Methodology: How Is Odds Analysis Done?

Last updated: October 7, 2026

This page explains how the analysis tools on this site work (Opening and Closing Odds Analysis, Automatic Analysis, Odds DNA, Filters, Odds Radar, corner/card/basketball analyses) and what they cannot do. The goal is to make the origin of every number visible.

1. Short definition

Iddaa odds analysis means comparing a match's odds with the real results of past matches played at the same odds. The question is: "How did matches that opened at these odds end in the past?" The answer is a count, not a prediction.

2. Data source

Odds: the opening and closing odds of the iddaa bulletin are imported. The opening odds are the odds at the moment the match first enters the bulletin; the closing odds are the last odds before kick-off.
Results: match scores come from the same source; corner, card and basketball statistics come from a separate match statistics provider.
No manual data: no odds, score or statistic is entered by hand or filled in by guesswork. Missing data is not counted as zero; it is left out of the calculation.
The current coverage of the archive (number of matches, leagues and odds) is shown live on the Data Centre page.

3. How is an analysis calculated?

Odds band: The selected odds (e.g. 1.85 for the home win) are searched not as a single number but as a small range. A very narrow range shrinks the sample; a very wide range mixes different kinds of matches.
Matching matches: Settled matches in the archive that opened within that range are found.
Counting: For each market, the number of matches in which it happened is counted and converted to a percentage.
Sample: Every percentage is given together with the number of matches it was calculated from. If the sample is small, the percentage is either not shown at all or shown faded with a "low sample" warning.
Base rate: The overall hit frequency of a market is shown too; the result of an odds band only means something when compared with this base rate.

Automatic Analysis runs the same process for every market of the selected match at once. Odds DNA searches for past matches whose odds "fingerprint" across several markets resembles the match, instead of a single price. Filters summarise the results of the most recent completed days by odds band. Each tool states on its own page which time window it uses.

4. Closing odds and odds movement

A change in the odds between opening and closing is the market pricing in new information about the match (line-ups, injuries, money flow). On this site the odds movement is shown as measured data; it is not presented as a rule such as "dropping odds win".

5. The bookmaker margin inside the odds

Odds include the bookmaker's margin. In our archive, the implied probabilities of the match result (1X2) odds add up to about 116.5% (opening) and 121.9% (closing) on average (measured in September 2026). In other words, the "fair" probability of an outcome is lower than the probability read from its odds. Because of this margin, even an outcome that happened often in the past may not be profitable in the long run.

6. Limits of models and predictions

Past frequency is not the future probability itself. Squads, motivation, weather, referees and luck are not visible in past odds.
The odds already contain the market's information. We do not claim that our own measurements found a signal that gives a lasting edge beyond the odds.
In tools that use artificial intelligence or a statistical model, the model output is a probability estimate, not a verified result.
No prediction is a guarantee. The future is unknown.

7. Predictions and backtests

Every published prediction is marked automatically with the real score when the match ends. Won and lost predictions are public per system on the Performance page; pending records are not included in the success rate.
A backtest applies a rule to past matches and measures the outcome. A reliable backtest must also be measured on a period separate from the one the rule was built on; otherwise the result may be "fitted" to the past. The period and sample used in every published backtest are stated on the relevant page. For Spor Toto, the method and results are on the Spor Toto Backtest page.
Weak or negative backtest results are not presented as "high success".

8. Errors and corrections

If you think a number is wrong, report it via the Contact page. The correction process is described on the Editorial and Corrections Policy page.