Methodology

Every calculator uses accepted financial mathematics. Nothing here is a proprietary formula, a weighting we devised, or a model with hidden parameters. Where a calculation has more than one accepted convention — position sizing, for instance, or how a break-even rate treats trading costs — the page states which convention it uses and why.

Each calculator page shows its formula in full, defines every symbol, and works through an example with substituted numbers so you can check the arithmetic by hand. If a page shows you a result you cannot reproduce from its own formula, that is a bug and we want to hear about it.

Calculation logic lives in pure functions, entirely separate from the interface. Each function takes numbers and returns numbers, which makes it directly testable and keeps display formatting from ever affecting a result.

Those functions are covered by an automated test suite that runs on every change. Tests cover ordinary cases, known worked examples, zero values, negative values, decimal inputs, boundary conditions such as a 0% or 100% win rate, and inputs designed to break the maths — a stop equal to the entry price, a divisor of zero, values large enough to overflow.

A calculator that prints “NaN” has failed twice: once in the arithmetic and once in telling you about it. Impossible inputs are rejected before they reach the maths, with a message naming the specific problem — for example that a stop-loss cannot equal the entry price, rather than a generic “invalid input”.

As a second line of defence, every number passes through a formatting layer that returns a placeholder for any value that is not finite. Both layers are covered by tests, so no combination of inputs can produce a nonsensical figure on screen.

Calculations are carried out at full floating-point precision and rounded only for display. Rounding is applied with a correction that removes binary representation artefacts, so a result of 0.30000000000000004 displays as 0.30 rather than being shown raw or silently truncated.

Where rounding affects what you can actually do — forex lot sizes, which brokers accept in 0.01 increments — the calculator rounds down and shows the resulting risk. Rounding up would quietly breach the risk limit you set.

Calculations are performed on numbers, independent of any currency symbol. The currency selector changes how results are displayed, not how they are computed, so switching between USD, USD and EUR never alters the underlying arithmetic.

No live exchange rates are fetched. Where a calculation needs to convert between currencies — forex pip values, margin in an account currency — you enter the rate yourself. That is deliberate: the rate your broker applies is the one that determines your actual result, and it will differ from a mid-market rate pulled from a third-party feed.

No calculator fetches prices, quotes or market data of any kind. Every figure comes from values you enter. This keeps results reproducible, keeps the site fast, and means a calculation you performed today can be repeated exactly tomorrow.

Every calculator states what it leaves out, on the page, in its own section. Common exclusions are spread, commission, slippage, overnight financing, platform fees, stamp duty and tax. These vary by broker, jurisdiction and account type, so building an assumption about them into the maths would make results less accurate rather than more.

Some calculators accept a cost figure directly — the break-even win rate calculator and the crypto fee calculator both do — because the effect of costs is the point of those tools.

One calculator, risk of ruin, reports the output of a Monte Carlo simulation rather than a closed-form formula. Standard risk-of-ruin equations assume even-money bets at a fixed stake, which does not describe fixed-fractional position sizing with an asymmetric payoff, so the system is simulated directly instead.

That page lists the model’s assumptions explicitly: independent trades, a constant win rate and payoff ratio, and a fixed risk percentage of the current balance. The simulation uses a fixed random seed, so identical inputs always produce an identical figure. It is the output of a model, not a forecast.

Each calculator carries a “last reviewed” date covering both its logic and its explanatory content. Reviews check that the formula still matches accepted practice, that the worked example reconciles, and that the stated limitations remain accurate.

If you find a result you believe is wrong, please tell us and include the inputs you used. Reported errors are prioritised above new calculators.