Guest interview: Flip Pidot
On the past, present, and future of prediction markets
Today’s interview features
, Co-Founder and CEO of American Civics Exchange, and Founder at Sharp Square Capital, LLC, a New York-based alternative investment management firm specializing in political and other event futures.Alongside his colleagues at
, Flip is working on some interesting projects including the recent rollout of their Trump Chaos Index, which is a real-time gauge of the severity of policy uncertainty associated with the second Trump administration.Be sure to subscribe to both
and .Will:
You’ve been behind the scenes of the proliferation of modern prediction markets. What’s that evolution been like?
Flip:
I’m an old-timer, having founded American Civics Exchange back in 2007, but the modern age of prediction markets probably starts a few years earlier with the founding of (now defunct) Intrade in 1999.
When the CFTC shut them down in early 2013, it seemed like we were entering a real prediction market winter, but it was actually less than two years later that we had the PredictIt launch, going into the 2014 midterms. Having been part of the PredictIt team, I’m biased in saying so, but I give them the lion’s share of the credit in mainstreaming these markets for American users, media and even policymakers. For the next five or six years, they were pretty much the only game in town - certainly the only legal way in the U.S. to put a non-trivial amount of money on electoral and other political outcomes. As a result, we saw a lot of interest from academics who wanted to study the markets’ predictive accuracy and from armchair pundits who wanted to try to profit from their political insights.
But of course, it was the surprise results of the 2016 presidential election that doused the markets with lighter fluid. Trading volumes went through the roof, legacy forecasters with egg on their face scrambled to argue that PI’s forecast wasn’t actually better than theirs (which was false for everyone except Nate Silver’s 538, which was arguably roughly tied, depending on how you score things).
As exciting as it was, we’d braced for trade volume and media attention to fall off a cliff after the election. But to our surprise (and thanks in no small to Trump’s Apprentice-style Cabinet appointments and general carnival atmosphere that would follow), there was no shortage of entertaining and economically meaningful uncertainties to trade through the transition and of course throughout Trump’s term in office.
Fast forward to the end of the Trump term and you get to this kind of 3rd wave, the current crop, of prediction markets with the advent of Polymarket, Kalshi, and others.
So now, you’ve got blockchain-based platforms, regulated platforms, and an ensuing explosion in the number of markets, trade volumes, media and cultural awareness, and of course in the level of sophistication of traders and trade strategies. I’d argue that’s also generating more meaningful, more useful price signals across an enormous spectrum of uncertainties - not just election outcomes, but economic results, international relations, weather, pop culture, scientific breakthroughs, etc.
And that’s happening at the perfect time, when we desperately need new and better mechanisms to handicap, prepare for, and guard against some truly massive societal uncertainties.
Obviously, there’s still work to be done to convince some of the more recalcitrant policymakers that these markets serve a meaningful economic and societal purpose (one that outweighs the superficial “ick” factor that their opponents tend to lean on), but I think the popular understanding is getting there.
Will:
The FBI raided the apartment of Shayne Coplan, CEO of the blockchain-based prediction market platform Polymarket, which has stirred controversy and speculation. While the exact reasons for the raid remain unclear, many have characterized it as “political retribution” by the outgoing Biden administration given Polymarket’s accurate prediction of Trump’s election victory, which could have been seen as unfavorable by the outgoing administration.
What is behind the emerging tensions between prediction markets and regulatory frameworks? And what do you make of the raid? Was it politically motivated, in your view?
Flip:
Well, it sucks. And it certainly strikes me as capricious. But I don’t think it’s particularly likely that it was an act of political retribution for Polymarket having correctly called Trump the favorite. The reporting suggests it’s related to Polymarket “allowing” (or rather, not sufficiently zealously excluding) U.S. customers, which is of course what got Intrade in trouble a decade ago. Yes, it came on the immediate heels of the election, but other markets by and large defied the legacy forecasters in favoring Trump too. And the feds don’t really need added motivation to want to shut Polymarket down. They’ve been pretty vocally opposed to the existence of these markets and have of course been fighting with PredictIt and Kalshi in court as well. This cycle, Polymarket established itself as the 800-pound gorilla, with something like a 75% global share in electoral markets. So if there’s a platform to crack down on, they’re the obvious target. You can still call that a political axe that they’re grinding, but it’s one I think they’d be swinging just as hard if Polymarket traders had favored Harris in the end.
That doesn’t mean I think the timing is coincidental. Even since Trump won, there’s been a ticking clock for both the anti-prediction market and anti-crypto factions of the federal bureaucracy. So there may be an incentive to get as much done before January as possible.
But remember that Polymarket has been operating under the terms of an order from the CFTC since January 2022, one that presumably sets out in some level of detail what mechanisms are sufficient to prevent or deter U.S.-based traders. If they can show they’ve been implementing those measures, I wouldn’t think they’re going to encounter Intrade-like problems.
Will:
You’ve claimed that prediction markets are indeed efficient mechanisms in forecasting the outcomes of major events. You maintained this position in the run-up to the election, even as Trump’s odds of winning diverged from polling odds in extreme fashion in October, on no apparent new news.
As more information has come to light in the wake of the 2024 elections, has anything surprised you? What explains the gulf between pre-election polling odds and prediction market odds? Are pollsters losing relevance?
Flip:
2024 was the latest in a string of hard-to-forecast elections. Hard even for sober-minded, poll-skeptical analysts with their own money on the line, which means they’ve been total brainbusters for pollsters and poll-credulous forecasters, who suffer from a lack of financial or reputational accountability in their predictions (and often the ideological bent of their sponsors).
