*931Oligopolies have always posed problems for conventional antitrust law: without something that can be called an agreement, they elude scrutiny under section 1 of the Sherman Act, 15 U.S.C. § 1, and yet no individual firm has enough market power to be subject to Sherman Act section 2, 15 U.S.C. § 2. Tacit collusion is easy in those markets, see In re Text Messaging Antitrust Litigation , 782 F.3d 867 (7th Cir. 2015), and firms have little incentive to compete on the basis of price, "preferring to share the profits [rather] than to fight with each other." Joe Sanfelippo Cabs, Inc. v. City of Milwaukee , 839 F.3d 613, 615 (7th Cir. 2016).
This appeal concerns the fine line between agreement and tacit collusion, or, put another way, conscious parallelism. Direct purchasers of containerboard ("the Purchasers") charged multiple manufacturers with conspiring to increase prices and reduce output between 2004 and 2010. We affirmed the district court's decision to certify a nationwide class of buyers. Kleen Prods. LLC v. Int'l Paper Co. , 831 F.3d 919 (7th Cir. 2016). Before and after that ruling, most of the defendants settled with the Purchasers. But two companies-Georgia-Pacific LLC and WestRock CP, LLC-decided to fight. They persuaded the district court that there was not enough evidence of a conspiracy to proceed to trial. We agree with that assessment and affirm the judgment dismissing the case.
I
A
Containerboard is the name of the material used in countless boxes: it consists of a corrugated layer of heavy paper sandwiched between two smooth pieces of linerboard. Demand is relatively inelastic, meaning that customers will not defect to other products even if the price goes up, because the available substitutes are inferior. Containerboard is manufactured at large, costly mills, which are hard to duplicate, given both the high cost of construction and the myriad of environmental laws that must be satisfied. A handful of major players dominate the industry. Those players include the original defendants in this suit: International Paper ("IP"), Georgia-Pacific, Temple-Inland, Inc., WestRock,1 Weyerhaeuser Co., Norampac Holdings U.S. Inc., and Packaging Corporation of America ("PCA").
During the early 2000s, prices for containerboard were low. But from February 2004 to November 2010, they rose dramatically. The original defendants attempted to institute price increases on 15 different occasions. The pattern was a common one. After one company announced that it *932would raise its prices for containerboard, the rest followed suit with identical or comparable increases in the ensuing hours, days, or weeks. (The one exception was a failed attempt in which there were three hold-outs.) Such efforts took place from time to time. For example, in March 2003, the defendants attempted an ultimately unsuccessful increase. Of the proposed hikes from 2004 to 2010, Georgia-Pacific, WestRock, and a non-defendant each led the effort twice. The price increases were sustained nine times, a 60% success rate.
While containerboard prices rose, containerboard production capacity fell in North America (despite the inelasticity of demand and growth throughout the rest of the globe). The initial defendants were not immune from this decline. The Purchasers' expert concluded that the defendant companies reduced their production capacity by an amount almost double that of non-defendants, though they used different strategies to accomplish this goal. They closed a significant number of mills during the class period-WestRock alone was responsible for more than a third of those closures. WestRock also took care, through measures such as buyer selection and machinery sales, to avoid adding containerboard supply into the market. Georgia-Pacific kept all its mills running, but it slowed the rate of production. It would periodically "slow back" production by idling or shutting down machines and taking extra downtime. While these practices diminished supply to the point that it sometimes pinched, in the end Georgia-Pacific never missed an order. And the company actually increased its overall capacity by acquiring a new mill in 2007.
During this period, the defendants were in regular communication. Company executives and other employees spoke by phone and at trade association meetings every few days. The record does not reveal the contents of all these conversations, but at least some dealt with the timing and pricing of interfirm trading of containerboard-a common practice.
Internal and public-facing statements made by the defendants' employees shed light on these economic developments. Some email exchanges may be read to imply that the defendants had foreknowledge of other companies' proposed increases before they were announced. For example, just before three price hikes, a PCA employee offered an opinion about how high prices would need to go over the next year and a half in order to recover the cost of capital. A Georgia-Pacific staffer wrote "the party begins" when discussing an increase attempt. A WestRock vice president emailed that the company "always follow[s] IP," even though in fact "always" was an overstatement. And a Weyerhaeuser employee discussed a specific increase two days before WestRock first made its new price public. Other statements support the inference that a coordinated plan was in place. For instance, a Weyerhaeuser employee wrote that he "made up a bunch" of information in a report about what was learned from customers about competition, asking others to "be more specific" to stay "out of anti-trust legal issues." A Norampac executive, discussing problems with the industry, said "you have to be ready to let go business if you want to keep the price up," and "everybody needs to do the same thing."
