available information. Wielgos does not say otherwise. What he offers are two variations on the theme that delay disqualifies an estimate: he notes that in a world of increasing costs any estimate that was almost a year old is bound to be wrong, and he adds for good measure that Commonwealth Edison knew it. Teams of employees were revising the estimate, and by the date of sale one team had made a higher projection, although, this had not been reviewed by upper levels of engineers and *514managers. None of this denies the issuer the shelter of Rule 175.
Commonwealth Edison made point estimates: Byron 1 and 2 will cost $3.34 billion and start in 1984 and 1985. Everyone understands that point estimates are almost certainly wrong. Things will not go exactly as predicted, and any deviation will cause the future to diverge from the estimate. Statisticians — and stock analysts— need confidence intervals to go with the maximum-likelihood estimate. Commonwealth Edison might have said, for example, that there is a two-thirds chance that the cost will fall in a given range, identifying the events that would push the cost up or down within (or outside of) that range. As new information comes to light, the firm and its observers may evaluate the consequences for themselves. Commonwealth Edison did not do this, and the projection it made did not assist anyone in estimating how likely (and how great) a departure from its point estimate would be.
Inevitable inaccuracy of a projection does not eliminate the safe harbor, however. Rule 175 does not say that projections qualify only if firms give ranges and identify the variables that will lead to departure. Like Form S-3 and the shelf registration rules, Rule 175 assumes that readers are sophisticated, can understand the limits of a projection — and that if any given reader does not appreciate its limits, the reactions of the many professional investors and analysts will lead to prices that reflect the limits of the information. Securities Act Release No. 5992, 43 Fed.Reg. 53246 (1978). A belief that investors — collectively if not individually — can look out for themselves and ought to have information that may improve the accuracy of prices even if it turns out to be fallacious in a given instance underlies the very existence of Rule 175.
Until 1978, when it adopted Rule 175, the SEC discouraged firms from making projections or commenting beyond the domain of “hard” information, such as last year’s sales. See Walker v. Action Industries, Inc., 802 F.2d 703, 707 (4th Cir.1986) (describing this history); Victor Brudney, A Note on Materiality and Soft Information Under the Federal Securities Laws, 75 Va.L.Rev. 723, 753-57 (1989). It did this because statements about the future are less reliable than statements about the past. If you view investors as easily misled and unable to appreciate the uncertainty of predictions, you try to keep such information out of their hands. You will not succeed. Investors value securities because of beliefs about how firms will do tomorrow, not because of how they did yesterday. If enterprises cannot make predictions about themselves, then securities analysts, newspaper columnists, and charlatans have protected turf. There will be predictions aplenty outside the domain of the securities acts, predictions by persons whose access to information is not as good as the issuer’s. When the issuer adds its information and analysis to that assembled by outsiders, the collective assessment will be more accurate even though a given projection will be off the mark.
Convincing the SEC of the utility of projections is one thing, and convincing enterprises that they ought to make projections is another. What’s in it for them? If all estimates are made carefully and honestly, half will turn out too favorable to the firm and the other half too pessimistic. In either case the difference may disappoint investors, who can say later that they bought for too much (if the projection was too optimistic) or sold for too little (if the projection turns out to be too pessimistic). Thus the role of a safe harbor: the firm is not liable despite error.
Safe harbors are not necessarily enough. Harbors could be impossible to enter. Suppose the Commission were to require the issuer to reveal all of the data, assumptions, and methodology behind its projections, so that participants in the market could assess them fully and react appropriately. Data could be proprietary, secrets whose revelation to business rivals could damage the firm and its investors. Assumptions about the firm’s own behavior could lead other firms to change theirs— for example, revealing the assumption that product line X will be discontinued at the end of the year could lead customers to *515stop buying it now or offer a rival a clue about how to capitalize. See Roger J. Dennis, Mandatory Disclosure Theory and Management Projections: A Law and Economics Perspective, 46 Md.L.Rev. 1197, 1211-18 (1987). The SEC therefore does not require the firm to reveal its data, assumptions, and methods. Rule 175(c)(4) allows the firm to reveal them, but their disclosure is not a condition of the safe harbor. Shelter without mandatory disclosure of the data and assumptions that may turn out to be important to evaluate the soundness depends on a belief that investors, collectively, are sophisticated. Readers may infer from what is said and what is omitted how reliable the estimate is. Firms that want to induce greater reliance may reveal more. Rule 175 leaves it to the issuer and the market to determine how much is revealed.
