The math behind staying in the game. Session limits, Kelly Criterion, and probability-adjusted analysis.
How big should your bankroll be? It depends on your win rate, variance, and risk tolerance.
| Profile | Buy-ins | Bankroll | Risk of Ruin | Who It's For |
|---|
The Kelly Criterion tells you the optimal fraction of your bankroll to risk on a single bet/session to maximize long-term growth while minimizing ruin.
| Fraction | Growth Rate | Max Drawdown | Best For |
|---|---|---|---|
| Quarter Kelly (25%) | ~44% of max | Minimal | New players, uncertain edge, high-variance games |
| Half Kelly (50%) | ~75% of max | Moderate | Most players — best risk/reward tradeoff |
| ¾ Kelly (75%) | ~94% of max | Significant | Experienced players with proven edge |
| Full Kelly (100%) | Maximum | Severe (50%+ swings) | Theoretical only — don't do this |
Pre-commit to session rules before you sit down. Discipline = profit.
| Rule | Limit | Why |
|---|---|---|
| Stop-Loss | 3 buy-ins | Prevents tilt-induced mega-losses. You're not playing your best after 3 buy-ins down. |
| Win Target | Soft — don't quit a good game | Leave if the game gets tough or you feel tired. Stay if fish are still paying you off. |
| Time Limit | 4-6 hours | Decision quality degrades after ~4 hours. Take breaks every 90 minutes. |
| Tilt Check | After every big loss | Lost a big pot? Walk away for 10 minutes. If you're angry, leave. |
| Table Change | If no fish remain | You're there for profit, not ego. If the weak players leave, find a better table. |
Enter your session results. Get probability-adjusted analysis, true win rate estimation, and confidence intervals.
| Situation | Action | Rule |
|---|---|---|
| You have 30+ buy-ins for next level | Take a shot ↑ | Play 3-5 sessions at the next stake. If you lose 5 buy-ins, move back down. |
| Bankroll drops below 20 buy-ins for current level | Move down ↓ | No shame. Rebuild, plug leaks, move back up when ready. |
| Win rate is positive over 50+ hours | Consider moving up | 50 hours minimum sample for live, 30K hands for online. |
| Win rate is negative over 30+ hours | Study, don't move up | Fix leaks before trying higher stakes. Variance isn't your problem if it's 30+ hours. |
| You lost 10+ buy-ins at a new level | Move down, no debate ↓ | The game is too tough. Go back, study the differences, and try again later. |
| Sample Size | Confidence | What It Means |
|---|---|---|
| 10 sessions / 5K hands | Very Low | Basically meaningless. Could be running hot or cold. Don't change anything based on this. |
| 30 sessions / 15K hands | Low | Slight indication of direction. Still highly variable. |
| 100 sessions / 50K hands | Moderate | Starting to see real patterns. Win rate estimate within ±5 bb/100. |
| 200 sessions / 100K hands | Good | Reasonable confidence. Win rate estimate within ±2-3 bb/100. |
| 500+ sessions / 250K+ hands | High | Strong confidence in your true win rate. This is where pros operate. |