Glossary

The betting glossary — plain English, no mystique

Betting jargon does two jobs: it compresses real ideas, and it keeps newcomers feeling like outsiders. This page does the first job and undoes the second. Every term here is one the site actually uses, defined the way we'd want it explained to us — and where a full guide exists, the definition links to it.

Accumulator · Base rate · Bet builder · Brier score · Calibration · Closing line · Closing line value · Confidence · Correlation · De-vig · Edge · Expected value · Expected goals · Favourite–longshot bias · Gubbing · Hit rate · Implied probability · Kelly criterion · Leg · Line · Overround · Per-90 rate · Price · Shrinkage · Stake factoring · True probability · Variance

Accumulator (acca)

Several bets multiplied into one: every leg must land or the whole thing loses. The odds multiply — and so does the bookmaker's margin, which is the part the adverts leave out. Bet builder value →

Base rate (baseline)

How often something happens overall, before you know anything specific — the share of all players booked, say. It's the naive guess a model must beat to be worth anything, which is why our track record grades every market against it. Track record →

Bet builder

A multi built from legs within a single match. Because same-match events move together, the bookmaker prices the combination, not just the parts — see correlation. Bet builder value →

Brier score

The average squared gap between what a model predicted and what happened — lower is better. On its own it flatters cautious models, so we always report it against the baseline's score; the difference is the skill. Track record →

Calibration

Whether your probabilities mean what they say: do the events you call 30% happen about 30% of the time? A calibrated model can still be beaten by the market, but an uncalibrated one is lying to you before you've even seen a price. How the model works →

Closing line

The final price before kick-off — the market's most informed opinion, with every late pound and team-sheet leak priced in. It's the benchmark serious bettors measure themselves against.

Closing line value (CLV)

The difference between the price you took and where the market closed. Beat the close consistently and you're genuinely finding value; profit alone can't tell you that for hundreds of bets. The scoreboard that matters. Closing line value guide →

Confidence (grade)

How much evidence sits behind a probability — sample size, lineup certainty, referee history. Edge says the price looks wrong; confidence says how sure we are of the number that says so. A big edge on a thin sample is a watchlist item, not a bigger bet. How to read the tool →

Correlation (same-match legs)

When legs rise and fall together — the game state that delivers a striker's shots also delivers the over. Multiply correlated legs as if they were independent and you'll overstate the combined chance; bookmakers reprice for it, and so should you. Bet builder value →

De-vig

Stripping the bookmaker's margin out of a price to recover the market's honest probability. Doing it naively flatters longshots; doing it properly (we use Shin's method) is the difference between finding edge and manufacturing it. De-vig explained →

Edge

Your probability × the decimal odds − 1: the expected profit per pound staked, if your number is right. A +5% edge means about 5p per £1 over many repeats. The "if" is the entire game. Expected value betting →

Expected value (+EV)

What a bet averages in the long run: win amount × win chance, minus loss amount × loss chance. A +EV bet loses plenty of individual nights and still makes money over enough of them — which is harder to live with than it sounds. Expected value betting →

Expected goals (xG)

The quality of chances, counted — each shot valued by how often that kind of chance scores. Steadier than goals themselves, because finishing luck comes and goes; it's what our goals model is fit on. Modelling team goals →

Favourite–longshot bias

Bookmaker margin isn't spread evenly: longshots carry more of it than favourites. Player bets are mostly longshots, so pretending margin is uniform quietly flatters exactly the prices you're most tempted by. De-vig explained →

Gubbing

Punter slang for the quiet removal of promotions from your account — free bets, boosts, money-back offers — usually the first rung on the restriction ladder. There's rarely a letter; the offers just stop. Why bookmakers limit winners →

Hit rate

The share of a player's games in which a line would have landed. The honesty column: a per-90 rate can flatter someone who piles stats into a few wild games, but a hit rate tells you how often the bet actually wins. Season outlook →

Implied probability

The probability a price asserts: 1 divided by the decimal odds. Odds of 2.50 imply 40% — but that includes the bookmaker's margin, so the market's true opinion is a touch lower. De-vig to find it.

Kelly criterion (fractional Kelly)

The stake size that maximises long-run growth if your probabilities are exactly right — and a wrecking ball when they're not. We surface a quarter of it: most of the growth, a fraction of the swings, and respect for the fact that every edge is an estimate. Bankroll management →

Leg

One selection inside a multi or bet builder. Each leg carries its own probability, its own price and its own slice of margin — which is why value has to be found leg by leg, not declared over the total.

Line

The threshold a bet clears: "1+ shots on target" asks for at least one. Bookmakers write it as over 0.5 — same bet. Each line is priced separately, and a player can be value at 1+ and a bad price at 2+ on the same afternoon.

Overround (vig, margin)

The bookmaker's built-in cost: add up a market's implied probabilities and they exceed 100%, and the excess is what the book keeps. It's why a fair-looking price usually isn't, and why de-vigging comes before any judgement about value. De-vig explained →

Per-90 rate

A stat normalised to 90 minutes rather than counted per appearance. It's the denominator that tells the truth: a player with heavy substitute minutes looks worse per game and honest per 90. Everything we model starts here. Modelling shots on target →

Price (decimal odds)

The odds, quoted decimally: your total return per pound staked, stake included. 2.50 returns £2.50 on a winning £1 bet. Every price is also a probability wearing a disguise — see implied probability.

Shrinkage

Blending a thin sample toward a sensible prior — a striker with 200 minutes gets priced mostly as "a striker" until his own record earns its say. It's what a careful human would do with small samples, done consistently. Modelling shots on target →

Stake factoring

A restriction where the bookmaker caps your stakes at a fraction of what you request — ask for £50, get offered £3.27. Usually unannounced; you discover it at the moment you try to bet. Why bookmakers limit winners →

True probability

The model's estimate of the real chance of an outcome, built from data rather than read off a price. The whole tool is the comparison between this number and the market's — when ours is meaningfully higher, the leg is flagged as value. How the model works →

Variance

The gap between what should happen on average and what happens tonight. It's why good bets lose, bad bets win, and short-run results tell you almost nothing — and it's the reason CLV, not profit, is the fast honest scoreboard. Expected value betting →