Quant research

Polymarket Favourite Bias

Year
2026
Role
Solo
PythonpandasPre-registration
Resolved markets
%
Favourites win
%
Return after fees
NOT PROVEN
Verdict

The question

Is blindly backing the favourite on Polymarket profitable? A pre-registered backtest over 2,418 resolved markets gives the answer: favourites win 90.6% of the time – and the strategy still loses about 1% after fees.

Why it matters

A 90% win rate that loses money is exactly the kind of result that never survives an unregistered analysis – there is always a filter that "fixes" it. The verdict rule was frozen before the data was pulled, so NOT PROVEN is the finding, not a disappointment.