POSNER, Circuit Judge.

This is an intricate insurance case. Ash-land Oil (and some other companies — but we’ll ignore them to simplify our opinion) brought a fraud suit against Toy Rex Ar-nett, accountant Donald Richards, and others. The suit was under RICO, with pendent law claims under Indiana law. It resulted in a substantial judgment (of which $175,000 was against Richards), which we affirmed in Ashland Oil, Inc. v. Arnett, 875 F.2d 1271 (7th Cir.1989). Meanwhile, Truck Insurance Exchange — which had issued a liability insurance policy to Richards, Isenberg & Co., a corporation of which Donald Richards was a one-half owner — had filed a declaratory judgment action against the corporation (also in federal district court under the diversity jurisdiction), Richards himself, and Ashland, seeking a declaration that the policy did not cover the judgment against Richards. That is the case before us. The district judge entered judgment for the insurance company on its motion for summary judgment. Ashland (and the other fraud plaintiffs — but we’re ignoring them, remember) have appealed. Mr. Richards has not appealed. Indiana law governs the substantive issues.

The basis of federal jurisdiction over the declaratory judgment suit is diversity of citizenship, and there is no question that the plaintiff, Truck Insurance Exchange, is a citizen of a different state from any of the defendants. But we must consider whether the parties are properly aligned — that is, whether the plaintiff and the defendants are the real adversaries. City of Indianapolis v. Chase National Bank, 314 U.S. 63, 62 S.Ct. 15, 86 L.Ed. 47 (1941); Fidelity & Deposit Co. v. City of Sheboygan Falls, 713 F.2d 1261, 1264-68 (7th Cir.1983). Ordinarily the victim of an insured is on one side of the lawsuit and the insured and his insurance carrier are on the other; and if we realigned the parties accordingly, the requirement of complete diversity would no longer be satisfied. But such a realignment would be improper in this suit, as opposed to the first one, the fraud suit against Richards. The insurance company brought the present suit in order to disclaim any liability it might have either to Richards, the insured (or rather the half owner of the insured), or to Richards’ victims; therefore the plaintiff in this suit really is the adversary of all the defendants. Bonell v. General Accident Fire & Life Assurance Corp., 167 F.Supp. 384, 385 (N.D.Cal.1958). So there is diversity jurisdiction, but, as we are about to see, there is another jurisdictional issue.

The insurance policy, issued in 1985, covered “claims arising out of professional services rendered prior to the inception of this coverage if claim is made or suit is brought during the policy period provided no insured had any knowledge of any circumstances which might have resulted in a claim at the effective date of the policy.” (Emphasis added.) The policy period began on June 19, 1985, and ended the same day the following year. Richards’ fraud had been committed years earlier, in 1981 and 1982, when he was conducting his accounting practice through a sole proprietorship called Richards & Co., rather than through the corporation (Richards, Isen-berg & Co.) that Truck Insurance Exchange had insured. The district judge held that coverage was excluded by the proviso that we have quoted and, indepen *789 dently, by the fact that the corporation was not liable for Richards’ torts.

Ashland argues that the proviso is against public policy because it makes the insurance policy illusory. Unless Richards committed fraud in his sleep, he would have to have known, if not that he was committing fraud (for one purpose of liability insurance is to insure against false claims), of some circumstance that might result in a claim whether against him or the corporation through which he was operating when he bought the policy. So the only insured events would be claims that both arose and were presented in the brief policy period, a period shorter than any of the statutes of limitations likely to apply to suits against accountants.

? initial curiosity is that this argument should be made by Ashland rather than by Richards or his corporation. Ash-land was not the insured. It was the insured’s victim. Ordinarily a victim cannot complain about the coverage of his insurer’s liability insurance policy. We shall see that he may have a practical interest and some limited legal rights in the insurance contract, but he is not a third-party beneficiary of it. Bennett v. Slater, 154 Ind.App. 67, 289 N.E.2d 144 (1972); Winchell v. Aetna Life & Casualty Ins. Co., 182 Ind.App. 261, 264, 394 N.E.2d 1114, 1117 (1979); Baker v. American States Ins. Co., 428 N.E.2d 1342, 1347 (Ind.App.1981); Eichler v. Scott Pools, Inc., 513 N.E.2d 665, 667 (Ind.App.1987). This point is obscured by the procedural setting. Ashland did not sue the insurer directly, for a declaration that the latter was liable on the policy. It was the insurer that, having decided to deny coverage, brought this declaratory judgment action and out of an abundance of caution named Ashland as an additional defendant. Naturally Ashland thought it must have some potential interest in the insurance policy to be made a defendant in a suit to determine the policy’s meaning; and the insurer — which revealed that it thought the same thing by naming Ashland as a defendant in the first place — has tacitly conceded Ashland’s standing by not contesting the appeal on the ground of lack of standing.

