Betting exchange markets
A research engine for
betting exchange markets
142,035,768 recorded price movements, and growing. You build the question yourself and the answer arrives with the working: how many bets it is built on, the t-statistic and p-value that say whether it could be luck, and how it fared on the seasons it was never allowed to see.
The stats board
Every league and every team — over and under, both teams to score, goal windows. Free, open to everyone.
A worked example
One question asked of the engine end to end, with the holdout and the figures it produced.
The calculators
Back and lay, trading in and out, and whether backing the over beats laying the under.
What it takes to be sure
How many bets it takes before a result means anything, and why most of what looks like an edge is not one.
What we are actually looking at
An exchange is a market. People post prices, other people take them, and the number moves because money moved it — the same way a number moves on any other exchange. The football is the commodity being priced. A team’s form matters here only in so far as it shifts what someone is willing to pay.
So what we search for is movement: where a price drifts and where it shortens, how far it travels between the market opening and the whistle, which prices finish near where they started and which never get close. Each of those 142,035,768 readings is a minute stamped with somebody changing their mind with money on it.
Here is the plainest example of it — what a price does, on average, in the hour before kick-off, split by where that price started.
What the market does in the hour before kick-off
every match-odds runner we hold · averaged · by starting price
130,750 runners, each indexed to its own price an hour before kick-off, then averaged minute by minute. Short favourites come in. The longer the price, the harder it drifts.
Measured to a T-Statistic of 21 across every runner we hold — and still not a strategy. That is what the tools are for.
Tested on data it has never seen
Seasons are held back from the search. The figure that survives meeting them is the only one quoted.
Judged on the closing price
A run of winners proves nothing. Beating the price the market settled at is the thing that does.
Costs counted before the claim
Every figure on this site is what would have reached your account, not what the market showed.
Proved wrong as often as proved right
Most of what gets tested does not work, and finding that out here costs an afternoon rather than a season.
What is behind each door
The lists are what is actually there — not a summary of it.
An account, not a trial. No card.
Open an account- My Desk — every game on the day's card, with the live prices
- League and team stats — over and under, both teams to score, W/D/L and goal windows, every competition
- Your own bet log and Betfair statement uploads — your record, kept on your account
- The calculators, the worked example and the hurdle
Details at launch
Sign-ups currently closed- The price database — 14 markets, every league, minute by minute since July 2019
- The strategy simulator — build a question, any league, season, team or side
- Price conditions with their own clock, joined by and / or
- Form rules built only from games played before each fixture
- Hold seasons back, then see how far the figure falls when it meets them
- Sample size, t-statistic and p-value on every answer
- Judged against the closing price, not against our profit
- Bank and staking models — flat, percentage, and the sizing sweep
- My Desk — today's card with our own recorded prices against it
- ELO, Poisson and the confluence board
- Save your strategies, and have them watched as new games arrive
- Pre-match entry, held to the result
Fifty seats only **
Sign-ups currently closed- Everything in Member
- In-play entry — open at any minute, on any price condition
- Close on a time, a price, a profit, or whichever price hits first
- Trade out rather than only hold to the result
- Our own built-in filters — the ones the engine runs on
- What our own engine found, as it finds it, with the working alongside
Submit a strategy and get it read back: what is actually driving the result, where the sample is thin, and the pitfalls it is walking into.
A model built on the same seasons of prices, run against the card, and judged in public on the closing price like everything else here.
** Elite is capped at fifty because in-play markets are thinner and far more dynamic than the steady pre-match ones. Money moves fast in them, and fastest of all at the moments worth trading — so a crowd taking the same price in the same minute moves it, and destroys the thing they paid for. Member has no cap, because two people building their own strategies are not competing with each other.
What a typical search hands back
Every one of these is a column on the answer, not a feature we are planning.
What happened
- Bets
- How many games the rule actually bet on.
- Win Rate
- How often it won, and the average price you got.
