Models, AI, and statistics

Active Model: how to interpret the estimates

Explore the model behind your probabilities, fair odds, EV, and historical results. Compare methods and see how a different model changes your view of a match.

Follow the active model: inputs Match data; method Active model; outputs Estimates. Illustrative example.
Illustrative example. Use the same model when comparing probability, fair odds and EV.

Direct answer

Active Model is the mathematical lens currently in use on Predora.si at that moment. It is what calculates the probability, a fair odds, the EV, the edge and the main analyzed outcome of the games in the selected market.

Definition in one sentence: Active Model is the currently selected model for calculating probability, fair odds, EV, edge, hit rate, and simulation within the market and league you are analyzing.

Quick summary

PointWhat it means
DefinitionIt is the model that is active and guiding Predora.si's calculations.
Where it appearsIn the Active Model card at the top of the grids and on the Models page.
What it calculatesProbability, fair odds, EV, edge, main analyzed outcome, and performance metrics.
Markets involved1x2, Double Chance, Over 2.5 goals, Both Teams to Score, and Over 9.5 corners.
Why it mattersSwitching models changes the reading of the games, the probabilities, the EV, and the highlighted scenarios.
Common mistakeUsing the Active Model without examining its method and history.
How Predora helpsShows the model in use, the Base, the Hit Rate, the Return, and the comparison with odds and factors.

Main concept: the Active Model is the app's lens

Think of the Active Model as the lens Predora.si uses to see the games. The game is the same and the bookmaker's odds may be the same, but the mathematical reading can change depending on the selected model.

In practice, the Active Model defines:

  • what probability the app estimates for the analyzed outcome;
  • what the fair odds would be for that probability;
  • whether the best available odds seem to have value;
  • what the EV of that scenario is;
  • which games stand out most in Upcoming Games with EV+, Upcoming Games with High Probability, and AI Radar;
  • which numbers appear in performance, such as base, hit rate, and simulation.
The Active Model is not just a name at the top of the screen. It is the engine guiding the current analysis.

Formulas used in the Active Model reading

The Active Model does not change the bookmakers' odds. What it changes is the estimated probability. Based on it, Predora.si calculates the main numbers in the analysis.

Fair odds = 1 ÷ estimated probabilityImplied probability = 1 ÷ decimal oddsEdge = model probability − implied probabilityEV = (estimated probability × odds) − 1

Estimated probability

It is the probability the Active Model gives for the analyzed outcome in the selected market.

Fair odds

It is the quote that would be consistent with the probability estimated by the model.

Implied probability

It is the probability the bookmaker's odds are pricing in at that moment.

EV and edge

They show whether the odds seem to pay above or below the estimated fair price.

Example in football

Imagine a Brasileirão game in the market 1x2, with the analyzed result Home team wins. The Active Model estimates a 53% chance, and the best available odds are 2.05.

Fair odds = 1 ÷ 0.53 Fair odds = 1.89 Implied probability = 1 ÷ 2.05 Implied probability = 48.8% Edge = 53% − 48.8% Edge = +4.2 pp EV = (0.53 × 2.05) − 1 EV = +8.65%

If the model thinks the fair odds would be 1.89, but the bookmaker is offering 2.05, the odds are paying above the estimated fair price. This may indicate an estimated difference.

The probability estimate and offered odds produce positive EV. Compare current match context and the model’s historical performance to understand the evidence behind the price.

What appears in the Active Model card

The Active Model card works as a control panel for the grid. The reading becomes stronger when you combine the model name, market, sample, hit rate, and simulation.

FieldHow to interpret
Model nameExample: Predora AI Pro, Predora Market Median, or one of your custom models.
LeagueIt may appear as All or as a filtered league.
MarketShows the current market: 1X2, Double Chance, Over 2.5, BTTS, or corners.
SampleHow many games are included in the calculation of that history.
HitPercentage of analyzed results that matched within that specific cut.
ReturnHistorical result considering the odds used in the calculation.
Weekly historyShortcut to see the evolution of hit rate and simulation over the weeks.

A model with a high Hit Rate and a Return below zero may be hitting very low odds. A model with a lower Hit Rate and a Return above zero may be finding better prices.

What changes when you switch the Active Model

When you switch the Active Model, Predora.si starts using a different lens to calculate games. Bookmaker odds still come from the market; what changes is the way the app interprets those odds.

Can change

  • probabilities;
  • fair odds;
  • EV and edge;
  • main analyzed outcome;
  • order of games in the grid;
  • games highlighted in Upcoming Games with EV+ and Upcoming Games with High Probability.

Does not change

  • odds published by bookmakers;
  • actual game result;
  • your saved custom models;
  • the risk and variance of an individual bet.

Active Model vs. comparison between models

Within Predora.si, you can see different models analyzing the same game. This helps you find consensus or notice divergence. But the Active Model is what drives the experience at that moment.

FeatureRole
Active ModelDefines the main grid calculations: chance, EV, fair odds, analyzed result, and filters.
Comparison between modelsShows other possible readings for the same game, helping you review context and divergence.

