Polymarket vs. Kalshi (and Betfair)

Who's following whom?

With hundreds of millions of dollars pouring into the presidential markets (in some cases almost singlehandedly), the two most in-focus venues are crypto-platform Polymarket and regulated futures market Kalshi.

Generally, the real-money markets have tended to agree that Trump’s odds are a bit better than suggested by other prediction platforms and media forecasters (hovering around 60-65% lately, contra the “true coin flip” narrative).

But we were curious whether the very different pedigrees and circumstances of the two markets (the qualifying jurisdictions of their traders, longevity, crypto vs. fiat denomination, media awareness, and of course, the Fredi factor) might be leading one to be more supple or sticky than the other in its price response to new news. If so, it might suggest that one is acting as the true price setter, with the other simply following suit.

At least on the main Presidential winner market, however, we see no consistent leader/laggard. What we do see is that Polymarket consistently gives Trump nearly 3 cents more of a premium than Kalshi does.

Hard to know which market’s got the sharper price, but if we ratchet Kalshi’s price up to cancel out the average cross-market spread, we can see the prices track each other very neatly and usually close to simultaneously.

It’s hard to tell with any real precision (it’s possible that fast-fingered arbs are just squeezing out the differential more quickly that we can discern, thanks in part to the chunkiness of Kalshi’s whole penny pricing), but that begs the question of why they aren’t squeezing out the persistent 3 penny differential.

Given jurisdictional and currency-related disparities between the markets, it may be hard for arbs to do so, but of course that would imply the directional harmony between the two markets should also not be attributed to arbitrage-driven efficiencies.

So neither of the two heavyweight exchanges can likely claim to be the de facto market price setter at the moment, which in terms of the credibility of the forecast, is a good thing.

It gives incremental comfort that market odds (diverging as they do from punditry and legacy forecasters) are not being systematically manipulated for electoral purposes and that - whether they turn out to be more predictive than other sources - they likely do reflect the aggregate consensus of market patricipants.

Update: At the request of commenter

MrBernoulli

, here’s the same comparison between Poly and Betfair (to which we also applied a price adjustment equal to their avevrage differential).

Their correlation and reaction timing tell a similar story.