Georgia-Pacific and WestRock made their own incriminating remarks. Because some details remain under seal in this court, some of our examples are a bit vague, but we have reviewed the sealed materials and they are consistent with the remainder of the evidence. A WestRock vice-president made remarks in an email that could easily be construed as an undertaking to follow-the-leader. A different *933vice-president complained that the company "ha[d] no choice but to support [a price increase] initiative" and that WestRock "ha[d] done [its] part." At one point, a company employee wrote that the "only way to get paid is to have a 1994-95 situation where the tide rises for all boats," perhaps referring to the containerboard industry's earlier run-ins with antitrust law. See, e.g. , In re Linerboard Antitrust Litig. , 305 F.3d 145 (3d Cir. 2002). Publicly, WestRock's CEO was reported to have said that the company had a restructuring plan to "cut supply enough at [WestRock] to force price increases throughout the industry." Georgia-Pacific's president gave a speech during the period in question urging the industry to resist customer requests for price breaks.
B
In September 2010, the Purchasers filed a putative class action alleging violations of section 1 of the Sherman Act. 15 U.S.C. § 1. The district court consolidated the suit with similar actions and denied several motions to dismiss. Discovery followed. Then in March 2015, the district court granted the Purchasers' motion for class certification under Federal Rule of Civil Procedure 23. It defined the class as follows:
All persons that purchased Containerboard Products directly from any of the Defendants or their subsidiaries or affiliates for use or delivery in the United States from at least as early as February 15, 2004 through November 8, 2010.
Kleen Prods. , 831 F.3d at 922. We affirmed the certification decision while making clear that we were not addressing the merits. E.g. , id. at 928.
Back in the district court, the litigation rolled onward. The court largely denied the parties' cross-motions to exclude each other's experts. Both sides moved for summary judgment. Before the court acted on those motions, some of the defendants settled with the Purchasers. The district court granted the remaining defendants, Georgia-Pacific and WestRock, summary judgment. In a lengthy opinion that delved deeply into the Purchasers' evidence, the court concluded that the record, viewed holistically in the light most favorable to the Purchasers, did not tend to rule out that the defendants had acted independently. With only the final approval of settlement agreements pending, the district court entered partial final judgment for the remaining defendants under Rule 54(b). The Purchasers ask us to revisit that ruling.
II
Section 1 of the Sherman Act prohibits every "contract, combination, ... or conspiracy in restraint of trade...." Courts have understood for more than a century that this language does not ban all contracts, but instead reaches only agreements that restrict competition. Copperweld Corp. v. Indep. Tube Corp. , 467 U.S. 752, 768, 104 S.Ct. 2731, 81 L.Ed.2d 628 (1984) ; see also Broadcast Music, Inc. v. Columbia Broadcasting Sys., Inc. , 441 U.S. 1, 99 S.Ct. 1551, 60 L.Ed.2d 1 (1979) ; United States v. Socony-Vacuum Oil Co. (1940). In the absence of an agreement, the antitrust laws forbid only monopolization or attempts to monopolize, see 15 U.S.C. § 2, as well as a few other arrangements including anticompetitive mergers and acquisitions, see 15 U.S.C. § 18. But this case concerns only section 1 ; the plaintiffs make no claim that any of the defendants has even attempted to monopolize, much less succeeded in such an effort. See Spectrum Sports, Inc. v. McQuillan , 506 U.S. 447, 113 S.Ct. 884, 122 L.Ed.2d 247 (1993). We can therefore disregard all other antitrust theories and focus on the question *934whether the district court correctly decided that the Purchasers did not present enough evidence to permit a trier of fact to find the agreement necessary for section 1 liability. As the Supreme Court put it in Bell Atlantic Corp. v. Twombly , 550 U.S. 544, 127 S.Ct. 1955, 167 L.Ed.2d 929 (2007), "at the summary judgment stage a § 1 plaintiff's offer of conspiracy evidence must tend to rule out the possibility that the defendants were acting independently." Id. at 554, 127 S.Ct. 1955, citing Matsushita Elec. Indus. Co. v. Zenith Radio Corp. , 475 U.S. 574, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986).
