A mathematical football prediction is a probability worked out from numbers by a fixed method. This page shows that method step by step, with the arithmetic, and says plainly what the numbers can and cannot tell you.
A mathematical football prediction is a statement about a match written as a probability. It says, for example, that the home team has a 48% chance to win. It comes from a calculation that anyone can repeat, not from an opinion about the teams.
On Zort the calculation has three steps. First, bookmaker odds are turned into probabilities. Second, the bookmaker margin is removed. Third, a goals model spreads those probabilities across every possible scoreline. The daily board at /predictions/ shows the result for every match priced that day.
Decimal odds hide a probability. To find it, divide 1 by the odds. Odds of 2.00 give 1 ÷ 2.00 = 0.50, which is 50%. Odds of 4.00 give 1 ÷ 4.00 = 0.25, which is 25%. This number is called the implied probability. Short odds mean a high probability. Long odds mean a low one.
Take a match priced at 2.00 for the home win, 3.50 for the draw and 3.80 for the away win.
Only one of the three results can happen, so the true chances must add up to 100%. The extra 4.89% is the bookmaker margin, the bookmaker's built-in share. It makes every result look a little more likely than it is.
The simplest way to remove it is to divide each figure by the total:
Now the three add up to 100%.
Zort uses a more careful way, called Shin's method. The simple method takes the margin off every result in the same proportion. In real prices more of the margin sits on the outsider, so the simple method overstates long-shot chances by roughly 1.2 percentage points. Shin's method corrects for this. For the same three prices it gives 48.19% for the home win, 27.02% for the draw and 24.79% for the away win. The favourite moves up a little and the outsider moves down. No team in this example is a real long shot, so the two methods stay close.
Zort does this with live odds from around 27 UK and EU bookmakers.
The probabilities without margin say who is more likely to win. They do not yet say what the score might be. For that, Zort converts the match result and goals probabilities into an expected goals figure for each team, then applies a Poisson model.
A Poisson model answers one question. If a team is expected to score a certain number of goals on average, how likely is it to score exactly 0, 1, 2 or 3? For a team expected to score 1.5 goals:
The rest is the chance of four goals or more.
In a plain Poisson model you do the same for the other team and multiply the two chances. That gives a probability for every scoreline: 0-0, 1-0, 1-1, 2-1 and so on. Add up the scorelines where the home team scores more and you have the home win chance. Add up the ones with three or more goals and you have the chance of over 2.5 goals.
Zort uses the Dixon-Coles version of the model. It adds a low-score correction, because a plain Poisson model underestimates low scorelines such as 0-0 and 1-1.
Each market has its own page: match result (1X2), over/under 2.5 goals, correct score and both teams to score.
A probability tells you how often something would happen if the same match were played many times. It does not tell you what will happen this time.
A 70% chance is not a promise. 100% − 70% = 30%, so the other result arrives about 3 times in 10. Two separate 70% chances both come true only 0.70 × 0.70 = 0.49 of the time. That is 49%, less than half.
The numbers start from bookmaker odds, so they carry what those odds already reflect and nothing more. They cannot know what will happen on the pitch: a red card, a deflection, a missed penalty.
Zort publishes no win rate, return or hit rate for these predictions and makes no claim that they win. Every board since 10 August 2026 is kept in the archive, so anyone can check them against the results.
Each row on the board is one match. The columns are:
The board is free, needs no signup and is published every day at 10:00 UTC. It typically covers 30 to 150 fixtures, depending on the day of the week. Tomorrow's matches are at /predictions/tomorrow/.
It is a football prediction written as a probability and calculated by a fixed method. On Zort, bookmaker odds are turned into probabilities, the bookmaker margin is removed with Shin's method, and a Poisson/Dixon-Coles model gives a chance to each scoreline.
Today's board is at zort-app.com/predictions/. It is published every day at 10:00 UTC, it is free and it needs no signup.
Yes. Tomorrow's matches are at zort-app.com/predictions/tomorrow/, calculated with the same method as today's board.
No source is right every time, so in that sense there is no best. What you can check is the method and the record. Zort explains its method on this page, publishes no win rate and keeps every board since 10 August 2026 in its archive, so anyone can compare the predictions with the results.
No. Sure wins do not exist. A 70% chance still fails about 3 times in 10, and two 70% chances both come true only 49% of the time.
It gives the most likely scoreline, not the actual one. A match can end in dozens of different scorelines, and across the market generally even the most likely one typically lands only nine to thirteen times in a hundred matches.
Zort is a bet tracker, not a betting company. It is an app for tracking the bets you have already placed, so you can see your real results in one place: profit and loss, return and win rate. Zort sells no picks and takes no bets. Read more about the bet tracker app.
Download Zort FreeMore: Free Football Predictions · Tomorrow's board · Match Result (1X2) · Over/Under 2.5 Goals · Correct Score Predictions · Both Teams To Score Tips · Today's board