About this site

An editorial statement

Baseball is the most numerically rich sport ever played. Every pitch, every swing, every stolen base generates data — and in that data, stories hide in plain sight.

Numbers Tell Stories finds those stories. Each morning, we scan the previous day's game logs for statistical anomalies — performances that z-score testing flags as genuinely unusual. Then we write about them. Clearly. In two languages.

The writing is done by AI. We are transparent about that. The numbers are real. The anomalies are real. The narratives are machine-generated, and you should read them as such.

Our four principles

Daily

One story per day. Not a recap, not a roundup — one anomaly, told well, every morning during the MLB season.

Transparent

Every article is labeled AI-generated. We use Claude Sonnet for writing and Claude Haiku for anomaly selection. No ghost-writing.

Statistical

Selection is driven by z-score, not narrative convenience. If a performance didn't register as statistically extreme, it doesn't make the cut.

Bilingual

Every article is published in both English and Japanese. MLB is watched globally. Stories should travel.

Data sources

MLB Stats API — official game and player statistics
Baseball Savant (Statcast) — pitch-level and batted ball data

Affiliate disclosure

This site participates in affiliate programs including Amazon Associates and MLB.tv. Articles may contain links that earn commissions. This does not influence editorial choices — the AI selects anomalies purely on statistical merit.