It is worth recalling that an antitrust plaintiff, like all others, is entitled to try to meet that burden with either direct or circumstantial evidence. Miles Distribs., Inc. v. Specialty Constr. Brands, Inc. , 476 F.3d 442, 449 (7th Cir. 2007). Antitrust plaintiffs do not face a heightened burden to defeat summary judgment. See Omnicare, Inc. v. UnitedHealth Grp., Inc. , 629 F.3d 697, 707 (7th Cir. 2011). As Rule 56 generally commands, we draw all reasonable inferences in favor of the non-moving party, here the Purchasers. It is the substantive law, however, that establishes what the plaintiff must address. The Purchasers needed evidence that would allow a trier of fact to nudge the ball over the 50-yard line and rationally to say that the existence of an agreement is more likely than not. Put more directly, they must put on the table "some evidence which, if believed, would support a finding of concerted behavior." Toys "R" Us Inc. v. FTC , 221 F.3d 928, 935 (7th Cir. 2000) ; accord Blomkest Fertilizer, Inc. v. Potash Corp. of Saskatchewan , 203 F.3d 1028, 1032 (8th Cir. 2000) (en banc ) (reading Matsushita to require that it be more reasonable to infer a price-fixing conspiracy than permissible activity).
Armed with bountiful circumstantial evidence, the Purchasers accuse the defendant manufacturers of agreeing to restrict the supply of containerboard and thereby to create market conditions that would support significantly higher prices. The district court properly considered "economic evidence suggesting that the defendants were not in fact competing, and noneconomic evidence suggesting that they were not competing because they had agreed not to compete." In re High Fructose Corn Syrup Antitrust Litig. , 295 F.3d 651, 655 (7th Cir. 2002). It then drew and compared the corresponding inferences from each data point. See, e.g. , Omnicare , 629 F.3d at 707-20. After determining that each piece of evidence individually did not rule out the possibility of independent action, it reviewed the evidence in the aggregate, as required. See id. at 720. The court concluded that because no individual piece of evidence tended to show collusion, the combined probative value was zero. We are not so sure of that. While no single piece of information may win the day, the whole may be greater than the sum of its parts in tending to exclude the possibility of conscious parallelism. See High Fructose Corn Syrup , 295 F.3d at 655 ("[E]vidence can be susceptible of different interpretations ... without being wholly devoid of probative value....").
Nonetheless, our assessment of the district court's decision is de novo , and so we need only satisfy ourselves that we have the proper standard in mind. In re Dairy Farmers of Am., Inc., Cheese Antitrust Litig. , 801 F.3d 758, 762 (7th Cir. 2015). Viewing the evidence and reasonable inferences in the Purchasers' favor, we ask whether they have produced any evidence that would rule out the hypothesis that the defendants were engaged in self-interested but lawful oligopolistic behavior during the relevant period. Despite the volume of evidence the Purchasers submitted in opposition *935to summary judgment, we find ourselves in agreement with the district court's ultimate conclusion. We discuss below only the most significant evidence against Georgia-Pacific and WestRock; those interested in a more comprehensive account should refer to the district court's opinion, Kleen Prods. LLC v. Int'l Paper , 276 F.Supp.3d 811 (N.D. Ill. 2017). We conclude that nothing in this record would permit a trier of fact to conclude that the defendants were colluding, rather than behaving in their independent self-interest.
