betaider.com

Model

Baseball: how well the predictions actually do, and what they were built from

The model serving this sport has not been backtested.
Nothing here has been checked against what actually happened, so treat every number as untested until you have run python -m pipeline.backtest --sport baseball.
Is the model any good?
Log loss scores how much probability the model put on what actually happened — lower is better. What matters is the comparison with baseline, which is a model that always predicts the league's average home/draw/away split. The edge between them is the only evidence the model has learned anything. If edge is zero or negative, ignore every other number on this site.

Backtests

No backtest has been run. Walk-forward evaluation is the only honest read on a prediction model — run it before trusting any number on this site:
./.venv/bin/python -m pipeline.backtest --sport baseball --store-predictions

Stored predictions

ModelPredictionsOn upcoming fixturesGenerated
deep 1 1 4 Sep 2026 10:08
mlb-dc 51 51 4 Sep 2026 08:49

Data behind it

Matches
5493
Played
5442
Scheduled
51
Shots recorded
0
Matches with shot data
0

Covering 2024-03-01 to 2026-09-06. Source: Understat, via the soccerdata scraper.

Calibration

No calibration yet. Run ./.venv/bin/python -m pipeline.picks --calibrate.

How the model works

Each team carries an attack and a defence rating on the log scale. The goals a home side is expected to score are exp(attackhome + defenceaway + home advantage), and the away side's are exp(attackaway + defencehome). Those two numbers feed a pair of Poisson distributions, combined into a full scoreline matrix from which every market — 1X2, over/under, both teams to score, exact score — is read off.

Two departures from the textbook version matter:

The rho parameter is Dixon and Coles' correction to the low-scoring cells (0-0, 1-0, 0-1, 1-1), where treating the two scores as independent is known to misfit.

betaider.com

Sport

Baseball sections

Preferences

Appearance
Aider Reading the Baseball model · online
Open the Radar
Bet slip
Open full page
to move to open esc to close See all results

On this page

Model

What this page is for

The evidence for or against trusting anything else on this site. Every probability elsewhere comes from the model described here, and this page is where it is held to account.

How to read the backtests

Log loss scores probabilities: lower is better, and it punishes confident mistakes far harder than cautious ones. The number to compare it against is the baseline beside it, which is what you get by predicting the league's base rates and nothing else. If the model does not beat that baseline, it has learned nothing, however good its accuracy percentage looks — and a banner will say so on every page of that sport.

The backtest is walk-forward: the model is only ever asked about matches later than the ones it was fitted on. That is what makes the score meaningful rather than a measure of how well it memorised the past.

How to read the data section

Coverage is a limit on everything above it. A model fitted on few matches, or on a league with patchy shot data, produces confident-looking numbers built on very little.

Where to go next

Calibration asks a different question — not whether the model is sharp, but whether its stated probabilities are honest.

Press esc to close Read the full guide