Tool · Free, no sign-up
Season simulator
Run a season of bets thousands of times at your price, edge and staking plan, and see the bad year before it happens: percentiles, drawdowns and the chance of finishing down.
This is the simulation template chapter 28 of the book promises. Say how many bets you expect to place, at roughly what price and with what edge, and whether you stake level or a fraction of Kelly. It plays the season out as many times as you ask and reports the spread of outcomes. It runs in your browser and stores nothing.
Worked example
Chapter 28's first exercise: 500 bets at 6/4 with a genuine 3 per cent edge, one unit a bet, against the same season with no edge at all. With the edge, 27.3 per cent of the thousand seasons still finished below the starting bank; without it, 48.3 per cent did. The normal approximation in the chapter 28 workings gives 29.3 per cent for the first figure.
| Percentile | Bank, 3% edge | Drawdown, 3% edge | Bank, no edge | Drawdown, no edge |
|---|---|---|---|---|
| 5th | 72.5 | 14.0 | 57.5 | 16.0 |
| 25th | 97.5 | 19.5 | 82.5 | 23.0 |
| Median | 115.0 | 25.0 | 100.0 | 30.5 |
| 75th | 135.0 | 33.5 | 117.5 | 40.5 |
| 95th | 160.0 | 48.5 | 145.0 | 60.0 |
How it works
Each bet wins with probability p = (1 + edge) ÷ decimal price. Level staking risks the same number of units every time and stops if the bank cannot cover a stake. Kelly staking risks your chosen fraction of (bp − q) ÷ b of the current bank, so stakes shrink after losses and grow after wins. Drawdown is the largest fall from a previous peak during the season.
The random numbers come from a small seeded generator, so a result can be repeated exactly. Staking sizes are explained in the chapter 29 workings, and the Kelly criterion calculator sizes a single bet.
Questions
Frequently asked questions
What does the 5th percentile mean here?
Sort all the simulated seasons from worst to best finishing bank and take the one a twentieth of the way up. One season in twenty ends at or below that figure. It is the bad year you should plan for, not the worst one possible.
Why does the same link always give the same answer?
The simulator uses a seeded random number generator, and the seed is part of the address. The same inputs and seed reproduce the same seasons exactly, so you can write a figure down and check it later. Press New seed for a fresh set of seasons.
How should I choose the edge?
Use the edge your own records support, measured against closing prices, not the one you hope for. If you have no record, run it at zero first: that shows what luck alone does to a season.
Does it store anything?
No. It runs in your browser, sets no cookies and sends nothing. The inputs live in the page address after the #, which never reaches a server.
Out now · free to read on Kindle Unlimited