Methodology

Principles

Three rules govern everything on this site. First, show the mathematics rather than assert a conclusion. Second, state assumptions and limits alongside every result. Third, never present an estimate, a model output or a point-in-time observation as though it were a certainty.

Where we cannot know something, we say we cannot know it. That is more useful to a reader than false precision.

How the calculators work

Every calculator on this site is a deterministic function of the numbers you type in. The same inputs always produce the same outputs. There is no hidden data source, no live odds feed, no bookmaker connection, and no forecast of any real event.

Calculations run entirely in your browser. The values you enter are not transmitted to us, and we do not store them, unless you are explicitly saving them to your own account in the betting journal.

The formulas are the standard, publicly documented ones used across the field. For example: decimal, fractional, American and implied-probability conversions; expected value as the probability-weighted return minus the stake; bookmaker margin as the sum of implied probabilities minus one; fair (no-vig) prices obtained by normalising implied probabilities back to one; arbitrage stake distribution by proportional allocation across inverse prices; Kelly stake as edge divided by odds; and parlay pricing as the product of the individual decimal odds.

Assumptions and limits of the calculators

A calculator answers a precise mathematical question. It does not answer the question “will this bet win”, and it cannot. These limits apply throughout.

• Any probability, edge or confidence value used in a calculation is the one you supplied, or an illustrative default. It is an assumption, not a measurement, and the output is only as sound as that assumption.
• Multiples and parlay tools assume the selections are independent unless the tool explicitly handles correlation. Real sporting outcomes are frequently correlated, which makes true parlay value lower than naive multiplication suggests.
• Arbitrage, hedging and exchange tools ignore stake limits, account restrictions, price movement between placements, partial matching, and exchange commission unless the tool provides a field for it.
• Bankroll, growth and drawdown tools are simulations under stated assumptions. They illustrate the shape of variance; they do not forecast your results.
• Results may be rounded for display. No tool accounts for tax, currency conversion, payment fees, or the time value of money.
• Expected value being positive does not mean a bet will win, and a long losing run is fully consistent with a positive-expectation approach.

Testing

The calculator engine is covered by an automated test suite that runs on every change to the codebase, checking the formulas against known worked examples. This protects against arithmetic regressions.

It does not, and cannot, validate the assumptions a user brings to a calculation. A correct formula applied to a wrong probability produces a confidently wrong answer.

How operator ratings are formed

Where we rate or rank betting operators, the assessment is editorial judgement informed by observable criteria, weighted according to how much they affect a bettor's long-run result. In broad order of importance: pricing and margin across main markets; payout reliability and speed; market depth and limits; licensing and regulatory standing; and usability.

These ratings are opinions, not certified measurements. Margins are sampled at a point in time from public prices and vary by sport, market and moment. Payout times vary by method, jurisdiction and individual account. Nothing about a rating is a guarantee of how any operator will treat you, and commercial relationships do not influence position — see our Affiliate disclosure.

Predictions and model output

Where this site publishes predictions, tips or model output, they are analytical illustrations, not advice and not insider knowledge. They are produced from the information available at the time of writing and from generic modelling assumptions.

Our systems have no live fixtures feed and no privileged data. Any stated confidence figure or expected-value percentage is a model artefact under stated assumptions, not a measured probability of a real-world event. No accuracy or profitability is claimed, promised or implied, and past or simulated performance says nothing about future results.

Content of this kind is generated or assisted by automation, as described in our Editorial policy, and should be treated accordingly.

Corrections

If you find a formula, figure or description on this site that you believe is wrong, report it to [email protected], naming the page and the specific value. Mathematical corrections are treated as urgent.