That’s not a new problem, but it was particularly pronounced during this unholy black swan of an election cycle. The Worst Debate Ever, an 11th hour incumbent withdrawal, assassination attempts, mugshots, criminal convictions, and so on.
So you had two consecutive cycles when polls understated Trump support both nationally and in the swing states, but pollsters were assuring people they’d finally fixed the Trump voter response rate problem. Meanwhile, you had this crazy stew of fundamentals (Is the economy great or terrible? Is immigration plunging us into violent chaos or is crime at an all time low? Is it better or worse to replace a manifestly non-functioning incumbent with his even more unpopular sidekick? Will anyone vote for a widely loathed convicted felon?) that seemed to make either outcome appear both impossible and inevitable.
That’s the kind of environment - messy, self-contradictory, without strong historical precedent - when markets tend to have the biggest comparative advantage vs. traditional (i.e. centralized, institutionalized) forecasting methods. As more nimble, diverse, meritocratic (admittedly imperfect) systems for discounting the future, they’re generally going to be able to sift through mountains of inconsistent data better than more rigid, opaque methods that use a small number of quantitative indicators, backfitted to a relatively small event sample.
So it shouldn’t surprise us that the markets did better than all of those legacy forecasters. And since polls provide their primary feedstock, it should be the case that pollsters will continue to lose relevance, unless and until they can figure out how to get better.
That doesn’t mean polls (even if they can’t improve) don’t have a place. Polls are certainly still among the most important inputs to prediction market traders and market prices would surely suffer if polling simply went away. It’s just that traders have shown they know better than to be blindly credulous of poll results. That’s a lesson the data journalism set hasn’t learned yet.
The black box, poll-aggregating forecasters have long complained that markets don’t provide any meaningful signal beyond what’s in their models - that they herd toward the forecasters, especially late in the season when the dumb, MAGA, Scottish teen money was done being sopped up by the few clever traders. That’s by now pretty clearly not the case, and it’s notable that it was most demonstrably not the case this cycle, the first one to feature truly diverse, deep, liquid markets.
Will:
I can envision a future where, let’s say, long/short hedge funds mitigate risk in unorthodox ways like using alternative instruments such as contracts on Tesla production/delivery targets to hedge their positions against an event. In light of Kalshi’s favorable ruling in the CFTC lawsuit, and given Robinhood’s subsequent rollout of their election prediction offering, I’d imagine other brokerage firms will begin developing prediction market products and offerings for U.S. investors.
As the asset class continues to emerge, do you foresee prediction markets increasingly catering to institutional investors (or the “smart money”), as opposed to predominantly retail speculators? Elections aside, how do you envision the future of prediction markets taking shape?
Flip:
I’m 100% certain we’ll soon see these kinds of products being used by large institutions to hedge event risk. Public policy risk alone represents the largest category of external financial risk that’s otherwise unhedgeable (excluding means we generally frown upon - lobbying, corporate PACs, outright corruption, etc.). That untapped market is the founding philosophy of American Civics Exchange and despite the huge strides made by the likes of PredictIt, Polymarket and Kalshi in the last few years - in the courts, among the policymakers, and in raising mainstream awareness of the benefits of these kinds of markets - we haven’t gotten there yet. But we will. And the good news is that the universe of highly impactful, quantifiable policy changes, from regulatory decisions, to legislative action, at the federal, state and local levels, doesn’t only visit us every two or four years.
Midterms and presidentials are the sizzle. The steak is the mundane humdrummery of the bureaucratic machine that goes on in between.
Will:
The 2024 election cycle proved a popular moment for prediction markets – at one point during the run-up to election day, Kalshi was the number one app on Apple’s app store.
Zooming out, what’s going to drive participation in these markets, and what are you and your colleagues at The Super Model currently working on?
Flip:
That steak that’s going to be delicious to the institutions, and which will ultimately drive much, much larger volumes, is also going to make most retail traders’ eyes glaze over. When those contracts emerge, we’re very likely to see a bifurcation of the prediction market space into the retail and the institutional. Some might cater to both, but they’ll be pretty different markets with very different use cases and participants, and will therefore likely optimize to different market microstructures and exchange architectures, so I’d expect to see most specialize in one or the other.
Now that the election has passed, the retail traders on Polymarket and Kalshi are mainly trading the recurring contracts - weather, sports, crypto prices, movie scores and Spotify charts. Plus the occasional international intrigue (though Kalshi is bound by Dodd-Frank to keep away from most military matters).
If the past is any guide though, once the second Trump administration is in full swing, it’ll provide a firehose of headline-driven, retail-friendly spectacles to trade. And the predictions flowing from those markets will be media catnip, especially coming off a cycle that so spectacularly increased awareness of these platforms.
The Super Model is our attempt to aggregate prediction market data to generate better models of the near future, the same way the legacy election forecasters used to aggregate polling data to generate better election models. They still do that of course; it just doesn’t really work that well any more (or at least the method is sufficiently baked into everyone’s model that there’s not much alpha signal left).
Our general thesis is that since markets are downstream of all other public data, and especially with their newfound depth, breadth, and liquidity offering talented forecasters massive financial incentives to help fine tune their pricing, they’ll inherently keep themselves right at the forecasting frontier. And to the extent we can apply an analytical layer on top of those markets, to squelch some of the noise, bias, and frenzy, and to amplify the sharpest underlying forecasters, then we can bottle some of that elusive alpha signal for ourselves and our subscribers.
***
Thanks for reading, and don’t forget to subscribe!
 | A guest post by
|