These are staking rules, not probabilities.
A 25% rule means: on each fight night you put a quarter of your bankroll at risk, split evenly across that night’s picks. Whatever you finish the night with rolls into the next card. Nothing here is a chance of winning.
1Size per eventone slice of the bankroll, split across the card
Try a staking rule
—
How the fixed rules compare
$100 start, split across every pick on the card and rolled into the next one. Same window, fights and closing prices as the curve above — only the fraction differs. Ruin and the typical year come from a bootstrap, which is why they are here and not on your curve.
| Staking rule | Ended at | Chance of ruin | Bad year (5th pct) | Typical (median) |
|---|---|---|---|---|
| 100% rulewhole bankroll each card | $0 | 87.6% | $0 | $0 |
| 50% rulehalf the bankroll each card | $9,389 | ~0% | $619 | $10,046 |
| 25% rulea quarter each card | $1,411 | ~0% | $386 | $1,435 |
| 10% rulea tenth each card | $316 | ~0% | $188 | $318 |
Ruin and the typical year come from a 10,000-path bootstrap resampling the same 42 cards. Betting the whole bankroll wipes out in 88% of simulated years — its enormous average comes from a few lottery paths while the median outcome is zero.
Why the 100% rule dies on card 5. Picks per card run 1 / 3 / 9 (min / median / max). On 13 Sep 2025 the whole roll went onto that card and did not come back — “split the bankroll across the picks” puts nearly everything on one fight whenever a card sits at the low end of that range. A hard 20% cap on any single fight would have prevented it.
2Size per fightthe same slice on every pick, capped
Here the stake is per fight, not per card.
A 5% per fight rule stakes 5% of your bankroll — as it stands when the card starts — on each pick. Nothing compounds mid-card: every fight that night is sized off the same opening number, so the order they finish in never matters.
The catch is that exposure now scales with the card. 5% across 9 picks is 45% of everything you have. The cap is what stops that: set a ceiling on how much of the bankroll can be at risk on any one card, and when the per-fight rule would breach it, every pick is cut to cap ÷ picks instead.
Try a per-fight rule
—
How the fixed per-fight rules compare
$100 start, the same fraction staked on every pick of each card and rolled into the next one, capped at 50% of the bankroll per card. Same fights, same closing prices and the same window as your curve above — only the sizing rule differs.
| Per-fight rule | Ended at | Cap |
|---|---|---|
| 10% a fight10% of the roll on every pick | $2,557 | capped on 7 of 42 cards |
| 5% a fight5% of the roll on every pick | $643 | cap never bound |
| 2% a fight2% of the roll on every pick | $217 | cap never bound |
No ruin column here: these are exact arithmetic over the recorded fights, not a bootstrap, so there is one path per rule rather than a distribution. The ruin percentages in section 1 come from resampling 10,000 alternative years and have no counterpart here yet. The same applies to your own curve above: one path, not a distribution.
Fractional staking assumes you know your edge. Ours is measured over one year, not guaranteed — if the true edge is half what the backtest says, the growth-optimal fraction halves with it. Overbetting is the usual way a real edge still goes broke. Simulated results; these bets were not placed.