- P&L Gross and Net
- Profit before and after commission, and what £100 a bet would have made.
- ROI Net
- What each pound staked returned.
- Per Bet, and On Risk
- Return on the stake and on the money at risk. On a lay they are very different.
- Drawdown
- The worst losing run it put you through, from a high point to the bottom.
Whether it is real
- T-Statistic
- How far the result sits from break-even, allowing for how much it swings.
- P-Value
- The chance of a result this good with no edge at all.
- Luck Would Give You
- The return pure chance could produce on this many bets. Beat it, or it is noise.
- Expected
- What the closing prices said these bets should return.
- Bets Needed
- How many bets it would take to prove the edge is real.
- Less the Best
- The same figure with its best season taken out, so one good year cannot carry it.
Whether it is an edge
- CLV
- How your price compared with the closing price, as mean, median, mode and last.
- Beat the Close %
- How often you beat the closing price, not just by how much.
- Edge %
- Your edge over the market, and over the league as a whole.
- Held Back
- The same figures on seasons it was never shown.
- Spread of Results
- Split by league, season, team, side and price range.
- Never Triggered
- The games it looked at and turned down.
One we ran
Lay the away team whenever its price is 10.00 or longer, one hour before kick-off. Hold it to the final whistle.
43,588 games searched · 2,499 qualified · 8 seasons · 95 milliseconds
Ninety-four winners in every hundred and nine per cent up after commission — and it is still not proof. A T-Statistic of 1.22 sits inside the range of ordinary noise. We would not run it.
What asking it looks like
Three questions on the left — what market, when you open, when you close. The answer on the right, in the time it takes to let go of the mouse.
| Competition | Bets | Win | Price | Net | T |
|---|---|---|---|---|---|
| Portugal · Liga NOS | 300 | 96.7% | 19.35 | +42.73% | 2.42 |
| Netherlands · Eredivisie | 286 | 96.2% | 20.05 | +37.24% | 1.91 |
| Italy · Serie A | 261 | 92.0% | 14.02 | −6.63% | −0.30 |
| England · Premier League | 247 | 91.5% | 17.23 | −24.30% | −0.91 |
| Spain · La Liga | 243 | 93.0% | 15.08 | −6.80% | −0.26 |
| Scotland · Premiership | 208 | 96.6% | 22.57 | +31.96% | 1.22 |
| Germany · Bundesliga | 192 | 93.8% | 18.06 | −16.46% | −0.46 |
| France · Ligue 1 | 150 | 92.0% | 16.75 | −13.84% | −0.44 |
Read the Net column. The same rule makes 42% in Portugal and loses 24% in the Premier League. A strategy with a real edge does not do that. A coin does exactly that — and this is how you catch one.
That is the same search the card above reports, split by competition. Nothing about it was set up to look good — it is what the engine returned, and it is the clearest picture of a strategy falling apart you will see on this page.
Why an exchange, not a bookmaker
Every price here came from an exchange, and that is a decision rather than a convenience.
A bookmaker closes you down
Win consistently with a traditional bookmaker and your stakes get cut to pennies, or the account goes. It is not a risk — it is their business model working as intended.
An exchange cannot
You are betting against other people, not against the house. Nobody there has an interest in stopping you, and nobody can.
The prices are real, not offered
Every price we hold is one two people actually met at. A bookmaker price is an offer that may never have been taken.
You can be on either side
Lay as readily as back, and close a position before the result. None of that exists with a bookmaker.
Most of what we test does not work
That is not an admission. It is the service.
An idea that looks good on the seasons it was found on and falls apart on the two it was never shown has cost you an afternoon. The same idea, backed for six months before you find out, costs a great deal more than that. Knowing which is which before you stake anything is the whole point of the place.
Everybody sells you the systems that worked. Finding out here that one does not saves you money too, and nobody else will tell you.
What this is
Statometrics is a data and research service. We do not accept wagers, hold stakes, place bets or operate any market. Nothing published here is advice or a recommendation.