III
A
We start with some structural evidence about the containerboard industry. As we noted in our earlier encounter with this litigation, the market has certain structural features that make it "conducive to successful collusion," such as a small number of manufacturers, vertical integration, inelastic demand, a standardized commodity product, and high barriers to entry. Kleen Prods. , 831 F.3d at 927-28. These characteristics make it easier for companies either to form a cartel or to follow the leader independently. Text Messaging , 782 F.3d at 871-72 ; In re Chocolate Confectionary Antitrust Litig. , 801 F.3d 383, 397 (3d Cir. 2015). We explained why this is so in our 2015 Text Messaging opinion:
[I]f a small number of competitors dominates a market, they will find it safer and easier to fix prices than if there are many competitors of more or less equal size. For the fewer the conspirators, the lower the cost of negotiation and the likelihood of defection.... But the other side of this coin is that the fewer the firms, the easier it is for them to engage in "follow the leader" pricing ("conscious parallelism," as lawyers call it, "tacit collusion" as economists prefer to call it)-which means coordinating their pricing without an actual agreement to do so. As for the apparent anomaly of competitors' raising prices in the face of falling costs, ... this may be not because they've agreed not to compete but because all of them have determined independently that they may be better off with a higher price. That higher price, moreover-the consequence of parallel but independent decisions to raise prices-may generate even greater profits (compared to competitive pricing) if costs are falling, provided that consumers do not have attractive alternatives.
Text Messaging , 782 F.3d at 871-72. Because of the competing inferences that can be drawn from this market structure, the district court properly found that the economic evidence did not tend to exclude the possibility of independent action.
B
Next, we turn to more specific evidence that the Purchasers offered. In establishing both defendants' failure to compete, the Purchasers rely heavily on the 15 price hikes that occurred over the class period. But one must take care with the inferences that can be drawn from such evidence. Following a competitor's price increases can be consistent with rational self-interest in oligopolies. See Text Messaging , 782 F.3d at 874-75. Each firm in a tight oligopoly might think that it will reap greater profits if it imitates, rather than undermines, its peers' price hikes. Brooke Grp. Ltd. v. Brown & Williamson Tobacco Corp. , 509 U.S. 209, 227, 113 S.Ct. 2578, 125 L.Ed.2d 168 (1993) ; Valspar Corp. v. E.I. Du Pont de Nemours & Co. , 873 F.3d 185, 191 (3d Cir. 2017). And it might reach that conclusion without any conscious coordination with its competitors. For that reason, "it is not a violation of antitrust law for a firm to raise its price, counting on its *936competitors to do likewise (but without any communication with them on the subject)...." Text Messaging , 782 F.3d at 876.
1
The task before any plaintiff is thus to find and produce evidence that reveals coordination or agreement (even a wink and a nod-formal agreements have never been required for purposes of Sherman Act section 1 ). See id. ; Blomkest Fertilizer , 203 F.3d at 1033. For instance, foreknowledge of price increases may be persuasive evidence that an agreement was afoot. See Chocolate Confectionary , 801 F.3d at 408 ; In re Coordinated Pretrial Proceedings in Petroleum Prods. Antitrust Litig. , 906 F.2d 432, 455 (9th Cir. 1990). The Purchasers here attempt to carry their burden by emphasizing the timing of the price increase attempts, which they describe as "lockstep." They urge us to draw the inference that such tight congruence of price movements could not have occurred unless the competitors had an inside scoop. But a close look at the record reveals that the Purchasers overstate how coordinated these hikes actually were. Different manufacturers, including non-defendants, led the attempts. See Reserve Supply Corp. v. Owens-Corning Fiberglas Corp. , 971 F.2d 37, 54 (7th Cir. 1992) (finding that a change in who is the price leader suggests that manufacturers could independently decide whether to participate in price increases). Sometimes companies followed suit over a month later. Even the attempts that saw quick turnaround times do little to raise suspicions. If it is in a company's self-interest to imitate a price leader's increase, why wait to enjoy the benefit? The Purchasers accuse the defendants of lying when they claim to have explored independently a possible increase. But there is no evidence supporting this allegation.
The Purchasers' "proof" of prior knowledge amounts to nothing more than speculation. They emphasize a March 2004 PCA memorandum that said "at least three $40-50 increases over the next 18 months" were needed to recoup the cost of capital. By September 2005, three attempts to raise prices had indeed occurred. But this supposed smoking gun could be nothing more than a somewhat accurate industry prediction. That two of the increases were for $50 is unsurprising, given that most of the 15 attempts were for $40 or $50. More tellingly, the PCA employee did not accurately predict three successful increases, since the second one failed and the third was only for $30.
