EASTERBROOK, Circuit Judge.

Indiana Gas Co. and Citizens Gas Coke, two utilities that supply natural gas to customers in Indiana, formed a joint venture (called ProLiance Energy) to manage the contracts by which they purchase gas and transportation services from the interstate pipelines that pass through that state. United States Gypsum (USG) purchases substantial quantities of gas for use in manufacturing; it buys gas at the wellhead and deals directly with the pipelines for transportation. In this litigation under sections 1 and 2 of the Sherman Act, 15 U.S.C. §§ 1, 2, USG contends that ProLiance is an unlawful combination that by contract controls a substantial fraction of the transport capacity between the gas fields and Indiana, and that it has used this market power to monopolize. Even though USG buys transportation directly from the pipelines, it alleges, the price the pipelines can charge for their services depends on what ProLiance has done with its portion of the capacity. According to USG, pipelines have been able to charge more for their residual capacity because of ProLiance's existence (and practices) than the pipelines would have been able to charge in its absence.

Indiana Gas and Citizens Gas have many customers with firm entitlements to gas. In order to assure delivery, Indiana Gas and Citizens Gas purchase more pipeline capacity than needed for daily deliveries; they hold the excess as reserve for the benefit of the uninterruptible customers during periods of peak demand, such as cold snaps or a business's high season. During times of average demand, Indiana Gas and Citizens Gas sold their excess transport entitlement on the spot market, where USG bought it at attractive prices and used it to secure gas that it stored for times when spot market prices were high. After ProLiance came into existence, however, it ended (or at least greatly curtailed) these spot-market sales, forcing USG to pay more for firm capacity from the pipelines (firm commitments always sell for more than interruptible or spot purchases).

There are several ways to characterize what happened. ProLiance contends that, by managing purchases on behalf of both Indiana Gas and Citizens Gas, it has achieved efficiencies: when one utility's demand peaks, the other's may be closer to normal, which means that less aggregate reserve capacity is needed. This is the way in which an insurer, by pooling many imperfectly correlated risks, creates a portfolio that is less risky than any insured standing alone. Thus ProLiance needs less standby capacity for peak periods and can provide more firm, uninterruptible commitments per unit of pipeline capacity than either Indiana Gas or Citizens Gas could do on its own. An increase in demand from the utilities' customer base then can be met without an increase in price. The upshot, however, is that third parties such as USG find fewer bargains in the spot market. As USG sees matters, however, the higher spot-market prices stem not from risk pooling but from ProLiance either holding reserve capacity off the market (a reduction in output that drives up prices) or bundling the release of reserve transport capacity with gas (which USG describes as a monopolistic tie-in sale).

Because all we have to go on is USG's complaint, it is too soon to determine whose understanding of these events is superior. The district judge concluded that it would never be necessary to examine these issues and dismissed the complaint, citing Fed.R.Civ.P. 12(b)(6), on three grounds: first, USG has not suffered antitrust injury because it does not buy from ProLiance; second, the suit is barred by the four-year period of limitations in 15 U.S.C. § 15b; third, USG could not prove its claims in light of adverse findings by the Indiana Utility Regulatory Commission in a proceeding to which USG was a party. None of these is a good ground on which to dismiss USG's complaint — and the latter two are not permissible even in principle, because the statute of limitations and issue preclusion are affirmative defenses. See Fed.R.Civ.P. 8(c). Complaints need not anticipate or attempt to defuse potential defenses. See Gomez v. Toledo, 446 U.S. 635, 100 S.Ct. 1920, 64 L.Ed.2d 572 (1980). A complaint states a claim on which relief may be granted when it narrates an intelligible grievance that, if proved, shows a legal entitlement to relief. See Swierkiewicz v. Sorema N.A., 534 U.S. 506, 122 S.Ct. 992, 152 L.Ed.2d 1 (2002); Bennett v. Schmidt, 153 F.3d 516 (7th Cir. 1998). A litigant may plead itself out of court by alleging (and thus admitting) the ingredients of a defense, see Walker v. Thompson, 288 F.3d 1005 (7th Cir. 2002) (applying this principle to the period of limitations), but this complaint does not do so; the district judge thought, rather, that the complaint had failed to overcome the defenses. As complaints need not do this, the omissions do not justify dismissal. What is more, all three grounds are unsound in application as well as in principle.