We are not bound by such a concession, but we think it correct. The policy is indirectly an asset of Donald Richards’ (indirect because the insured is his corporation, not himself) — or at least it may be, depending on how the questions of coverage are resolved. And Richards is a judgment debtor of Ashland. If Richards were forced into bankruptcy, the trustee in bankruptcy would succeed to Richards’ assets and would attempt to realize on them for the benefit of Richards’ unsecured creditors, including the victims of his fraud— including therefore Ashland. Richards is not in bankruptcy; nor has Ashland obtained, through postjudgment proceedings or otherwise, an assignment of the insurance policy. But it may have been in anticipation of Ashland’s obtaining such an assignment that the insurance company added Ashland as a defendant in this lawsuit.

In these circumstances, Ashland’s interest is not so remote, improbable, or impalpable that it lacks standing to litigate the coverage of the policy — especially when we consider that- Ashland named Richards, Isenberg & Co., the actual owner of the insurance policy, as an additional defendant in its fraud suit against Donald Richards. Although an insured’s tort victim cannot (except in direct-action states) sue the insurance company directly, he has a legal right to protect his potential interest in the policy, for example by paying the insurance premiums to make sure the policy doesn’t lapse. That right is conferred in recognition that a tort victim has a practical, albeit only a potential, financial interest in the tortfeasor’s insurance policy, and the impairment of such an interest is an injury that will support standing under Article III. Maryland Casualty Co. v. Pacific Coal & Oil Co., 312 U.S. 270, 61 S.Ct. 510, 85 L.Ed. 826 (1941); Hawkeye-Security Ins. Co. v. Schulte, 302 F.2d 174, 176-77 (7th Cir.1962); Federal Kemper Ins. Co. v. Rauscher, 807 F.2d 345, 353-54 (3d Cir.1986).

That brings us to the merits. We have some difficulty even understanding *790 Ashland’s argument that the insurance policy is against public policy because the coverage it provides is illusory. Contracts are against public policy when they create incentives to commit acts that society has made illegal or, at least, disapproves of. Sayres v. Decker Automobile Co., 239 N.Y. 73, 145 N.E. 744 (1924); United States v. King, 840 F.2d 1276, 1283 (6th Cir.1988); Graham v. James F. Jackson Associates, Inc., 84 N.C.App. 427, 352 S.E.2d 878 (N.C.App.1987); E. Allan Farns-worth, Contracts § 5.2, at pp. 352-53 (2d ed. 1990). That is one of the reasons you can’t insure yourself against the consequences of your deliberate wrongdoing, Zuckerman-Vernon Corp. v. Rosen, 361 So.2d 804, 806 (Fla.App.1978), and why you can’t insure a person’s life unless you have an insurable interest in it. Lakin v. Postal Life & Casualty Ins. Co., 316 S.W.2d 542, 549 (Mo.1958). In either case the contract would create an incentive to commit bad acts. We might not care if the victims of such acts always had completely adequate legal remedies against the actor, but they don’t.

Contracts that are “illusory” in the sense of hopelessly or deceptively one-sided fail, if at all, on different grounds, such as fraud, unconscionability, or undue influence. The basic difference is that between a contract in which the parties combine to harm others (including the diffuse others evoked by such terms as “the public interest” and “public policy”) and a contract in which one party exploits the other. The former is the domain of the public policy defense, the latter of fraud, incapacity, and related defenses. Like most attempts at generalization in law, this one is only partially successful; for it is on grounds of public policy that courts refuse to enforce waivers of intentional wrongdoing. Farnsworth, supra, § 5.2, at p. 353. So the line is a little blurred; but it is clear on which side this case falls.

This misclassification is not critical in itself. If a claims-made policy that had a short policy period and no retroactive coverage would be deemed so one-sided, so illusory, as to be unenforceable on fraud or related grounds, Ashland might have a good ground for its appeal and merely have misnamed it. Let us explore this possibility. Whereas an occurrence policy protects the insured against the financial consequences of an accident or other liability-creating event that occurs during the policy period, no matter when the claim is made— it might be many years later — a claims-made policy protects the insured against the financial consequences of a legal claim asserted against him during the policy period. Sol Kroll, “The Professional Liability Policy ‘Claims Made,’ ” 13 Forum 842 (1978); Comment, “ ‘Claims-Made’ Liability Insurance: Closing the Gaps with Retroactive Coverage,” 60 Temple L. Q. 165 (1987). Given that there must be some interval between a wrongful act and the claim arising out of it, a claims-made policy might seem illusory if its coverage were confined to claims made during the policy period arising out of wrongs also committed during that period and the period was extremely short. Yet there would be nothing exploitive about such limited coverage if the insurance premium were correspondingly small, and in fact it is commonplace for issuers of claims-made policies to limit retroactive coverage by specifying a cut-off date, such as the date of the first claims-made policy issued by the insurer to this insured, so that claims based on occurrences before that date are excluded from coverage. National Cycle, Inc. v. Savoy Reinsurance Co., 938 F.2d 61, 62 (7th Cir.1991). For protection against old occurrences the insured must look to his occurrence policies. Claims-made policies that lack retroactive coverage are attractive mainly to new entities, such as Richards’ corporation, or young professionals just beginning their careers. They don’t need retroactive coverage.