The evidence that Georgia-Pacific provided or received advance notice is even weaker. The best that the Purchasers offer is a comment made by a Georgia-Pacific employee that "the party begins," following a discussion that a few manufacturers had announced an increase. This remark could merely express enthusiasm about the upward trend in pricing.
Another aspect of the Purchasers' argument is that the rising prices throughout the class period reflect an abrupt change in business practices. If that is an accurate description of what happened, it might support an inference of conspiracy. But before the inference can be drawn, we have looked for a shift in firm behavior, as opposed to external market conditions. See Toys "R" Us , 221 F.3d at 935 ; Chocolate Confectionary , 801 F.3d at 410. In the present case, the Purchasers' evidence reveals only changed market conditions. For instance, they point to complaints that manufacturers made about aggressively competitive pricing that took place before, but not during, the class period. Yet the shift may be explained by external factors, such as the emergence from the economic downturn of 2008, which occurred in the *937middle of the class period. And in terms of the companies' behavior-the relevant inquiry-the manufacturers had attempted to raise prices before the class period as well. A continuation of a historic pattern-including of parallel price increase announcements-does not plausibly allow one to infer the existence of a cartel. Valspar , 873 F.3d at 196 ; Chocolate Confectionary , 801 F.3d at 410.
A further strike against the Purchasers' case is the failure rate of the manufacturers' efforts: 40% of the attempted increases did not hold. The district court pondered why a company would risk treble damages by colluding on an often-ineffective plan when tacitly following price hikes had no downside risk. Cf. Text Messaging , 782 F.3d at 878. Perhaps the Purchasers have a good answer to that question: when potential profits are in the billions, even 60% odds provide a substantial incentive. But that at best leaves matters in equipoise.
If this was a cartel, it would have tried to impose disciplinary measures on the "cheaters" who did not go along with the price increases. But that type of evidence is conspicuously absent, even though nearly half the price hikes failed. See Petruzzi's IGA Supermarkets, Inc. v. Darling-Del. Co., Inc. , 998 F.2d 1224, 1233 (3d Cir. 1993) (noting that "a cartel cannot survive absent some enforcement mechanism"). The Purchasers propose two possible mechanisms for enforcement, but they have not pointed to any evidence indicating that either one was used. It is true that a cartel may exist with only soft measures of control or ineffective enforcement. See In re Broiler Chicken Antitrust Litig. , 290 F.Supp.3d 772, 796-97 (N.D. Ill. 2017) (evidence of an enforcement mechanism is not always required). Even if that is so, however, the absence of evidence about enforcement does nothing to dissipate the inference of independent behavior. We are still left with price increases that appear to be just as consistent with independent action as with collusion.
2
The second half of the Purchasers' theory focuses on supposedly coordinated reductions of output through mill closures and machine slowdowns. Supply behavior is highly relevant because price-fixing arrangements often function through restrictions of output. See Westinghouse Elec. Corp. v. Gulf Oil Corp. , 588 F.2d 221, 226 (7th Cir. 1978). But not every supply-side change is equally suggestive of a conspiracy. Conduct that is easily reversed may be consistent with self-interested decision-making.
An example illustrates the point. Suppose Company X takes its machines offline more frequently in order to reduce its supply. If competitors follow suit, and industry-wide production falls, all companies can charge more for the commodity and potentially reap greater profits. See Text Messaging , 782 F.3d at 877. But what if instead the competitors maintain their supply and woo Company X's customers. Because Company X's reduction strategy was flexible, it can quickly get its machines running and filling orders again, minimizing any losses. In contrast, if Company X had lowered its production by selling its mills or equipment, it could not rapidly undo its efforts while competitors came knocking on customers' doors. This inability to stave off potential losses has earned the name "perilous leading." PHILLIP E. AREEDA & HERBERT HOVENKAMP, ANTITRUST LAW ¶ 1425d (4th ed. 2018). Firms take significant risks by reducing their output in an inflexible manner, unless there is an enforceable agreement in place to ensure that competitors will follow suit. Petroleum Prods. , 906 F.2d at 463 ;
*938In re Plasma-Derivative Protein Therapies Antitrust Litig. , 764 F.Supp.2d 991, 1002 (N.D. Ill. 2011). Because perilous leading makes "little economic sense" absent coordination, evidence of less-reversible supply restrictions supports an inference of conspiracy. Broiler Chicken , 290 F.Supp.3d at 798.
