Kelly vs flat-stake — honest 12+ month experience reports?

[BreakPointBrenda]

Question for anyone who's been doing this for more than a year: in real life — not theory — has Kelly staking actually outperformed flat-stake for you, or did the variance break you?

I switched from 1% flat to 0.25-Kelly (quarter-Kelly) at the start of 2025 because the maths is undeniable in the long run. In practice my drawdowns have been roughly 1.6x what flat would have produced. ROI is up but the emotional cost has been significant — I made bad decisions during the worst drawdown that I wouldn't have made on flat staking because the stake size made the loss feel like a stake-decision rather than an edge-decision.

I'm not asking 'which is right' — I know what the maths says. I'm asking how it played out for you in the messy real-world version. Has anyone reverted to flat after running Kelly? Or do you discipline yourself with hard drawdown caps?

Open to anything — sportsbook, exchange, tennis-specific, football, whatever. Particularly curious to hear from people who switched mid-year rather than starting on day one with Kelly.

[PuntingProfessor]

I've run fractional Kelly (0.2-0.3) for about six years on tennis. Two honest findings: 1) ROI is meaningfully higher than flat over multi-year windows but the year-to-year noise is much wider. 2) The emotional cost you mention is real and the literature underweights it. I capped daily exposure at 4% of bankroll and weekly at 10% — that's the discipline lever, not the staking formula itself.

If your edge estimates are even slightly biased, full-Kelly will wreck you. Quarter-Kelly is the right compromise. If quarter-Kelly is causing emotional dysfunction, that's a signal that your bankroll size or your edge confidence is mis-calibrated, not that Kelly is wrong.

[OddsArchitect]

I went the other direction — flat for years, switched to quarter-Kelly for tennis only last September, kept flat for football. Tennis ROI improved measurably; football I had no edge estimate I trusted enough to size dynamically. The takeaway for me was that Kelly is only as good as your probability estimates. If you can't articulate why your number is what it is, flat-stake.

[ServeAndValue]

Reverted from quarter-Kelly to flat in 2024 after a 24% drawdown that I handled badly. Flat is mathematically suboptimal but I sleep better and I make better in-game decisions because the per-bet stress is lower. There's a real psychological argument for flat that has nothing to do with maths.

[RafaFan_Bet]

Half-Kelly here for two years. Drawdowns hurt more than flat but I've never felt out of control. The trick that worked for me was a hard rule: if I've lost more than 8% of bankroll in seven days, I drop to 0.1-Kelly for the next week regardless of how good the bets look. Mechanical brake on emotional decisions.