A private plaintiff must show antitrust injury — which is to say, injury by reason of those things that make the practice unlawful, such as reduced output and higher prices. The antitrust-injury doctrine was created to filter out complaints by competitors and others who may be hurt by productive efficiencies, higher output, and lower prices, all of which the antitrust laws are designed to encourage. See, e.g., Atlantic Richfield Co. v. USA Petroleum Co., 495 U.S. 328, 110 S.Ct. 1884, 109 L.Ed.2d 333 (1990); Cargill, Inc. v. Monfort of Colorado, Inc., 479 U.S. 104, 107 S.Ct. 484, 93 L.Ed.2d 427 (1986); Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 97 S.Ct. 690, 50 L.Ed.2d 701 (1977). A plaintiff who wants something, such as less competition or higher prices, that would injure consumers, does not suffer antitrust injury. In Midwest Gas Services, Inc. v. Indiana Gas Co., 317 F.3d 703 (7th Cir. 2003), we held that the antitrust-injury doctrine prevents a suit by one of ProLiance's business rivals. USG, by contrast, is a consumer of gas; it is in the class of persons protected from reductions in output and higher prices. And USG contends that it has been required to pay higher prices. Its injury (if any) is antitrust injury. That at least one of ProLiance's rivals has sued, and that none of its indirect purchasers (the customers of Indiana Gas and Citizens Gas) has done so, may be informative, but it does not prevent USG from attempting to show that ProLiance has anticompetitive consequences.

Portions of the district court's opinion equate the antitrust-injury doctrine of Brunswick and its successors with the direct-purchaser doctrine of Illinois Brick Co. v. Illinois, 431 U.S. 720, 97 S.Ct. 2061, 52 L.Ed.2d 707 (1977), and Hanover Shoe, Inc. v. United Shoe Machinery Corp., 392 U.S. 481, 88 S.Ct. 2224, 20 L.Ed.2d 1231 (1968). USG may suffer from higher prices but does not buy from defendants, which the district judge thought dispositive. If USG were seeking damages, and ProLiance's direct or derivative customers also wanted (or could seek) monetary relief, then defendants would have a point. See Kansas v. UtiliCorp United Inc., 497 U.S. 199, 110 S.Ct. 2807, 111 L.Ed.2d 169 (1990) (reserving the possibility of suit by an indirect customer if the direct customer is a participant in the cartel); cf. Paper Systems Inc. v. Nippon Paper Industries Co., 281 F.3d 629 (7th Cir. 2002). But the direct-purchaser doctrine does not foreclose equitable relief, nor does it apply when no purchaser could obtain damages, for then there is no risk of double recovery (and no need to calculate elasticities in order to apportion damages among multiple tiers).

A cartel cuts output, which elevates price throughout the market; customers of fringe firms (sellers that have not joined the cartel) pay this higher price, and thus suffer antitrust injury, just like customers of the cartel's members. We noted and reserved in Loeb Industries, Inc. v. Sumitomo Corp., 306 F.3d 469 (7th Cir. 2002), a number of potentially difficult issues about the design of relief when the customer of a fringe firm sues the (supposed) cartel members and the injury is derivative. See also Associated General Contractors of California, Inc. v. California State Council of Carpenters, 459 U.S. 519, 103 S.Ct. 897, 74 L.Ed.2d 723 (1983); Blue Shield of Virginia v. McCready, 457 U.S. 465, 102 S.Ct. 2540, 73 L.Ed.2d 149 (1982). Courts sometimes label this "antitrust standing," despite the potential for confusion with Article III standing (which requires only injury in fact plus redressability.) We did not resolve these issues in Loeb and need not do so here either. It is enough to reiterate, as Loeb holds, that the buyers from fringe firms suffer antitrust injury, that their complaints cannot be dismissed at the outset under the Illinois Brick doctrine, and that the potential to establish injury through elevation of price in the affected market satisfies any distinct "antitrust standing" requirement. See also Metallgesellschaft AG v. Sumitomo Corp. of America, 325 F.3d 836 (7th Cir. 2003).