During the class period, the North American market saw a drop in overall capacity for containerboard. The original defendants collectively were responsible for 19 mill closures. Yet Georgia-Pacific not only kept its mills open; it also purchased a new mill. The Purchasers respond that Georgia-Pacific underutilized its machines, but Georgia-Pacific has an answer for that: its run-to-demand strategy. Under this strategy, which dated back to 1999, Georgia-Pacific aimed to produce just enough containerboard to fill orders without creating excess inventory. Internal communications suggest that this strategy led to some close calls when filling orders, but Georgia-Pacific always found a way to meet its customer demand. Moreover, its acquisition of a mill allowed it to increase its production capacity over the class period.
The Purchasers' strongest evidence undercutting Georgia-Pacific's account are comments in performance reviews that credit employees for getting price increases by keeping inventory low. Because Georgia-Pacific did not have sufficient market power to alter containerboard pricing on its own, the Purchasers insist that these statements can be understood only as proof of an anticompetitive agreement. But underusing machinery is the kind of flexible behavior that is consistent with rational attempts to raise prices through watchful attention to one's competitors' actions. Far from perilous, had Georgia-Pacific's efforts not paid off, it could have increased its output quickly. Georgia-Pacific's supply behavior does not point towards its having a role in any conspiracy.
3
As ammunition against both defendants, the Purchasers cite the frequent contacts that company executives had by phone and at trade association meetings. They allege that the defendants' regular communications and trades served as opportunities for collusion. Some courts have held that this type of information flow, especially between executives, may be probative of conspiracy. E.g. , Stanislaus Food Prods. Co. v. USS-POSCO Indus. , 803 F.3d 1084, 1093 (9th Cir. 2015). But having the opportunity to conspire does not necessarily imply that wrongdoing occurred . Monsanto Co. v. Spray-Rite Serv. Corp. , 465 U.S. 752, 762, 104 S.Ct. 1464, 79 L.Ed.2d 775 (1984) ; Weit v. Cont'l Ill. Nat'l Bank & Trust Co. of Chi. , 641 F.2d 457, 462 (7th Cir. 1981). Especially when companies have legitimate business reasons for their contacts, plaintiffs must offer some evidence that moves beyond speculation about the content of what was conveyed. See Text Messaging , 782 F.3d at 878 ("And as there is no evidence of what information was exchanged at these [trade association] meetings, there is no basis for an inference that they were using the meetings to plot prices [sic ] increases."); accord Chocolate Confectionary , 801 F.3d at 409 ; In re Musical Instruments & Equip. Antitrust Litig. , 798 F.3d 1186, 1196 (9th Cir. 2015).
The Purchasers have no evidence indicating that the executives discussed illicit price-fixing or output restriction deals during their calls or meetings. They rely instead on the frequency and timing of the contacts. For example, 20 calls were made in the days around a Georgia-Pacific-led price increase, despite the fact that the company had predicted flat pricing just *939two weeks earlier. That is not enough. We cannot put much stock in the frequency of contacts, given the amount of trading that was taking place among the firms. See Dairy Farmers of Am. , 801 F.3d at 763 (where companies were not only competitors but also among each other's largest suppliers and largest customers, plaintiffs needed to point to a communication that "suggest[ed] a meeting of the minds to fix prices"); High Fructose Corn Syrup , 295 F.3d at 659 (finding nothing suspicious about occasional interfirm trades unless firms could supply their customers at a lower cost).
Furthermore, we hesitate to impugn the companies' intentions solely from the timing of the contacts. To be clear, we do not see the frequency of the calls and meetings as evidence tending to exclude collusion. Such a rule would create an incentive for businesses to make constant phone calls in order to immunize themselves from antitrust liability. Here, however, though some trade association meetings occurred before price-increase proposals, most of Georgia-Pacific's announcements were not preceded by meetings. The Purchasers' speculation about the content of the frequent interfirm contacts is not enough to create a jury issue.
