1. Home
  2. About
  3. Evaluation Method

How We Evaluate Predictions

Every prediction gets an accuracy score from 0-100%. Here's exactly how we calculate it.

5 minute read

The Formula

Accuracy = Base Score + Bonuses − Deductions

The base score depends on whether we got the result right. Bonuses reward getting the details right (total goals, margin, xG alignment). The final score is capped based on prediction type.

The Five Prediction Types

Every match gets one of five predictions based on our model's confidence:

1
Strong Home— Clear home favorite, confident pick
1X
Home or Draw— Lean home with draw as safety net
?
Avoid— Too close to call, no confident pick
X2
Away or Draw— Lean away with draw as safety net
2
Strong Away— Clear away favorite, confident pick

Base Score Ranges

The base score is determined by what we predicted and what actually happened. Scoring starts at the minimum of the range—bonuses push it toward the maximum.

Strong Predictions (1 or 2)

OutcomeBase RangeMeaning
Predicted team wins85–100%Nailed it
Draw50–70%Close miss, one goal shy
Other team wins0–50%Wrong call

Range Predictions (1X or X2)

OutcomeBase RangeMeaning
Primary hit (leaning team wins)75–90%Got our lean right
Secondary hit (draw)75–90%Safety net worked
Miss (other team wins)0–50%Wrong call

Avoid Predictions (?)

OutcomeBase RangeMeaning
Any outcome50–75%Bonuses determine final score

For "avoid" picks, we didn't back a specific outcome—so bonuses like "ended in draw" or "tight 1-goal win" determine the final score.

Bonuses That Increase Your Score

Getting the result right is only part of the story. These bonuses reward predictions that got the details right too.

Close Miss Bonus

+8 to +14%

When the result was just 1 goal away from being a hit. For example: we predicted "1" (home win), they drew 1-1. One more goal and we'd have been right.

Avoid Outcome Bonus

+6 to +15%

For "avoid" predictions that proved correct. Draw = +15% (we called it tight, it was). 1-goal win = +6-10% (still a close game).

xG Direction Bonus

+2 to +10%

When match statistics (xG) back up our prediction. If we said "home to win" and the home team created more expected goals, we get a bonus—even if they lost.

Total Goals Bonus

+2 to +10%

When we predicted the right number of total goals. If we predicted 2.3 total goals and the match ended 1-1 (2 goals), that's a near-perfect call.

Dominance Bonus

+3 to +8%

When we predicted the right winning margin. If we predicted home by 1.2 goals and they won 2-1, we nailed the margin.

Chaos Bonus

+4 to +8%

When a match goes wild (5+ goals) and we predicted a reasonable scoring game (2.0-4.5 goals). High-scoring chaos is hard to predict exactly—we get partial credit for being in the ballpark.

Upset Alert Bonus

+6 to +12%

When we flagged upset risk and the underdog delivered. Even if we backed the favorite, the upset warning showed we saw the risk coming.

Contrarian Bonus

+4 to +6%

When we went against popular opinion and were right. Only awarded for correct predictions— being contrarian and wrong doesn't earn a bonus.

Possible Deductions

These reduce your score when the prediction got key details wrong. Important: wrong predictions (misses) don't get deductions—starting at 0% is punishment enough.

xG Direction Mismatch— Match stats went against our call
−6%
Total Goals Off— Predicted 2 goals, got 5
−2 to −4%
Wrong Team Dominated— Predicted tight game, one team crushed it
−4%

Score Caps

Even with bonuses, each prediction type has a maximum possible score:

100%
Strong picks (1, 2)
Confident and correct
90%
Range picks (1X, X2)
Hedged with safety net
75%
Avoid picks (?)
No confident prediction

Evaluation Categories

The final accuracy score determines the evaluation category:

75%+
Success
Prediction was on point
50–74%
Partial
Close but not quite
0–49%
Wrong
Missed the call

See It In Action

Every finished match on the Results page shows its accuracy score with a full breakdown. Click any match to see the base score, bonuses earned, and deductions applied.

View Results

Related