‹ HomeIT

Fischietto finale

Bundesliga · matchday 8

Real dataStatistics onlyUpdated 26/10 08:39

Every prediction is saved before the match and compared with the result.

In short

Over 20 matches the model got the outcome (1, X or 2) right 50% of the time. Always picking a home win would be right about 51% of the time. Its percentages are 2% less wrong than those taken from league averages alone.

With fewer than 30 matches this is only a first look: the numbers can change a lot.

How the last round went

23 October – 25 October · outcome right 4 out of 9 (44%)

The surprises

Stuttgart – M'gladbach · 2-3outcome given 13%
Augsburg – Union Berlin · 1-1outcome given 19%
Mainz – Bremen · 2-3outcome given 22%

The best predictions

Dortmund – Frankfurt · 1-01 at 73% ✓
RB Leipzig – Elversberg · 4-01 at 59% ✓
Freiburg – Bayern · 1-32 at 55% ✓
Model performance
50% 1X2 right
Sample still growing

10 correct predictions out of 20. Last 12: 4 correct. More matches are needed before judging form.

50%
✓✕✓✕✕✓✓✕✕✕✓✓
1X2 outcome right
10 of 20
50%
Average stated probability
54%
if they match, the model is calibrated
Model reliability
Sample still growing

Real results are needed to calculate a reliable score. No correction is applied.

Lab · look back

20 predictions

Comparison on predictions actually saved before the result. No prediction is rebuilt after the fact.

40%Last 10 · 10 matches
Brier 0.65
50%Last 25 · 20 matches · incomplete sample
Brier 0.59
50%Last 50 · 20 matches · incomplete sample
Brier 0.59
MarketHitActualGap
111/2055%gap +5%
25/2025%gap -4%
X4/2020%gap -2%
1X15/2075%gap +4%
X29/2045%gap -5%
Over 1.512/2060%gap -25%
Over 2.511/2055%gap -11%
Under 2.59/2045%gap +11%
Under 3.513/2065%gap +10%
Both score11/2055%gap -9%

Model quality

Model error
0.59
the lower, the better
Simple method
0.60
league averages only
Over/Under 2.5
12 of 20
60%
Both teams score
12 of 20
60%

The model is 2% less wrong than the simple method. With 20 matches the figure is still shaky: it needs at least fifty matches, ideally a hundred.

When I say 60%, does it happen?

40%60%80%100%0%50%100%Stated probabilityActual hitsunder 45%: stated 42%, actual 0% over 3 matches345-55%: stated 50%, actual 67% over 9 matches955-65%: stated 58%, actual 50% over 4 matches465% or more: stated 70%, actual 50% over 4 matches4

Each point is a band of predictions (the number next to it says how many matches). The closer the points are to the dashed line, the more reliable the percentages.

ProbabilityHitsStatedActual
under 45%42%0%
45-55%50%67%
55-65%58%50%
65% or more70%50%

If "Stated" and "Actual" are close, the percentages are reliable.

Does the news help?

Over 16 matches: with the news the error is 0.42, with statistics alone 0.42. Too early to say. Outcome right: 7 of 16 with the news, 7 of 16 with statistics alone.

New model or old?

Over 20 matches, statistics only: error 0.41 for the current model, 0.47 for the first version. Too early to say.

Reliability by market

A 0–100 indicator built from calibration and Brier score. It is descriptive, not a guarantee of results. The model's picks use the history only when there are at least 10 observations per category.

1X2—
44 of 100 selections · 20 matchespred. 44%actual 44%
Both score—
20 of 40 selections · 20 matchespred. 50%actual 50%
Over/Under—
45 of 80 selections · 20 matchespred. 60%actual 56%
Combination—
58 of 160 selections · 20 matchespred. 41%actual 36%
Exact score—
1 of 20 selections · 20 matchespred. 10%actual 5%

Sample still too small: the system keeps collecting data without changing the model's probabilities.

Latest predictions

newest → oldest
Freiburg – Bayern2 given 55% · final score 1-3✓
Hoffenheim – Leverkusen2 given 44% · final score 1-0✕
Dortmund – Frankfurt1 given 73% · final score 1-0✓
Mainz – Bremen1 given 57% · final score 2-3✕
1. FC Köln – Schalke1 given 41% · final score 0-0✕
RB Leipzig – Elversberg1 given 59% · final score 4-0✓
SC Paderborn – HSV1 given 49% · final score 1-0✓
Augsburg – Union Berlin1 given 66% · final score 1-1✕
Stuttgart – M'gladbach1 given 70% · final score 2-3✕
M'gladbach – Hoffenheim1 given 42% · final score 1-2✕
Leverkusen – Freiburg1 given 56% · final score 2-1✓
Bayern – RB Leipzig1 given 70% · final score 6-1✓
Union Berlin – Dortmund2 given 58% · final score 0-0✕
HSV – Stuttgart2 given 48% · final score 1-0✕
Bremen – SC Paderborn1 given 51% · final score 2-2✕
Elversberg – Augsburg1 given 49% · final score 4-3✓
Schalke – Mainz1 given 46% · final score 0-1✕
Frankfurt – 1. FC Köln1 given 52% · final score 1-0✓
Freiburg – Schalke1 given 53% · final score 2-1✓
1. FC Köln – M'gladbach1 given 52% · final score 2-1✓