Simple example: the Active Model may give 53% to the home team, another model may give 48%, and one of your custom models may give 56%. That comparison is useful, but the Active Model is what decides the EV displayed as the main grid reading.

Predora.si models vs. custom models

On Predora.si, the Active Model can be an in-house model or a model created by you. None of them is always the best. The best model depends on the goal, the market, and the observed performance.

Model typeWhat isWhen it makes sense
Predora AI ProIn-house model based on machine learning and historical data.When you want a ready-made reading that is more independent from bookmaker consensus.
Predora Market MedianConverts bookmaker odds into margin-free probabilities and takes the median for each outcome. With fewer than three valid bookmakers, it uses one bookmaker's margin-free probabilities.When you want to compare a bookmaker's price with a margin-free market baseline.
Custom modelUser-created model based on the Predora Market Median database, with boosts by factor and by market.When you want to test your own analysis lens with more control.

How to interpret the Active Model in the long run

The Active Model should be evaluated calmly, not in the heat of a single game. A bet with positive EV can lose. A bet with negative EV can win. That happens because football has a lot of variance.

To evaluate better, look mainly at:

  • Sample: how many games are included in the calculation;
  • Hit rate: how many analyzed outcomes matched;
  • Return: what the result would have been considering the odds;
  • Market: 1x2, goals, BTTS, Double Chance, or corners;
  • League: where the model performs better or worse;
  • Segment: all games, positive EV, or high probability.
A small sample is more volatile statistically. A larger sample tends to give a more reliable reading.

How to apply the Active Model on Predora.si

A simple flow for using the Active Model on a daily basis:

1. Check which model is active

Before analyzing any game, look at the Active Model card.

2. Choose the right market

Chance and EV always apply to the selected market.

3. Look at sample, hit rate, and simulation

If you filtered a league, check whether the sample is still sufficient.

4. Compare probability with odds

The main question is whether the best odds are above the model's fair odds.

5. Use the right page for your goal

Upcoming Games broadens the view, Upcoming Games with EV+ focuses on EV, and Upcoming Games with High Probability focuses on high chance.

6. Open the game and compare context

See odds, factors, team comparison, and AI + Stats.

Want to see the Active Model reading right now?

Open the grid, check the model in use, and compare chance, EV, and best odds in the day's games.

How to switch the Active Model

  1. Go to Models.
  2. Choose a model under Predora.si Models or My Custom Models.
  3. Click Select.
  4. Done: that model becomes the Active Model.

From then on, the analysis pages recalculate probabilities, fair odds, EV, and highlights based on that model. Switching models does not delete your custom models and does not change bookmaker odds.

Common mistakes when using the Active Model

1. Using a model without understanding its assumptions

It connects probabilities, prices, and match evidence in a structured analysis.

2. Looking only at the hit rate

Hit rate matters, but simulation also matters. Hitting bad odds may not pay off.

3. Ignoring the base

A model can look amazing in 20 games and average in 300.

4. Comparing different markets as if they were the same

A model can do well in goals and be only okay in 1X2.

5. Switching models after a bad streak

Bad streaks happen even with consistent models. Compare by history, not by emotion.

6. Confusing positive EV with high probability

Positive EV is about price. High probability is about chance.

7. Not comparing with factors and context

The model provides the mathematical reading, but the game context still matters.

Practical checklist

  1. What is the Active Model right now?
  2. Which market is selected?
  3. Is the sample large enough to take the history seriously?
  4. Does the Return align with the Hit Rate?
  5. Is the best odds above the fair odds?
  6. Is the EV positive, or does the scenario appear only because of high chance?
  7. Does the filtered league change the reading much?
  8. Do the game factors converge or diverge from the analyzed outcome?
  9. What price is available at the time of the analysis?
  10. Are you interpreting the whole picture or reacting to a single number?

FAQ about the Active Model

What is the Active Model on Predora.si?

It is the model currently selected to calculate probabilities, fair odds, EV, edge, analyzed result, and performance metrics on Predora.si pages.

Can I have more than one Active Model at the same time?

No. You can have several available models, including in-house models and custom models, but only one is active at a time.

Does switching the Active Model change bookmaker odds?

No. The odds come from the bookmakers. What changes is the probability estimated by the model and, consequently, the fair odds, the edge, and the EV.

How do I assess the Active Model’s prediction?

Review how the model produces its probability, compare the relevant match factors, and explore its historical results for the same market.

Why does the EV change when I switch models?

Because EV depends on the estimated probability. If another model calculates a different chance for the same analyzed result, the EV also changes.

Does the Active Model apply to all markets?

Yes, but always within the selected market. The 1X2 reading is not the same as BTTS, Over 2.5, Double Chance, or corners.

Which metric should I look at first: hit rate or simulation?

Read both together with the Sample. Hit Rate shows the frequency of correct results, Return shows the model's simulated historical result considering the odds, and Sample shows the size of the dataset.

Conclusion

The Active Model is one of the most important pieces of Predora.si because it defines the mathematical lens used to read games. It calculates chance, fair odds, EV, and performance within the selected market.

The idea is not to follow numbers on autopilot. The strongest path is to use the Active Model to identify divergences, compare it with the observed odds, and contextualize the estimate with factors, stats, and history.