4
Incriminating remarks by defendants' employees can support the inference that a conspiracy existed. See, e.g. , High Fructose Corn Syrup , 295 F.3d at 662. The Purchasers flag a few comments that they see as particularly inculpatory. For starters, a Weyerhaeuser employee wrote that he "made up a bunch" of information in a report and instructed others to "be more specific" to keep the company "out of anti-trust legal issues." In addition, while discussing problems with the industry, a Norampac executive said "you have to be ready to let go business if you want to keep the price up," and "everybody needs to do the same thing."
Yet even if these statements are enough to create a triable question about the presence of an agreement generally, they are not enough to show that Georgia-Pacific was a part of that cartel. On this front, the Purchasers present little proof. They point to a speech in which Georgia-Pacific's CEO supposedly suggested that the industry should "say 'no' on deals" that, though competitive, are not profitable. But that is hardly an earthshattering insight, even if proof of the statement were possible without the use of hearsay contained in newspaper articles reporting on the speech. See Eisenstadt v. Centel Corp. , 113 F.3d 738, 742 (7th Cir. 1997). And the account of an attendee rejects the rumors, stating that this message "was not said nor implied." And the CEO claimed in his deposition that he was speaking only about Georgia-Pacific , not the industry, needing to decline deals. Like their economic proof, the Purchasers' noneconomic evidence-even when viewed with the parallel conduct-does not exclude the possibility that Georgia-Pacific acted in a self-interested but permissible way.
C
Some of the Purchasers' evidence was particular to WestRock, to which we now turn. There is a wrinkle in its potential liability: in June 2010, just shy of the close of the class period, WestRock received a discharge in bankruptcy, for which it had filed in 2009. At that moment, it was free of any antitrust liability incurred up to the date of discharge. See In re Travel Agent Comm'n Antitrust Litig. , 583 F.3d 896, 902 (6th Cir. 2009). Nonetheless, WestRock is potentially liable for the alleged conspiracy if there is evidence it rejoined the cartel post-discharge.
*940Kleen Prods. , 831 F.3d at 930 ; see also O'Loghlinv. Cnty. of Orange , 229 F.3d 871, 875 (9th Cir. 2000) ("A 'fresh start' means only that; it does not mean a continuing licence [sic ] to violate the law."). We must therefore look to see if there is evidence that would permit a trier of fact both to find the initial agreement, and to find that WestRock rejoined that agreement after its discharge.
1
As we have explained, the parallel price hikes alone do not suffice to permit a jury to find a cartel. But the Purchasers cite evidence hinting at WestRock's conspiratorial involvement. For example, they point to an email from Weyerhaeuser that discussed a $50 increase just days before WestRock announced it. Though a publication's email blast had made public the existence of a future attempt, it had gotten the amount wrong. Weyerhaeuser had the right number.
Yet even if this is enough to create a fact question about WestRock's original participation in the alleged agreement, it does nothing to establish that it rejoined the agreement post-discharge. During the relevant period, in July 2010, WestRock did participate in an unsuccessful attempt to hike prices. The Purchasers insist that this reveals more than parallel conduct, given an email between WestRock staff that the company "always follow[s] IP [International Paper]." But, as we have noted several times, merely following a leader is not the same as agreeing to do something. Also of little probative value is the fact that a WestRock vice president met with other manufacturers on the day between the first defendant's joining the increase and WestRock's decision to follow suit. Before this meeting, the increase had been floated by a non-defendant. See Valspar Corp. v. E.I. Du Pont de Nemours , 152 F.Supp.3d 234, 247 (D. Del. 2016), aff'd 873 F.3d 185 (declining to infer that a meeting provided an opportunity to conspire when price increase announcements occurred before and after the meeting). And WestRock offered evidence that since April it had been contemplating that prices would go up. Purchasers' parallel pricing evidence does not provide a hook for WestRock's liability.
2
Perhaps the most compelling evidence of collusion is WestRock's supply restrictions. During the class period, WestRock closed seven of its mills and took other steps to reduce capacity. WestRock attempts to rationalize this behavior in various ways. It claims that the closures were part of a 2003 restructuring plan to get rid of inefficient plants in light of its purchase of a highly efficient mill. It also asserts that it made certain sales decisions in light of a green marketing plan where buyers would assume a mill's associated environmental liabilities. And WestRock reminds us that it sold the mills while it was under the oversight of the bankruptcy court, a committee of creditors, and financial advisors.
