Bankroll & Kelly
Fee-aware position sizing, hedges, and a drawdown simulation.
The tool runs at /bankroll and needs a $49.99/mo membership. There is no free tier and no trial, so opening it signed out lands on the pricing page rather than the board. This page is the open documentation of what it does.
Traders rarely blow up because their reads were wrong. They blow up because their sizing did not survive the reads that were wrong. This page is three pieces of arithmetic for that problem: how much to stake, what a hedge would actually lock, and what a hundred trades at that sizing tends to feel like.
Everything is fee-aware, because on these venues the fee is not a rounding error. Kalshi charges a per-contract fee that peaks in the middle of the price range, and a Polymarket entry costs gas and spread even though the venue charges no commission.
The Kelly calculation
You give it the price of the contract you are buying, your probability that it settles at a dollar, your bankroll, a Kelly fraction and a hard per-market cap. It computes the all-in cost per share including the venue fee, the edge per share, the breakeven probability — which is simply that all-in cost — expected value per dollar staked, the full-Kelly fraction, the fraction after your chosen multiple and cap, the dollar stake, the whole contracts that stake buys, and the expected log-growth at that sizing.
When the edge after fees is gone it returns zero, and the correct stake is zero dollars. That is a real answer, not an error state, and it is the one the fee model exists to produce.
Fractional Kelly and the cap
Quarter, half and full Kelly are offered, with quarter as the sane default. Full Kelly is optimal only if your probability is exactly right, which it never is; fractional Kelly is what you use when you are honest about that.
A separate hard cap as a percentage of bankroll binds before the Kelly maths in the many cases where a large stated edge produces an absurd stake, and the result tells you explicitly when the cap was the binding constraint rather than the arithmetic.
The hedge and exit maths
Given a position — contracts held, the price you paid — and a current quote for the opposite side, the hedge panel computes the profit in both branches for any hedge fraction, and the single number a full hedge locks in regardless of outcome.
It treats sunk cost as sunk: what you already paid is a constant in every branch, so it is folded into the reported profit rather than into the decision. The useful landmark it prints is the breakeven opposite price — hedging locks a profit while the opposite side's all-in cost is below one minus your entry, and locks a loss above it.
The drawdown simulation
Two thousand Monte Carlo paths of a hundred trades each, at the fraction the calculator just computed, compounding multiplicatively. It reports the median, fifth-percentile and ninety-fifth-percentile ending bankrolls, the probability that a path dipped below half its starting value at any point, the median worst peak-to-trough drawdown, and forty sampled equity curves to look at.
The number most people find useful is not the median. It is the risk-of-halving figure, which is what a sizing choice feels like from the inside on a bad run.
What it cannot tell you
Every tool has a boundary, and knowing where it sits is the difference between using one well and being misled by it. For Bankroll & Kelly:
- Kelly is only as good as your probability, and the calculator neither supplies one nor can check the one you typed. Garbage in, confidently sized garbage out.
- The simulation fixes the stake fraction and ignores whole-contract rounding, the dollar cap and whether the same opportunity exists next time.
- It assumes independent, identically distributed trades. Real markets correlate, and a portfolio of correlated positions is riskier than the same trades taken one at a time.
- Fees are the venues' published formulas, not your actual fills. Slippage on a thin book is not modelled at all.
- Expected value and expected growth are averages over paths that will not happen. They are not projections of your account.
- Bankroll settings live in one browser's storage, shared with the portfolio exposure block. They do not sync anywhere.
What it is built on
- The venues' published fee structures: Kalshi's per-contract formula, and a gas-and-slippage allowance for Polymarket.
- The shared staking maths in lib/staking.ts, which is the same module that prices the arbitrage splits, so a lock and a Kelly stake can never disagree about cost.
- Your own inputs. Nothing on this page reads a live market — it is arithmetic on numbers you supply.
Related reading
- Kelly Criterion for Prediction Markets: A Fee-Aware Position Sizing Calculator for Polymarket and Kalshi — Guide
- How to Split Stakes on Both Sides of a Prediction Market Arb (Polymarket vs Kalshi) — Guide
- Bankroll Management for Prediction Market Trading: Position Sizing, Kelly Staking, and Drawdown Control — Strategy
Tools that pair with it
Paste a Polymarket wallet and read its book, live.
Practice book at real venue prices. No money at risk.
The same event priced differently on Polymarket and Kalshi.
WhaleTracks is informational analytics, not financial advice. Market data comes from Polymarket, Kalshi and Manifold's public APIs; WhaleTracks is not affiliated with any of them. Past performance does not guarantee future results.