Match Lens: A Probability Calculator, Not a Fortune-Teller

Match Lens 3 is a small program that calculates probabilities in percent. You fill in everything yourself: who plays whom, the average goals for home and away, who gives the handicap, and odds if you have them. The program only calculates.

We are calculators, not fortune-tellers. Nobody knows what will happen on the pitch, so every result below can be right or wrong. Calculations run in your browser, and a calculation is saved only if you press the save button.

The program area can be scrolled. Entries you save stay in this browser and are not sent anywhere.

1. The match. Write the names of the home and away teams.
2. Average goals per match. Enter goals scored and goals conceded for the home team, then for the away team, from data you have. The home factor can stay at 1.00 (neutral).
3. Handicap and handicap odds (optional). Choose who gives the handicap and its size, for example 0.5 or 0.75, then enter the odds on both sides.
4. Win-draw-lose odds and the total goals line (optional). Fill these in to see the comparison with implied probabilities and the total goals result.

Unreasonable numbers are rejected with a message explaining why. Every number comes from you; the program fetches no match data from anywhere.

The program uses a Poisson model, a common mathematical way to count goals. The steps:

1. The home goal rate is the average of the goals the home team scores and the goals the away team concedes. The away goal rate is computed the same way.
2. From the two rates, the probability of each score is computed, for example 1–0, 2–1 or 0–0.
3. The score probabilities are added up into the probability of a home win, a draw and an away win, the probabilities of the number of goals, and the handicap and total goals results.

The program also shows a version with a low-score correction (Dixon–Coles style). The gap between the two versions shows how much the result depends on assumptions. The model is simple and has not been validated for any particular league or match.

The result is in percent. The headline sentence always states the scenario rated most likely and the probability of the others, for example “Home win (45%). That means 55% of the probability falls on other results.”

• The range. The number in brackets is the probability range if the goal rates are off by 15%. A wide range means little can be said for sure.
• The handicap. Handicap results are split into full win, half win, push, half loss and full loss, according to the line you entered.
• Total goals. Lines such as 2.5 or 2.75 are calculated including their half results.

The highest probability is not certainty. A result with a 30% probability still happens about three times in ten.

If you enter odds, the program calculates their implied probability (1 ÷ odds) and then corrects for the margin with proportional scaling, one simple method. That implied probability is shown next to the model’s probability, with the difference.

The difference is only a comparison of two estimates: one from a model that has not been validated, one from a price. A small error in either can flip its sign. The program does not recommend bets, does not flag matches, and does not calculate stake sizes. The basics of implied probability and margin are in the (Indonesian) Panduan Sportsbook, and the open expected-value calculation is in Value Bets and Expected Value.

So that you can see for yourself how often these calculations match reality, the program provides a log in your browser.

1. Calculate a match, then press “Save this calculation to my log”.
2. After the match, enter the final score.
3. The report card counts how often the scenario rated most likely actually happened, with a 95% confidence range, the average probability given to the result that happened, the Brier score, and a calibration table.

The report card is built from your own log, so nothing is hidden. The data stays in this browser. You can export it as a JSON file or delete it at any time. When there are few entries, the report card warns you, because percentages from a few matches are very noisy. The statistical reasons are covered in Sample Size and Uncertainty.

The program only sees the numbers you enter. It does not see:

• Conditions on the pitch. Line-ups, injuries, fatigue, weather and team motivation.
• Decisions on the pitch. The referee, red cards and unforeseen events.
• Matches whose results are fixed. Match-fixing by criminal networks has been proven in many countries and sports. A simple model like this program only uses the numbers you enter and has no way to tell a fixed result from ordinary competition. Integrity bodies use statistical and market analysis as a filter, but still need investigation and other evidence. Full reading: Match-Fixing Is Real.

So the program’s output is a statement about a model and the numbers you enter, not about reality.

Laws differ in every country. In Indonesia, gambling, including online, is prohibited by law. This program was written for calculating and learning, not as an invitation to bet.