Bet Sizing Lab — Milkman Trades

Original idea, model & spreadsheet by the Playing For Doubles newsletter — thank you! 🙏 This page ports their workbook to the browser so every output (including the 10,000-path Monte Carlo) updates the instant you move a slider.

Inputs — drag or type

Win rate (w) %

W/L ratio (R = avg win ÷ avg loss) ×

Bet size (stake, % of equity) %

Average loss (% of stake on a loser) %

Number of trades (N)

Starting equity $

Target multiple ×

Advanced — Monte Carlo settings

Stair-step trigger (resize after this gain) %

Dimmer floor (smallest bet fraction)

Simulations (paths per run)

Average win (% of stake)—

Risk per trade (% of equity)—

Full-Kelly risk—

Your risk ÷ Kelly—

Per-Trade Mechanics

Win: gain on equity

—

×R payoff

Loss: cost on equity

—

= bet × avg loss

Arithmetic EV / trade

—

per trade, % of equity

Geometric growth / trade

—

what compounds

Full Compounding — after N trades

Median ending multiple

—

Mean ending multiple

—

skewed by lucky paths

Trades to target

—

Arithmetic breakeven win rate

—

= 1 / (1+R)

Growth breakeven win rate

—

to compound up at this risk

Sizing Schemes — deterministic estimates

Typical ending: median for compounding, mean for flat, typical-path estimate for stair-step. Full distributions, drawdowns & the dimmer rule are in the Monte Carlo below.

Metric	Compounding	Flat	Stair-step
Typical ending (after N)	—	—	—
Trades to double (2×)	—	—	—
Breakeven win rate	—	—	—

Monte Carlo — 3 sizing schemes, identical random draws

Compounding — fixed % of current equity Stair-step — flat $ until +26%, then resize Dimmer — halve after loss, double back after win —

Compounding

Stair-step

Dimmer

Equity multiple (log scale) over N trades. Bright line = median path; bands = middle 50% and 90% of simulations; faint lines = sample paths.

Where you end up — distribution of ending multiples (log scale)

Metric	Compounding	Stair-step	Dimmer

Assumes independent trades. With independent trades, stair-step & dimmer mainly affect drawdowns, not raw growth.

Sensitivity — win rate × W/L ratio

Both grids use your current risk per trade (—) and N (—). Click any cell to load that win rate & R into the model. Green = good/fast, red = poor/slow.

MEDIAN ENDING MULTIPLE

TRADES TO DOUBLE (2×)

GROWTH BREAKEVEN WIN RATE

Minimum win rate needed to compound upward at the current risk per trade.