Commonwealth Edison made a projection but did not disclose data, assumptions, or methods. The cautions accompanying the projection were so much boilerplate. No one could have deduced from the cloud of legalese how much uncertainty the firm perceived in its estimates, and the market would have discounted them accordingly. Especially because professional analysts knew — although perhaps Wielgos did not— that Commonwealth Edison’s estimates were biased. Not wrong in the sense of inaccurate; biased in the sense of having a predictable kind of inaccuracy. Like clockwork, every January the firm increased its estimate of the cost of getting the Byron and Braidwood reactors into service. Some years the estimates went up more than others, but up they went. Lack of a normal distribution of errors could suggest the absence of a “reasonable basis” or even of “good faith”. More likely, though, the pattern of errors allowed the market to infer two assumptions: that there would be no new regulations or unanticipated delays. Commonwealth Edison was estimating the costs it would experience if nothing went wrong and nothing unexpected happened. These were poor assumptions. Something always goes wrong, and in the nuclear power business the unexpected is the norm. Perhaps the firm had no way to estimate the timing and gravity of coming jolts, but if so that made the estimate less useful.
Because Rule 175 does not require a firm to reveal assumptions, Commonwealth Edison did not need to tell the market it was making these. Anyway, professional investors and analysts surely deduced what was afoot. Once they did so, they supplied their own assumptions about the likelihood that the firm will encounter trouble or that the rules will change. Issuers need not “disclose” Murphy’s Law or the Peter Principle, even though these have substantial effects on business. So too issuers need not estimate the chance that a federal agency will change its rules or tighten up on enforcement. Securities laws require issuers to disclose firm-specific information; investors and analysts combine that information with knowledge about the competition, regulatory conditions, and the economy as a whole to produce a value for stock. See Acme Propane, Inc. v. Tenexco, Inc., 844 F.2d 1317, 1323-24 (7th Cir.1988). Just as a firm needn’t disclose that 50% of all new products vanish from the market within a short time, so Commonwealth Edison needn’t disclose the hazards of its business, hazards apparent to all serious observers and most casual ones.
If Commonwealth Edison were doing significantly and unexpectedly worse than the industry as a whole — in completing its reactors or in making estimates about their costs — that might signal the presence of important, firm-specific information that it would have to reveal. Wielgos does not contend that either the estimates or the performance of Commonwealth Edison fall substantially below the norms of the industry. (Half of all firms do worse than average, and finishing in the lower half does not imply a securities problem, but Wielgos does not even suggest that Commonwealth Edison is in the lower half.)
Because the estimates in question have a reasonable basis once they are understood as projecting forward from past experience rather than trying to predict what new things can go wrong, they are covered by Rule 175 unless, by the time Commonwealth Edison used the estimates by selling *516the stock on the basis of documents incorporating them, they “no longer [had] a reasonable basis.” Release No. 5992, 43 Fed.Reg. at 53250. Wielgos observes that by December 1983 a team of employees had developed a different estimate; Commonwealth Edison responds that although the old estimate was stale, the new one was tentative and subject to review by higher echelons before release. Panter v. Marshall Field & Co., 646 F.2d 271, 291-93 (7th Cir.1981), holds that firms need not disclose tentative internal estimates, even though they conflict with published estimates, unless the internal estimates are so certain that they reveal the published figures as materially misleading. See also Kademian v. Ladish Co., 792 F.2d 614, 625 (7th Cir.1986). Estimates in progress in December 1983 were not of that character. Recall that the published projections included not only the one revealed in January 1983 but also the information in the quarterly report of September 1983 that the projections would be revised upward by an amount then under consideration.