Now we turn to the statute of limitations. ProLiance was formed in March 1996, and USG did not file this suit until October 2000. The statute of limitations is four years — but, as the district judge recognized, this time runs from the most recent injury caused by the defendants' activities rather than from the cartel's inception. See, e.g., Zenith Radio Corp. v. Hazeltine Research, Inc., 401 U.S. 321, 91 S.Ct. 795, 28 L.Ed.2d 77 (1971); United States v. E.I. du Pont de Nemours Co., 353 U.S. 586, 77 S.Ct. 872, 1 L.Ed.2d 1057 (1957). Cf. Klehr v. A.O. Smith Corp., 521 U.S. 179, 188-91, 117 S.Ct. 1984, 138 L.Ed.2d 373 (1997) (describing how this approach works). The district court wrote that the complaint was deficient because USG failed to "show some injurious overt act within the limitations period" — but, as we have observed already, complaints need not allege facts that tend to defeat affirmative defenses. The right question is whether it is possible to imagine proof of the critical facts consistent with the allegations actually in the complaint. See Hishon v. King Spalding, 467 U.S. 69, 104 S.Ct. 2229, 81 L.Ed.2d 59 (1984); Conley v. Gibson, 355 U.S. 41, 78 S.Ct. 99, 2 L.Ed.2d 80 (1957). Proof that ProLiance had committed an anticompetitive act after October 1996 would not contradict any of the complaint's allegations. Obviously USG hopes to show that ProLiance regularly keeps some capacity off the market, ties gas and transport together, or performs other acts that could be thought to violate the antitrust laws. Otherwise what's the point of USG's suit?

To the extent that defendants believe that even new anticompetitive acts and fresh injury within the four years before suit are insufficient, if the joint activity began earlier, that position cannot be reconciled with du Pont, which held that old activity (in du Pont, a stock acquisition preceding the suit by 30 years) is not immunized, if the potential for a reduction in output is created or realized more recently as market conditions change. Cooperative ventures may begin innocently but acquire market power (or begin to exercise it) afterward; if this occurs, a suit within four years of any anticompetitive activity is timely. This is clear enough if we apply the label "cartel" to what Indiana Gas and Citizens Gas call a "joint venture." Choice of terminology does not shorten the time for suit. A merger may be complete at closing, see Concord Boat Corp. v. Brunswick Corp., 207 F.3d 1039, 1050-53 (8th Cir. 2000), but a joint venture or cartel is a continuing cooperative activity that may be discontinued, or amended, from time to time. (According to the state agency, ProLiance's basic agreements had to be renegotiated in 2000. Opinion at 57.) The parties' decision to keep a joint venture in operation or manage the operations in ways that may violate antitrust rules is one that may be challenged when adverse effects are felt.

As for issue preclusion (collateral estoppel): USG's principal argument is that the state commission did not have "jurisdiction" to resolve a federal antitrust claim, so as a matter of federal law its findings must be disregarded. That's wrong, for two reasons. First, the preclusive effect of a state judicial decision depends on state rather than federal law. See 28 U.S.C. § 1738. (A state agency acting in a judicial capacity is a court for this purpose. See University of Tennessee v. Elliott, 478 U.S. 788, 799, 106 S.Ct. 3220, 92 L.Ed.2d 635 (1986). USG does not contest the district judge's conclusion that the Indiana Utility Regulatory Commission, whose decision was affirmed by the state's highest court, United States Gypsum, Inc. v. Indiana Gas Co., 735 N.E.2d 790 (Ind. 2000), was acting in such a capacity.) State law controls with respect to preclusion even when a federal court has exclusive jurisdiction of a federal claim that may be affected by the state's decision. Marrese v. American Academy of Orthopaedic Surgeons, 470 U.S. 373, 105 S.Ct. 1327, 84 L.Ed.2d 274 (1985). Second, USG confuses issue preclusion with claim preclusion. An agency or court that lacks authority to decide whether ProLiance has violated the antitrust laws nonetheless may resolve a disputed issue — such as whether ProLiance has market power — that has significance beyond the particular adjudication in which the issue is addressed. Indiana gives preclusive effect to issues actually and necessarily decided in a contested adjudication. See McClanahan v. Remington Freight Lines, Inc., 517 N.E.2d 390, 394 (Ind. 1988). All doubts about the proper use of Rule 12(b)(6) to one side, the right question to ask is what, in particular, the state agency actually decided.

When Indiana Gas and Citizens Gas formed ProLiance, USG and several other customers asked the Indiana Utility Regulatory Commission to block the plan. They offered two lines of argument: first, that ProLiance would itself be a utility that could not come into existence without the Commission's permission; second, that Indiana Gas and Citizens Gas (which are utilities subject to the Commission's jurisdiction) did not satisfy the "public interest" standard when forming ProLiance. The Commission rejected the first on grounds that do not matter to this antitrust litigation. It rejected the second after finding that ProLiance serves the public interest by enabling Indiana Gas and Citizens Gas to make better use of their joint reserve capacity. Petition by Ratepayers of Indiana Gas Co., No. 40437, 1996 WL 754134 (Sept. 12, 1997), affirmed under the name United States Gypsum, Inc. v. Indiana Gas Co., 735 N.E.2d 790 (Ind. 2000).