These explanations are all plausible. Yet they do not overcome the inference of conspiracy given that, unlike Georgia-Pacific's reversible cuts, WestRock's supply behavior could not be undone easily. Such perilous leading risked significant losses. Furthermore, a vice president wrote that WestRock "ha[d] done [its] part," implying it played a role in a larger agreement. And other company emails state that restricting supply would help raise prices, something no manufacturer could do alone.
While this discussion may suggest that the Purchasers win the day, the insurmountable problem is one of timing: these events occurred pre-discharge. Even assuming *941(favorably to the Purchasers) that WestRock was part of a cartel, they fall short on presenting evidence that WestRock was involved post-discharge. In those months, WestRock did some machine maintenance, but it did not close any mills and generally operated at a high level of production and capacity.
3
Last, we consider the other evidence that the Purchasers lodged against WestRock. This includes incriminating statements made by WestRock employees and an article discussing the CEO's statements that the company needed a restructuring plan to "cut supply enough at [WestRock] to force price increases throughout the industry." The latter is inadmissible hearsay. Eisenstadt , 113 F.3d at 742. The Purchasers maintain nonetheless that the CEO's deposition testimony confirms that he was signaling indirectly the company's commitment to a price-fixing plan. It does not.
Worse for WestRock are two vice-presidents' remarks that the company had "little choice" or "no choice" but to join the price increases, even though one was not supported by supply and demand. The most plausible explanation for its decision to go along with a price hike out of obligation, when that hike is not economically justified, is that WestRock had committed to an agreement. See In re Flat Glass Antitrust Litig. , 385 F.3d 350, 369 (3d Cir. 2004) (finding probative documents emphasizing that competitors had to hold the line on price increases that were not economically supported). Also inculpatory was a staff member's email to a vice-president, among others, that "the only way to get paid is to have a 1994-95 situation where the tide rises for all boats." The Purchasers encourage us to read this statement as a reference to the containerboard industry's earlier antitrust violations. See Linerboard , 305 F.3d 145. These statements are some of the Purchasers' strongest evidence that WestRock's price hikes and mill closures were more than just conscious parallelism. But again, they all occurred pre-discharge, and so they say nothing about WestRock's post-discharge conduct.
The Purchasers finally offer some economic and noneconomic evidence that could suggest suspicious activity. But even if it is credited, it is not enough to permit a trier of fact to find impermissible coordination. Statements by Georgia-Pacific staff are not enough to cast a cloud over its follow-the-leader price increases and flexible production adjustments. And while some of WestRock's behavior, particularly its mill closures, gives us pause, the Purchasers fail to establish that anything the company did post -discharge amounts to rejoining any existing conspiracy. This case shares many similarities with our decision in Text Messaging . In both situations, the plaintiffs did not discover a "smoking gun or ... additional circumstantial evidence that further tilts the balance in favor of liability." Id. at 871 (citation omitted). The Purchasers may be right that the containerboard industry got savvier at hiding its antitrust violations. But unfortunately for them, they "failed to carry the burden" of "establishing a prima facie case of explicit collusion," offering "no more than a plausible interpretation" of the defendants' anticompetitive conduct. Id. at 876.
IV
The outcome of this case flows directly from both the limitation in section 1 of the Sherman Act to anticompetitive agreements and the Supreme Court's cautions against interfering with individual firm behavior in ways that could inadvertently distort incentives to compete. In Matsushita, the Court warned against "mistaken *942inferences ... [that] chill the very conduct the antitrust laws are designed to protect.' " Matsushita , 475 U.S. at 594, 106 S.Ct. 1348 ; see also Text Messaging , 782 F.3d at 874 (expressing pragmatic concerns about prohibiting conscious parallelism).
Scholars, lawmakers, and courts have yet to agree on a regulatory regime that can address oligopolistic behavior that leads to higher prices and reduced consumer choice, without stifling normal business activity. For now, we follow established law to the effect that " 'conscious parallelism' has not yet read conspiracy out of the Sherman Act entirely." Twombly , 550 U.S. at 552, 127 S.Ct. 1955 (citation omitted).
Because the evidence proffered by the Purchasers does not tend to exclude the possibility that Georgia-Pacific and WestRock engaged only in tacit collusion, we AFFIRM the judgment of the district court.