Issuers need not reveal all projections. Any firm generates a range of estimates internally or through consultants. It may reveal the projection it thinks best while withholding others, so long as the one revealed has a “reasonable basis" — a question on which other estimates may reflect without automatically depriving the published one of foundation. Because firms may withhold even completed estimates, they may withhold in-house estimates that are in the process of consideration and revision. Any other position would mean that once the annual cycle of estimation begins, a firm must cease selling stock until it has resolved internal disputes and is ready with a new projection. Yet because large firms are eternally in the process of generating and revising estimates — they may have large staffs devoted to nothing else — a demand for revelation or delay would be equivalent to a bar on the use of projections if the firm wants to raise new capital. Rule 175 is designed to release enterprises from such binds.
It was no secret that the estimate prepared in December 1982 was too low. The firm said so in September 1983. Proceedings in the ASLB, including the staffs demand for re-inspections of Byron l’s plumbing — costly to perform, costly because of delay — were public knowledge. The market price of the firm’s stock, which Wielgos and his class paid in the shelf offering, reflected this information. Prompt incorporation of news into stock price is the foundation for the fraud-on-the-market doctrine and therefore supports a truth-on-the-market doctrine as well. In re Apple Computer Securities Litigation, 886 F.2d 1109 (9th Cir.1989); Flamm v. Eberstadt, 814 F.2d 1169, 1179-80 (7th Cir.1987); Rodman v. Grant Foundation, 608 F.2d 64, 70 (2d Cir.1979). Knowledge abroad in the market moderated, likely eliminated, the potential of a dated projection to mislead. It therefore cannot be the basis of liability.
B
Wielgos’s remaining claim is that Commonwealth Edison and its underwriters violated § 11 by omitting from all of their filings the fact that the application for a license to operate Byron 1 was pending before the Atomic Safety Licensing Board. In the district court Wielgos argued that Item 103 of Regulation S-K, 17 C.F.R. § 229.103 — which calls for the issuer to disclose “any material pending legal proceedings” although not “ordinary routine litigation incidental to the business” — covered the license application. See generally Louis Loss & Joel Seligman, 2 Securities Regulation 648-62 (3d ed. 1989) (describing the requirements of Item 103). In this court Wielgos relies particularly on Instruction 5B to Item 103, which says that “an administrative or judicial proceeding” under any environmental laws or rules that “involves ... capital expenditures ... and the amount involved ... exceeds 10 percent of the current assets of the registrant” cannot be treated as a “routine” proceeding (which needn’t be disclosed). Judge Shadur did not decide whether the application is “routine”; Instruction 5B qualifies only that exception. *517The judge decided instead that the status of the application was not “material”.
Never in its history had the ASLB, let alone the NRC as an institution, denied an application for a license. Often the ASLB tells applicants to do more work; when it sold the stock, Commonwealth Edison was gearing up to reinspect the welds at Byron 1 in light of deficiencies that had been discovered. Obviously the ASLB would not recommend that the NRC issue a license until that program had been completed to its satisfaction. Deferral and denial are worlds apart. As the court saw things, the probability of an outright denial in January 1984 was remote, making the proceeding itself not material. It would be foolish, the judge thought, to require issuers to predict that administrative officials will make costly errors and be reversed. How do you predict blunders?
Information is material when “there is a substantial likelihood that a reasonable shareholder would consider it important”. TSC Industries, Inc. v. Northway, 426 U.S. 438, 449, 96 S.Ct. 2126, 2132, 48 L.Ed.2d 757 (1976). At first glance it is incongruous to hold that an investor would not “consider important” a proceeding that cut 20% off the price of the issuer’s stock. If Commonwealth Edison knew what the ASLB was going to do, the proceeding would have been important indeed. It didn’t. Materiality depends not only on the magnitude of an effect but also on its probability. The anticipated magnitude (the size if the worst happens, multiplied by the probability that it will happen) may be small even when the total effect could be whopping. Reasonable investors do not want to know everything that could go wrong, without regard to probabilities; that would clutter registration documents and obscure important information. Issuers must winnow things to produce manageable, informative filings.
