## The Complete System
At the expert level, staking is not decided bet-by-bet. It is determined by a unified bankroll management system that integrates staking strategy, risk limits, portfolio rules, and review protocols into a single operating framework.
## The System Architecture
**Core staking rule:**
Stake = (EV% / 3) × bankroll × Kelly fraction (25%)
This produces stakes proportional to edge, scaled by bankroll, constrained by the 25% Kelly fraction. Reference EV of 3% = 1 unit (standard stake).
**Hard limits (override the formula if exceeded):**
- Maximum per bet: 2% of bankroll
- Maximum correlated exposure per event: 3% of bankroll
- Minimum per bet: 0.25% (avoid administrative cost on trivial stakes)
- Outright market maximum: 1.5% per market
**Stop-loss triggers:**
- Daily: pause at −5 units
- Weekly: pause at −15 units
- Drawdown: review at −25 units from peak
**Review schedule:**
- Weekly: balance reconciliation, stop-loss status check
- Monthly: ROI review, CLV review, account health update
- Quarterly: calibration test, Kelly fraction review, model performance
## The Calibration Review
Quarterly, review your Kelly performance:
- Calculate what your results would have been at flat 1% staking
- Compare to what they actually were at your variable staking
- If variable staking outperforms flat: your edge estimates are directionally correct
- If flat outperforms: your edge estimates are adding noise — simplify to flat staking
## The Long-Run Proof
A unified staking system, applied consistently over 3–5 years:
- Provides a complete, auditable record of every betting decision
- Enables rigorous performance attribution (selection quality vs staking quality)
- Builds the statistical sample required to validate edge with high confidence
- Demonstrates professional-grade risk management to potential investors or partners
This system is not glamorous. It is the operational backbone that separates sustainable professional operations from high-variance hobbyist betting.
Create a free account to track your progress and save bookmarks.
