Research service
We test your strategy against real market data.
You have a football trading strategy. You have a sense that it works. What you probably do not have is a rigorous answer to whether it held up across a few hundred markets you had not already seen.
We run it against Betfair Exchange tick data and write up what happened — the method in full, the sample size, and the result whether or not it is the one you were hoping for. The report is written to be published, because a report you would be embarrassed to publish is not one worth commissioning.
How the testing works
The split is chronological, never random. Earlier markets are used to form the view; later markets are left untouched until the end and are the only ones that count as evidence. Splitting at random leaks the future into the training half, and it is the most common reason a backtest flatters a strategy that does not work.
Fills are modelled pessimistically. An order fills only at the price actually shown on the tick, never better, and a limit the market never reached does not fill at all. Optimistic fill modelling is the easiest way to make a losing strategy look profitable.
Commission is charged on net market winnings, which is how the exchange actually bills. Charging it per bet overstates costs badly on hedged books and would understate every strategy that greens up.
Every headline figure carries a confidence interval, bootstrapped over whole markets rather than individual bets — bets inside one market are correlated, and treating them as independent makes an interval look far tighter than it is.
Every report runs a fixed-rule control that ignores all signal. If your strategy cannot beat it, the result is more likely exposure to an underlying market bias than anything the strategy is doing, and the report says so in those words.
What you get
- A written report: method, sample, period, competitions covered, result.
- Train and test figures side by side, so you can see whether it held.
- The control comparison.
- A plain statement of what the test does not show.
- Yours to publish, with or without our name on it.
What we will not do
We do not sell selections, tips or predictions, and we do not tell you what to bet. We will not describe any strategy as profitable, guaranteed or risk-free — no test of past markets can support that about the future, and we will say so in the report rather than leave it implied.
If the result is that your strategy showed no edge over the sample, that is what the report will say. You are commissioning a measurement, not a verdict you have chosen in advance.
Scope and pricing
Priced per strategy, because a single pre-off rule and an in-play system with several exit conditions are not the same piece of work. Tell us the rules in plain English and the competitions and period you want covered, and we will come back with a scope, a price and a delivery date before anything starts.
Sample sizes depend on available market data for the competitions you name. Where the sample is too small to support a conclusion, the report states that plainly instead of presenting a number that invites one.
Nothing on this page is betting advice and no outcome is predicted or promised. Past market behaviour is not a forecast. 18+ · BeGambleAware.org