How many bets do you really need before your sports betting ROI means anything?

I'd be careful with any fixed number. Even 500 bets won't tell the whole story if you're mixing different markets and odds ranges.

I'd rather track results by market, compare predictions with actual outcomes and check closing prices. ROI matters, of course, but a good month can still be mostly variance.

Do you guys track probability calibration too, or mainly ROI and CLV?
 
@Dennymucho one way to make "is it real?" concrete is to work out how wide the luck band is for your sample before you look at the result.

Rough maths for flat 1-unit stakes at about evens: say you bet at 2.06 on outcomes that truly win 50% of the time, which is a genuine +3% edge. One bet has a standard deviation of about 1.03 units, so after 600 bets you have about 25 units of noise either side of an expected profit of 18 units. Even with that real edge, there's roughly a 24% chance you'd still be behind at 600 bets. At longer prices (around 5.0) the noise per bet doubles, and you need about four times as many bets for the same confidence.

So monthly swings between -2% and +6% ROI are well inside what luck alone produces. What I'd watch instead at your sample size: your price against the closing price on every bet (CLV), and only split by month or market once each split has a few hundred bets.

Disclosure: I work on Pinnacle Terminal, so that's my bias. If you want to plug in your own edge, odds and bet count, we have a free variance simulator: Betting Variance and Drawdown Simulator · Pinnacle Terminal
 
Agree with not trusting one fixed number. The thing I'd add is splitting it by price range too. 100 picks mostly on short favourites and 100 picks around evens are not the same sample at all, the first one can look great on hit rate and still be flat on return. I wouldn't read much into anything under a few hundred settled, and even then I'd keep hit rate and ROI as two separate numbers.
 
The thing that helped me most was splitting by how sure I was before the game, not by month or league. The picks I rated as near certain behaved very differently from the ones I talked myself into, and that gap showed up long before the overall number settled down. Has anyone else tracked their own confidence rating next to the result and seen the same?
 
Back
Top
GOALLLL!
Odds