Why quantitative traders are using complex math models to hijack your weekend sports bets

Chicago-based trading giant DRW has spent decades profiting from mismatches between different asset classes, and now it’s building a dedicated prediction market desk targeting platforms such as Polymarket and Kalshi.

The move is one of the clearest signs yet that sophisticated “quantitative trading” firms — traders that use complex math and analysis to set up strategies — are increasingly viewing prediction markets as a legitimate trading venue rather than a niche betting product.

The firm that has been a dominant force in derivatives, fixed income and crypto markets since 1992, recently posted a job listing requiring candidates to monitor prices in real time across both platforms simultaneously, identify gaps where one is mispricing an outcome relative to the other and react quickly to profit before the pricing converges. The strategies listed in these posts — including microstructure arbitrage, cross-platform arbitrage, and news-driven momentum trading at sub-second speeds — are techniques honed in crypto derivatives markets and now being applied to sports and political events.

DRW is not alone. Wintermute, the algorithmic market maker that processes billions in daily crypto volume, is hiring algorithmic traders with experience in prediction markets. IMC, another proprietary trading firm, is also looking for quantitative traders comfortable operating across binary event contracts. Meanwhile, traditional crypto exchanges like OKX and Crypto.com have also recently posted job listings.

The hiring wave suggests institutional trading firms increasingly believe prediction markets have matured into a serious asset class and are ripe for profit.

Exploiting the mismatch

So what’s driving the sudden push? The catalyst is the volume being traded on these platforms.

Polymarket alone processed between $22 billion and $40 billion across political, economic and sports markets in 2025, up from virtually nothing three years ago and a growing share of that is concentrated in sports.

As of last week, Polymarket’s market on the UEFA Champions League Winner has processed $256 million, the 2026 NBA Champion market has done $399 million, and the 2026 NHL Stanley Cup market sits at $79 million after wild swings that saw Carolina Hurricanes rise from sub-10% implied probability to around 50% as they emerged from the Eastern Conference.

Combined, those three markets alone represent over $730 million in volume on sports outcomes, approaching the annual trading volume of some mid-sized European sports betting exchanges.

But the real reason traditional firms are pushing into this industry may not be to predict outcomes better than everyone else, market observers say.

“I don’t expect the institutional capital is contributing meaningfully to the accuracy of these markets, especially in the case of sports,” said Harry Crane, a statistics professor at Rutgers University who studies prediction market calibration.

“The accuracy of the markets is driven by specialized sports betting groups, which are much sharper at pricing sports outcomes.

Instead, Crane argues, firms such as DRW are likely applying trading techniques developed in traditional financial markets to exploit pricing mismatch.

“To the extent they are profitable, the institutions are likely applying techniques on short-term market dynamics and other technical aspects of trading that capitalize on short-term market fluctuations without insight into the event outcome.”

Simply put, DRW is not trying to predict who wins the Champions League. It is trying to profit from the way prices move before that question is answered.

A recent example appeared in the market for Britain’s next prime minister.

On the morning of May 14, Andy Burnham’s odds of becoming the next U.K. leader in the betting of “Next UK Prime Minister” on Polymarket surged from 24 cents to 43 cents as political speculation intensified around a potential Labour leadership challenge. But Betfair, the London-based betting exchange with over a billion pounds in annual volume, had already identified the move, pricing Burnham at the equivalent of 50 cents while Polymarket still showed 24 cents.

It took Polymarket hours to catch up.

For casual bettors, the gap was an interesting anomaly, but to a sophisticated quant trader, it was a textbook cross-market inefficiency waiting to be exploited.

In theory, a trader could have bought $10,000 of Burnham contracts on Polymarket at 24 cents after noticing the mismatch, before locking in $7,900 worth of profit in a matter of hours by selling when it caught up to Betfair, which would have made a profit without the event even needing to take place.

It’s a technique that has been used for decades by traditional trading firms: finding a mispriced asset across exchanges and either simultaneously buying/selling, as in arbitrage, or buying the underpriced asset and waiting for it to catch up.

Prediction markets, however, introduce an additional challenge. Betfair settles in sterling while Polymarket settles in crypto, requiring infrastructure capable of moving capital across currencies, exchanges and settlement systems.

That kind of complexity plays directly into the strengths of large trading firms, such as DRW

What’s driving them?

Beyond outright arbitrage, traders point to two structural features that make prediction markets attractive today.

The first is information lag. Traditional betting exchanges often react more quickly than decentralized prediction platforms, creating windows where prices have not yet fully adjusted.

The second is liquidity fragmentation. Champions League, NBA and Stanley Cup markets can trade simultaneously across Polymarket, Kalshi and traditional sportsbooks, meaning no single venue necessarily reflects the full market consensus.

For traders focused on forecasting outcomes rather than market structure, the toolkit looks increasingly familiar to quantitative finance.

Soccer traders often rely on “Dixon-Coles Poisson” models. The toolkit, developed in a 1997 academic paper, estimates team attack and defense strength and generates probability distributions for potential scorelines. This is something similar to how a weather forecaster assigns precise probabilities to every possible outcome rather than making a single prediction.

Meanwhile, Basketball traders frequently use “Bayesian Hierarchical” models that update assessments of team strength as new information arrives.

The goal for both models is to identify discrepancies between a model’s estimated probability and the probability implied by market prices.

A trader whose model values Arsenal’s Champions League chances at 47% while contracts trade at 43 cents may buy and profit if the market eventually converges toward that estimate.

The concept is known as closing line value, or CLV.

Crane explains why the CLV matters: “It incorporates all known pre-game information, such as injuries and lineup changes, and the sharpest players tend to wait until closer to game time to place bets because that is when the limits tend to be highest.”

Competition is here

Still, Crane remains skeptical that institutional firms will dominate sports prediction markets simply because they have arrived with larger balance sheets.

“Right now, the sharpest players in the sports betting markets are not the institutions,” he said. “The sharpest players have been in these markets for decades, and the prevailing market prices are likely driven by the same groups and the same information sources since long before prediction markets existed.”

Despite the skepticism, the talent migration is already underway.

Crypto market makers are studying sports analytics and expected-goals models, while traditional sports betting specialists are increasingly being recruited by crypto firms seeking expertise that took years to develop.

And it’s not just theoretical.

HyperLiquid, the onchain perpetuals exchange that processed over $10 billion in daily volume at its peak, is already preparing to launch prediction markets ahead of the 2026 World Cup, featuring 64 games over six weeks and generating thousands of correlated binary outcomes.

The infrastructure is being built, and the desks are now being staffed, with models working on potential outcomes.

The main question is whether institutions can outperform veteran sports bettors by finding their edge and applying sophisticated trading models used in traditional finance. But on latency, market structure and cross-platform inefficiencies, the competition has already begun.

Read more: Hyperliquid is emerging as a challenger to traditional exchanges and prediction markets, says FalconX