💰 Capital & Risk
Having a validated edge is half the picture; the other half is how much you risk on each trade. This branch —its own menu, just before Help— answers that question. Its north star is not the size that would have made the most in the past, but the one most likely to survive and grow. Confidence, not profit.
Conceptually inspired by tools like Adaptrade's MSA, but original work of our own. It brings together a quick Calculator, a Sizing Lab that chains four tools (comparing methods, Monte Carlo, optimizing the robust parameter and projecting the capital forward) and a Confidence-based Allocation that takes sizing to the whole portfolio.
Position Size Calculator
A quick, standalone utility (it needs no data loaded): you tell it your account, the risk you accept and the trade, and it gives you back how many contracts or shares to buy. Live calculation, for stocks and futures.
Capital and currency; risk as a % of the account (with 0.25 / 0.5 / 1 / 2% quick buttons) or as a fixed amount, plus a "max risk per trade" cap.
A contract catalog (ES/MES, NQ/MNQ, YM, RTY, FDAX, FESX, CL, GC, ZB, 6E… with point value, tick and margin) or Custom. On stocks, a leverage field (1 = cash, 2-4 = margin).
Long/Short, entry price and a stop by Price, Points, % or ATR×. An optional target that works out the R-multiple.
Round-trip commission + slippage, which are taken off the real risk per unit.
Sizing Lab
It takes the real sequence of trades from a strategy in your Portfolio (re-testing it) or from an imported CSV, and simulates several sizing methods on it at once so you can compare them. That way you see, on the very same trades, what would have happened with each way of risking.
Fixed size, Fixed % risk (fixed fractional), Fixed $ risk, Fixed Ratio (Ryan Jones), Kelly (full, ½, ¼), Stop-based (using each trade's real stop) and % Volatility (sizing by the ATR at each entry). Each with its parameters live.
A table with final capital, return, CAGR, Max DD, MAR (return ÷ risk) and max contracts. It highlights the best MAR — which is almost never the one that makes the most.
Overlaid capital curves (log scale optional) and an underwater chart: how deep and for how long each method falls from its peak.
It reorders or resamples the trades thousands of times and measures the risk of ruin, P(< starting capital) and the percentiles of drawdown and final capital. A fixed seed: reproducible results and paired comparison.
Sizing Optimizer · the robust plateau
It sweeps a method's parameter (the % risk, the Kelly fraction, the delta…) and looks for the best setting — but not the peak, which is almost always a coincidence of the past, rather the center of a wide, stable zone: if the market shifts a little, you do not fall off a cliff there.
It picks the setting whose worst neighbor is still good. The gold dot is the past's optimum; the green band is the recommendation. They are usually different.
Every setting goes through a Monte Carlo; the ones that exceed your risk-of-ruin threshold are painted red and ruled out.
It optimizes on the first half and checks on the second: verdict ROBUST / DOUBTFUL / FRAGILE.
Capital Simulator · projection and minimum safe capital
It looks forwards: it resamples your real trades into the future (bootstrap) and draws the fan of where your account could end up. It does not predict — it shows what is plausible if the edge holds.
A fan chart with the p5–p95 and p25–p75 percentiles and the median, over the horizon you choose (in trades ≈ years). Cards for median final capital, bad scenario (p5) and good one (p95), P(double), P(ruin) and median CAGR.
Independent of the method: it projects at 1 contract, measures the $ drawdown of the worst percentile and computes capital = drawdown ÷ tolerance%. It answers: "how much do I need so that a bad run is a scare and not the end?". It scales to N contracts.
Confidence-based Allocation
The suite's differentiator: instead of risking the same on everything, it splits your risk budget among a portfolio's strategies according to their confidence — more to the robust ones, less to the doubtful ones. It is sizing applied to the whole portfolio.
One grade per strategy, estimated from its trades alone: sample (number of trades), regularity (positive months), consistency (both halves winning by similar amounts), risk (Ret/DD) and edge (Profit Factor). With an A/B/C/D letter. It is a quick proxy, not the full Survival Test.
The risk budget (the heat, a % of capital) is split in proportion to the confidence, and optionally penalizing correlation (Pearson on monthly P&L): the ones that move together weigh less. Those that do not reach the minimum confidence are excluded.
A table with confidence, Ret/DD, correlation, weight, risk % / $ and contracts per strategy; allocation bars; and a reading in plain language. It warns about the ones that do not even fit at 1 contract.
The method that would have made the most is usually the one that leverages hardest — and the one that takes you out of the game soonest in a bad run. That is why the Lab does not reward final capital, but the MAR and the risk of ruin: pick the size you would hold on to without bailing out through the worst stretch. A risk of ruin under 5% is the bare minimum; under 1%, comfortable. And if the strategy's Kelly is ≤ 0, no sizing fixes it: the edge first, the size afterwards.
AniQuant can be tried free for 30 days, with every module and no card.