Are you actually any good? Measuring it properly

"I'm hitting 58%" is the most common self-assessment in betting and one of the least informative. It does not say what odds you were taking, how confident you were, or whether 58% is good for the markets you play. A forecaster who says 90% and is right 90% of the time is excellent. One who says 90% and is right 60% of the time is dangerous — and both can post the same win rate depending on what they bet.

Brier score: one number that works

The Brier score is the mean squared error of your probabilities. For each resolved call, take your probability, subtract the outcome (1 if it happened, 0 if it did not), square it, and average.

Brier = mean( (forecast − outcome)² )

Lower is better. Zero is perfect. The number to beat is 0.25, which is what you get by saying 50% to everything — if you cannot beat that, your opinions are worth less than a coin.

Calibration: are your 70s really 70s?

Bucket your forecasts by confidence and check how often each bucket came in. Well-calibrated means the things you called 70% happened about 70% of the time.

That table is the single most useful diagnostic in forecasting, and the pattern in it is the most common one there is: fine when hedging, overconfident when certain. If that is your curve, the fix is not to research harder — it is to pull your high-confidence numbers toward the middle and size them smaller.

How many calls before this means anything?

More than you would like. Ten resolved calls tells you nothing — a 60% forecaster goes 4-for-10 often enough that you would never notice. Rough guidance:

• Under 30 resolved calls: you are reading noise. Do not change your process, and do not tell anyone your hit rate.
• 30–100: a Brier score starts to be suggestive. Calibration buckets are still thin.
• 100+: the calibration curve begins to be trustworthy, bucket by bucket.

This is why anyone advertising a hit rate off a handful of results is either new or selling something. A track record is a slow thing to build and there is no way to accelerate it.

Closing-line value: the faster signal

Because outcomes are slow, professionals lean on a proxy: did you get a better price than the market closed at? If you consistently buy at 55c things that close at 60c, you are finding real information before the market does — and CLV converges far faster than win rate, because every bet gives you a reading rather than one bit of win/lose.