One month after the state Supreme Court's decision, USG filed this antitrust action, only to be met by the argument that the Commission's decision knocks out essential elements of the federal claim. The district court wrote that USG loses because "the issue sought to be precluded — the improper creation and operation of ProLiance — is the same as that involved in [the] prior action that was before the" Commission. But "the improper creation and operation of ProLiance" is not an "issue"; that is far too lofty a level of generality. Putting the matter this way suggests that the district court has equated issue preclusion with claim preclusion. Indiana did not require USG to present its federal antitrust claims to the Commission, so the rules of merger and bar do not block this litigation. Unless the agency decided some concrete issue that also bears on the antitrust claim, USG does not encounter a problem with preclusion.

A finding that "X is in the public interest" is compatible with subsequent antitrust litigation. See California v. FPC, 369 U.S. 482, 489, 82 S.Ct. 901, 8 L.Ed.2d 54 (1962); United States v. Radio Corp. of America, 358 U.S. 334, 351-52, 79 S.Ct. 457, 3 L.Ed.2d 354 (1959). It might mean simply that Indiana has decided that cartels serve the public interest, a conclusion that under the Supremacy Clause must yield to contrary federal policy. (Antitrust law makes an exception for state policies that compel monopolistic organization of regulated industries, see Cantor v. Detroit Edison Co., 428 U.S. 579, 96 S.Ct. 3110, 49 L.Ed.2d 1141 (1976); Parker v. Brown, 317 U.S. 341, 63 S.Ct. 307, 87 L.Ed. 315 (1943), but no one argues that Indiana required the utilities to form ProLiance.)

Defendants do not rely on the district court's understanding. Instead they contend that the agency made a favorable, concrete finding: that ProLiance lacks market power. If that is so, then USG's antitrust claim fails at the threshold. See, e.g., Jefferson Parish Hospital District No. 2 v. Hyde, 466 U.S. 2, 104 S.Ct. 1551, 80 L.Ed.2d 2 (1984); Elliott v. United Center, 126 F.3d 1003 (7th Cir. 1997); Digital Equipment Corp. v. Uniq Digital Technologies, Inc., 73 F.3d 756 (7th Cir. 1996); Chicago Professional Sports Limited Partnership v. National Basketball Ass'n, 95 F.3d 593 (7th Cir. 1996); Polk Bros., Inc. v. Forest City Enterprises, Inc., 776 F.2d 185 (7th Cir. 1985). We have searched the agency's decision in vain for such a finding, however. Although the agency mentioned market power as a factor worth consideration, it did not find that ProLiance has none. What it did say is that (a) the pipelines' transportation capacity to Indiana is unaffected by ProLiance, so that no matter how much of the capacity has been committed to ProLiance by contract, total deliverable supplies cannot fall; and (b) ProLiance had to date acted to make better use of the existing capacity by pooling amounts held in reserve.

"To date" is a vital qualifier. The Commission issued its opinion in September 1997. More than six years have passed since then. What is ProLiance doing today? It does not take a leap of fancy to envisage a joint venture behaving itself long enough to win regulatory approbation, and only then applying the squeeze in the market. The agency found that in 1997 ProLiance was beneficial to consumers and that a "thriving robust . . . secondary market" (opinion at 40) protected third parties such as USG. It wrote: "[m]ost important to our decision is witness Feingold's uncontradicted evidence that, post-ProLiance, the market place continues to function with no ill effects." Id. at 41. "[T]he affected markets are as robust after the formation of ProLiance as they were prior to its formation." Id. at 55. That was 1997. What of 2003? The agency recognized that its record had been compiled quickly and reflected only the initial months of ProLiance's operations. "There simply is little experience with the actual operation of the alliance. . . . [E]xperience under the current agreements may indicate that their actual operation does not comport with the public interest even though we find that they do so now." Id. at 57. Reviewing this decision, the Supreme Court of Indiana made a similar point, observing that, if circumstances change, the agency may revisit the subject. 735 N.E.2d at 804. These reservations foreclose any argument that Indiana would deem the agency's decision preclusive with respect to the economic effects of ProLiance in the period after September 1997. If the findings made in 1997 would not block Indiana itself from revisiting the subject in 2003 — and they don't — then under § 1738 they do not block adjudication in federal court either. It may be that a fresh look will lead to the same conclusions reached six years ago, but nothing in the agency's decision prevents a federal court from taking that fresh look in antitrust litigation.

VACATED AND REMANDED.