How Our Football Predictions Work

Every number on GoalStatsHub is a probability: our estimate of how likely an outcome is, from 0% to 100%. We don't publish bookmaker odds and we don't sell tips. This guide covers where the numbers come from, what each market means and how to use the numbers sensibly.

Where the numbers come from

For each fixture, our model looks at the recent record of both teams. Here is what it looks at:

• Goals scored and conceded, at home and away separately. Many teams play very differently in each setting.
• Recent form: results and scorelines from the latest matches.
• Head-to-head history: previous meetings between the same two clubs.
• Match statistics such as corners, cards and half-time scores. These feed the secondary markets.
• Referee statistics, when available. They help with the cards and booking points markets.

From this data the model estimates how many goals each side is likely to score. It then turns those estimates into a grid of every realistic final score, such as 0-0, 1-0 and 2-1. The goal markets on the site are built from that grid. For example:

• The home win probability is the sum of all scores where the home team is ahead.
• The "over 2.5 goals" probability is the sum of all scores with three or more goals.

Because these markets come from one grid, they stay consistent with each other. A match with a high over 2.5 figure will usually also show a high both-teams-to-score figure.

Confidence

A prediction is only as good as the data behind it. Next to each match on the Tips page, up to three dots show how much data the model had:

• Three dots: both teams have a long, recent record in the competition.
• One dot: the model had only a little data, for example early in a season, after promotion or in a cup tie between teams that rarely meet.
• No dots: the percentages are only weakly grounded. Treat them as a rough guide.

External estimates

Sometimes our own model has nothing to work with, for example in a lower division without enough history. In that case we may show an estimate from an outside data provider, marked ext. We show it so that the fixture isn't left blank, but it is not our work. It is never counted in our results.

Settled results

We don't delete predictions after a match. When the final score comes in, each tip and each likely score is marked:

• green if it landed;
• struck through if it didn't.

The misses stay on the page next to the hits, so you can judge the model on its whole record, not a curated selection.

The markets explained

1X2: match result

The most common football market has three possible outcomes after 90 minutes plus stoppage time:

• 1: home win;
• X: draw;
• 2: away win.

The three probabilities always add up to 100%. The coloured bar on each match shows the same split: green for home, grey for the draw, orange for away. Extra time and penalties don't count. A cup tie level after 90 minutes is a draw in this market.

Double chance

Double chance combines two of the three results, so you cover two outcomes:

• 1X: home win or draw;
• X2: draw or away win;
• 12: either team wins, so anything except a draw.

Its probability is the two parts added together. With home 45%, draw 27% and away 28%, 1X comes out at 72%.

BTS: both teams to score

BTS (often written BTTS) is the chance that each side scores at least once. The final result doesn't matter: 1-1 and 3-2 both count as "yes", and 3-0 counts as "no". This market rewards attacking teams with leaky defences. Two sides that both keep clean sheets often will produce a low BTS figure.

Over / under goals

This market is about total goals in the match, whoever scores them. Lines end in .5, so there is always a clear answer:

The "under" probability is simply 100% minus the "over" figure. If +2.5 shows 61%, under 2.5 is 39%.

Exact score

Each match lists its most likely final scores with their probabilities, for example "1-0 14%". Even the top score rarely goes much above 15%, because football has many possible scorelines and none of them is likely on its own. Use these figures to see the shape of a match (tight and low-scoring, or open), not as a single answer.

On the Matches page and on each match page, the full score matrix shows every scoreline in a grid, with home goals on one axis and away goals on the other. Darker cells are more likely.

Half-time markets

The same ideas apply to the first 45 minutes: half-time result and first-half goals. First halves usually produce fewer goals than second halves, so half-time draws are much more common than full-time draws.

Corners and cards

Corner and card lines work like goal lines. "Over 9.5 corners" means ten or more corners in total. Booking points are a weighted count of discipline: a yellow card is usually worth 10 points and a red card 25, so one sending-off moves the total a lot. The referee's record has a strong effect on card numbers, which is why we show referee statistics when they are available.

Result and goals combined

Some markets combine a result with a goals condition, for example "home win and over 1.5 goals". Both parts must come true, so the probability is always lower than either part on its own.

Probabilities versus bookmaker odds

Bookmakers show prices, not probabilities. You can convert decimal odds into the probability the price implies:

Implied probability = 1 ÷ decimal odds. Odds of 2.00 imply 50%; odds of 1.50 imply about 67%; odds of 4.00 imply 25%.

Bookmakers build a margin into their prices. As a result, the implied probabilities of all outcomes in a market add up to more than 100%. Our figures have no margin, so they always add up to exactly 100%.

Comparing the two can show where an estimate differs from the market. It is not a guarantee of anything.

Reading the numbers responsibly

• A 70% outcome fails three times in ten. A probability describes many matches, not one. Any single high-probability pick can and often will lose.
• Models can't see everything. Late injuries, rotation before a big game, weather and motivation at the end of a season are only partly reflected in historical data. Check team news before a match.
• Small samples mislead. Early in a season the figures rest on very few games. The confidence dots show when that is the case.
• Nothing here is financial advice. GoalStatsHub is an information service. If you bet, only stake what you can afford to lose, and read our responsible gambling page.

Where to go next

• Guides: longer articles with worked examples, from how the model builds a scoreline grid to comparing probabilities with bookmaker odds.
• Matches: every fixture for today and two days either side, with full statistics, form, head-to-head records and all markets.
• Tips: a short board with the key probabilities and suggested selections for today and the next two days.
• About us: who runs GoalStatsHub and why.