It’s Time to Prove That Political Gamblers Are More Accurate Than “Experts”

I became interested in political prediction markets when I realized that conventional sources of policy analysis aren’t all that good at forecasting.

I am particularly troubled by the record of think-tanks.

Think-tanks have proliferated since the 1980s on the promise that they can develop policy-relevant analyses. These institutions have matured enough to attract many of the most prestigious and widely feted policy experts.

Yet inaccurate predictions from the think-tank community have informed costly policy decisions from the Iraq War to the 2008 financial crisis to COVID-19 responses.

These policy failures raise questions on whether think-tanks are providing reliable forecasts, and whether alternate sources of expertise might provide better predictions.

The Untested Promise of Political Gamblers

The longer I have been involved in the political gambling community, the more I have wondered whether “degens” are producing better analysis than establishment policy professionals.

Researchers continue to investigate how political prediction market prices compare to other signals such as polls.

These studies are important, but only scratch the surface of the wisdom that may reside in the political gambling community.

Although the market prices in political gambling exchanges have a mixed record in providing accurate signals, they have succeeded in an important and under-appreciated respect: They have cultivated a community of political gamblers who are consistently beating the markets.

Long-term winners in political gambling exchanges are able to read the markets well because they have a solid grasp of how events will unfold.

I am curious how the predictions of consistent winners in political gambling markets compare to think-tank experts.

Here are two ideas on how to measure this:

Systematic Tracking

Forecasts from neither think-tanks nor the political gambling community are being tracked in a consistent, systematic way to my knowledge.  

Both American and European exchanges offer lines related to important geopolitical issues—the kinds of questions think-tank scholars write about. COVID markets on Kalshi, Polymarket, and Smarkets are an example.

Tracking and documenting forecasts from reputable think-tanks and political gambling forums on these questions may yield useful insights and comparisons.

I’m thinking something as simple as a spreadsheet like this:

QuestionThink-Tank PredictionsPolitical Gambler ForecastsMarket Price at Time of Prediction
What will be 7-day COVID case average on March 1?Brookings Institution Report, February 15: [X]Keendawg, Star Spangled Gamblers, February 15: [Y]Above 50,000: 86%
Above 100,000: 20%

It’s an open question how many direct comparisons this exercise would yield.

Think-tanks have a different set of incentives than gamblers and are typically more in the business of producing analysis and advocacy than quantifiable predictions.

I suspect that think-tank scholars—between their writings and commentary on TV/radio—are making enough predictions that certain trends would become apparent over time.

But if think-tanks are not in fact providing meaningful forecasts on important questions like the gambling community is, documenting this would itself be a notable finding.

The main challenge here is the manpower required to track this data. I suspect one researcher could pull it off if they were provided with small grant and access to relevant databases like TipRanks and NexiUni.


Forecasting tournaments would provide a more direct way of comparing think-tank experts and political gamblers.

Perhaps one of the betting exchanges could partner with a think-tank or a political risk consultancy and sponsor a tournament?


What do you think? Is this feasible? Suggestions on research design? Please contact us directly with your ideas.


Support SSG. Members get access to early picks and exclusive content.
Become a patron at Patreon!

Leave a Reply

Your email address will not be published. Required fields are marked *