Claims-made liability insurance is relatively new. It reflects primarily the combined impact of inflation and of the growing interval, under modern products law, between the introduction of a product to the market and a breakdown or other event giving rise to liability for defective design or manufacture. The longer the interval, the more difficult it is for the issuer of an *791 occurrence policy to forecast potential liability and determine the premium accordingly. And the higher the rate of inflation is during the interval, the likelier is the insured’s liability to exceed the policy limits. So both insurer and insured suffer, creating a demand for a type of liability insurance that avoids these problems. A claims-made policy does this, because it insures against events (the claims) occurring in the near rather than the remote future. True, if there is no retroactive coverage the benefits to the insured are severely limited — unless, of course, he is new to the activity and therefore has no concern with the possible inadequacies of earlier occurrence policies — but as long as the insured knows or should know this, it is unclear what grounds there are for public intervention.

But whatever the merits or demerits of claims-made policies that lack retroactive coverage, the policy that Truck Insurance Exchange issued to Richards, Isenberg & Co. provided such coverage. That is confirmed by the passage from the policy that we quoted earlier. The language to which Ashland objects merely protects the insurer against the form of fraud that consists of taking out claims-made insurance after you know or should know that a claim is about to be made against you. Like the exclusion of a known preexisting condition from a health insurance policy, the exclusion from a claims-only policy of claims based on conduct that occurred before the policy was issued and that was known to have claim potential is uncontroversially proper. Gereboff v. Home Indemnity Co., 119 R.I. 814, 820, 383 A.2d 1024, 1027 (R.I.1978); Stine v. Continental Casualty Co., 419 Mich. 89, 114-15, 349 N.W.2d 127, 138 (1984); Zuckerman v. National Union Fire Ins. Co., 100 N.J. 304, 321, 495 A.2d 395, 404 (1985); James J. Brogger & Associates, Inc. v. American Motorists Ins. Co., 42 Colo.App. 464, 595 P.2d 1063 (1979). Even the New Jersey Supreme Court, which in Sparks v. St. Paul Ins. Co., 100 N.J. 325, 495 A.2d 406 (1985), invalidated a claims-made policy because it provided no retroactive coverage (very much a minority view), at the same time upheld a policy that provided retroactive coverage with an exception, much like that in this case, for claims based on facts that the insured “knew, or could have reasonably foreseen, might lead to a claim or suit.” Zuckerman v. National Union Fire Ins. Co., supra, 100 N.J. at 319-20, 495 A.2d at 403.

Donald Richards almost certainly knew, when he had his corporation buy a claims-made insurance policy from Truck Insurance Exchange in 1985, that he was likely to be charged with fraud. The fraud suit had been brought the previous year and was pending. Richards had not yet been named as a defendant, but the fraud revolved around a corporation, owned by the Arnetts, that had been a client of Richards and had made a secret loan to him — about which he already had been deposed. He was named as a defendant in 1986, and that is the claim alleged to trigger coverage under the policy. The only basis for doubting that when he obtained the policy he knew he would soon be sued for fraud is that he failed to take out the policy in his own name. The insured was his corporation, and it was less likely to be sued than he was because it had been formed after the alleged fraud. But we must recall the broad wording of the policy proviso: “provided no insured had any knowledge of any circumstances which might have resulted in a claim at the effective date of the policy.” A corporation formed by a tortfeasor to carry on his profession is, as we are about to see, a likely target for the tort victim— enough so to bring Richards, Isenberg & Co. under the proviso. In fact, it was sued.

Although it does not matter to the outcome, we confess our unease about the district judge’s alternative ground — that the policy does not cover acts of Richards, as distinct from his corporation. Of course the named insured is the corporation, but the corporation was the successor to Richards & Co., and hence to Richards himself, since Richards & Co. was a sole proprietorship; and successor corporations are sometimes liable for the torts of their predecessors, even when, as here, they do not expressly assume such liability. EEOC v. Vucitech, 842 F.2d 936, 944-46 (7th Cir. *792 1988); Knapp v. North American Rockwell Corp., 506 F.2d 361 (3d Cir.1974); Mark J. Roe, “Mergers, Acquisitions, and Tort: A Comment on the Problem of Successor Corporation Liability,” 70 Va.L.Rev. 1559 (1984). There is a difficult tradeoff to be made in these cases. If successors are not liable, firms can externalize many of their costs of doing business, including the harms they inflict tortiously on third parties, by selling their assets to other firms, which take the assets free and clear, and by then dissolving. But if successors are liable, the transfer of assets becomes encumbered by contingent liabilities that may be difficult to value for purposes of fixing the sale price.

