Tools

Full Kelly assumes you know your edge. You estimated it from a sample, so the growth-optimal bet is smaller — and the calculator shows by how much. Enter a win rate and sample size (or a Sharpe and track-record length); get the full-Kelly fraction, the shrinkage that maximizes expected growth under estimation noise, the odds that the edge is zero, the odds that full Kelly is already over-betting, and the probability of a 20%, 50% or 80% drawdown at each fraction.

Every strategy has a size at which its edge equals its impact, and almost no paper reports it. Enter the edge, the turnover, the daily volume you can actually touch, your participation cap and how fast the signal decays; get the capacity in dollars, the AUM where the Sharpe halves, the size that maximizes dollar alpha (the fund-sized number), and where the strategy sits on the HFT-to-factor hierarchy.

Paper alpha meets the broker. Enter a gross return, volatility, turnover and the four cost components (fees, spread, timing, impact from your size against ADV); get the annual cost drag, the net Sharpe, the per-side cost at which the strategy breaks even, the turnover at which it breaks even, and the 1×/2×/3× cost-sensitivity table the production checklist asks for.

Vol targeting is a leverage rule wearing a risk-management name. Paste your daily returns or generate a calm-then-stressed regime series, choose the estimator, the target, the cap and the rebalance band, and watch what the layer actually does: the equity curve, the leverage path, the drawdown it smooths, the gap it does not, and the turnover it adds.

Enter an observed Sharpe, the sample length, the return distribution’s skew and kurtosis, and how many strategy variants you tried. Get the probabilistic Sharpe ratio, the luck ceiling E[max SR] for your trial count, the deflated Sharpe ratio, and the minimum track record length at 95% and 99% — the numbers behind the deflated-Sharpe guide.

How big is your tick archive going to be, and what will it cost to keep? Pick an asset class and data type, enter your universe and history, edit the dated assumptions, and get GB per day, year, and total; a side-by-side storage cost table; and an architecture verdict — laptop plus Parquet, a single server, or a managed warehouse.

Given a win rate, a reward-to-risk ratio, and a position-sizing rule, what does the distribution of outcomes actually look like? Simulate up to 1,000 equity curves, see the percentile fan, and get drawdown and risk-of-ruin statistics — the fastest way to build intuition for position sizing.

Should you buy a GPU, rent one, or neither? Enter your workload, edit the assumptions (all defaults dated September 2026), and get a side-by-side monthly budget with the break-even point — the numbers behind our local-vs-cloud guide.