Materiality is hard to pin down in the abstract. Basic, Inc. v. Levinson emphasizes that materiality is a “fact-specific” inquiry, 485 U.S. at 239-40, 108 S.Ct. at 987-88, and that close cases should not be resolved by summary judgment. Even very improbable events may be material if the injury is great enough. Just as tort cases often must go to trial so that juries may decide whether the cost of taking precautions was less than the gravity of the loss discounted by its probability (the definition of negligence), securities cases presenting similar questions may require trials.
Our case may be decided, however, without regard to materiality. We think that Commonwealth Edison revealed all that Item 103 requires. Its documents said that it was building five nuclear reactors, which it could not operate without licenses from the NRC. It told investors that it did not have licenses and that environmental groups were opposing its applications. What it did not say is that the application for Byron 1 was before the ASLB rather than some other part of the NRC, and that if the ASLB denied its application costs would go up while it tried to obtain a reversal. This is rather like revealing pending litigation without saying that the case is pending before a magistrate, and that costs will go up if the magistrate should make an adverse (and erroneous) but influential recommendation.
Issuers of securities must reveal firm-specific information. Investors combine this with public information to derive estimates about the securities’ value. It is pointless and costly to compel firms to reprint information already in the public domain. Issuers needn’t print the Code of Federal Regulations, which parcels authority among employees of the NRC. Acme Propane, Inc. v. Tenexco, Inc., 844 F.2d at 1323-24. Item 103 does not call on registrants to describe the internal organization of courts or administrative bodies or even to state the status of the pending case. It says instead that the registration statement must “[djescribe briefly” (emphasis added) and continues:
Include the name of the court or agency in which the proceedings are pending, the date instituted, the principal parties thereto, a description of the factual basis alleged to underlie the proceeding and the relief sought.
*518Nothing there about the status of the litigation within the tribunal, or how the tribunal is organized, or the probability that the tribunal will deliver a particular decision. Commonwealth Edison reported that it needed licenses from the NRC, and the “factual basis” for obtaining (or denying) the licenses flowed like a river through its papers. It would have been otiose to do more.
Wielgos treats the registration statement and the documents it incorporates as guarantees. They did not reveal that higher costs lay ahead; they did not predict a legally erroneous decision of the ASLB; therefore, Wielgos says, Commonwealth Edison must pay. The securities acts do not have this ex post perspective. Their approach is ex ante. Issuers and underwriters must decide what information will be useful without burying investors under a blizzard of paper. See Greenapple v. Detroit Edison Co., 618 F.2d 198, 210 (2d Cir.1980). No investor absorbs sheafs of dense type, which Commonwealth Edison printed or incorporated in connection with this shelf offering. Descriptions in Forms 10-K and registration statements are almost useless to individual investors. They require absorption by professional traders and investors. What these professionals need is new information specific to the issuer. Telling them over and again how the NRC works, or that costs are rising in the nuclear power industry, or even that Commonwealth Edison had run into trouble with its welds (which became known a few months before this stock was sold), has nothing to do with the accuracy of prices in the market. Investors who buy 500 shares of stock rely on the market price; Wielgos concedes that he did not read Commonwealth Edison's disclosures. Everything we can see demonstrates that the market had in its possession all significant information about Commonwealth Edison. That firm lived up to the technical requirements of Item 103. No one’s interests would be served by requiring the details Wielgos demands from the privileged position of hindsight, as opposed to the “brief[]” description the SEC solicited.
Appeal No. 88-2527 is dismissed for want of jurisdiction. On appeal No. 88-1900, the judgment is
Affirmed.