Where as here the successor is a corporation formed by the tortfeasor himself in what could be an effort to escape liability for fraud by transferring his assets to the corporation, the balance may well tip in favor of successor liability. The district judge, however, thought not. Her conclusion may well be correct, because there is no indication that Richards, who was of course personally liable for any frauds that he committed, transferred to the corporation assets that would otherwise have been available to persons seeking to obtain damages for his frauds. But the judge’s ground was different — that the question was foreclosed by the doctrine of collateral estoppel — and questionable.

Ashland had joined the corporation along with Richards himself in the fraud suit, and when the defendants’ counsel moved for a directed verdict, pointing out that there was no evidence of fraudulent conduct by the corporation, the judge (a different district judge from the one in the present case) had interjected: “I hope I don’t have to get into successor liability.” Then he granted the motion as to the corporation. We remarked in our decision in the fraud case that “no showing of successor liability was made.” 875 F.2d at 1282.

Collateral estoppel applies, precluding relitigation of a question, only if the question was actually adjudicated and the adjudication was necessary to the judgment. Brown v. Felsen, 442 U.S. 127, 139 n. 10, 99 S.Ct. 2205, 2213 n. 10, 60 L.Ed.2d 767 (1979), and cases cited there. (If unnecessary, the loser might have had no shot at getting the answer changed on appeal. White v. Elrod, 816 F.2d 1172, 1174 (7th Cir.1987).) It is sometimes difficult, however, to distinguish in practice between a case in which a question was not adjudicated and one in which the balance of evidence inclined so decisively on one side of the question that the adjudication was over in a wink (maybe because the party who had the burden of proof presented no evidence), in which event collateral estoppel still applies. Harris Trust & Savings Bank v. Ellis, 810 F.2d 700, 705 (7th Cir.1987); Jacques v. Local 1418, 404 F.2d 703, 705 (5th Cir.1968) (per curiam).

A helpful approach to drawing this distinction is to separate adjudication into litigation and decision. 18 Charles Alan Wright, Arthur R. Miller & Edward H. Cooper, Federal Practice and Procedure §§ 4419, 4420 (1981). A question must be litigated, but it also must be decided, for collateral estoppel to bar its relitigation. The first requirement will usually be satisfied merely by the designation of the question as one for trial (for example by its being listed on the pretrial order as an issue that is to be tried), even if no evidence is introduced; for then it is a case where the party with the burden of proof has failed to carry his burden. This may be such a case. Ashland named the Richards corporation as a defendant but put in no evidence that the corporation had committed any frauds. Logically, then, the only reason for including the corporation was on a theory of successor liability — but no evidence of that was introduced either. It is not that Ashland put in some flimsy evidence that was hooted down or even that it was found to have failed to present any evidence of successor liability. Although the complaint identifies the corporation, accurately, as the “successor” to Richards & Co., the issue of successor liability was never mentioned by any of the parties at any stage of the litigation. Only the judges, with their well-known eagerness to expand the issues in a lawsuit (we are *793 joking), mentioned successor liability — the district judge to express his relief that it was not in the case, and we to repeat that it was not in the case. It was not in the case not because it had been resolved against the party bearing the burden of proof but because it had never been put into issue. It was neither framed by the pleadings as an issue nor decided by the district judge.

Granted, there is a sense in which it had to be in the case because why else had Ashland named the corporation as a defendant when it had no evidence of the corporation’s having committed fraud? But maybe Ashland thought it could prove fraud against the corporation, never dreaming that there was another basis for imposing liability on the corporation if proof of fraud failed. Given that possibility, it is an exaggeration to say that the issue of successor liability necessarily was litigated. And litigated or not, it was not decided— the judge made that clear when he said there was no issue of successor liability in the case. The issue was extruded rather than resolved.

The grant of a directed verdict to Richards, Isenberg & Co., which we affirmed, established that it was not guilty of fraud, but not that it had no liability for the fraud of Richards & Co., its predecessor. That question remained, and remains, an open (though no longer a live) one. If the corporation did have successor liability for Richards & Co.’s fraud, then the policy issued by Truck Insurance Exchange would have come into play — were it not for the exclusionary proviso that, we have held, is both proper and applicable and requires that the judgment of the district court be

Affirmed.