About thecapper.io

What this is

thecapper.io publishes AI-assisted MLB and WNBA betting research. Every analytical piece is produced using a structured pipeline that pulls from over 300 data sources: Baseball Savant, FanGraphs, Baseball Reference, beat reporter coverage, sportsbook data, and projection systems. The output is verified, cited, and built for bettors who want to know the actual numbers behind a pick, not just a gut call.

This is not a tipster service or a tout operation. There are no "lock of the week" promotions, no selling of picks, no paid tiers. We publish our methodology, show our data sources, and track performance transparently. Every published pick and its outcome is in the public record — wins and losses alike.

Who operates it

thecapper.io is operated by Craig Fountain, a sports analytics researcher based in Ohio. The site launched in early 2026 as a public beta. I built the research pipeline, the quantitative models, and this site. I review the AI-generated content before it publishes and maintain the correctness of the underlying data pipeline.

I am not affiliated with any sportsbook. I do not accept payments for picks. The analytical outputs reflect what the models say, not what would be convenient to publish.

You can reach me directly at hello@thecapper.io or through the contact page.

How articles and picks are produced

The site runs a daily pipeline during the MLB season. At a high level:

1. Data collection — the pipeline pulls fresh lineups, pitching probables, odds, and Statcast data multiple times per day.
2. Model scoring — five quantitative models score each player or game situation: hits, strikeouts, home runs, NRFI/YRFI, and first-five innings runs. Each model produces a probability estimate and a confidence tier.
3. Publication gate — picks are only published when they clear an explicit publication contract: verified odds, minimum probability threshold, and at least projected lineups available. The gate prevents publishing noise or incomplete data.
4. AI-assisted editorial — game previews and recaps are drafted by an AI research system instructed to cross-reference statistical sources, beat reports, and sportsbook context. The drafts are reviewed before publishing.
5. Public grading — after games end, outcomes are recorded against the published picks. No retroactive edits. The performance page shows the full season record.

The full technical workflow — including how each model works, what data sources feed it, and the publication contract rules — is documented on the Methodology page.

AI content disclosure

A significant portion of the editorial content on thecapper.io — including game previews, recap narratives, and player and team analysis — is produced with AI assistance. We disclose this openly because we think it is the honest thing to do, and because the AI layer is genuinely central to how this site works, not a footnote.

AI assistance lets the pipeline process a volume of statistical sources and cross-reference a breadth of context that a single human analyst could not do cost-effectively. A single preview article draws on pitching matchup data, batter splits against specific pitch types, bullpen workload, park factors, umpire tendencies, and current sportsbook lines — all synthesized into a coherent analytical narrative.

At the same time, AI-generated content has real limitations. It can miss qualitative context, overweight statistical patterns that no longer apply, or occasionally produce a factual error. Every AI-generated piece goes through a human review step before publication. When we find errors, we publish corrections on the Corrections page.

The quantitative model picks (hits, strikeouts, home runs, NRFI, first-five) are not AI-generated in the large-language-model sense. They are outputs from a deterministic statistical pipeline (XGBoost and logistic regression models trained on historical outcomes) that I built and maintain. AI assists with the editorial narrative around those picks, not the picks themselves.

Analytical philosophy

The models trust pitching over hitting, value run prevention over run scoring, and look for market inefficiency in props more than in game lines. These aren't arbitrary preferences. They reflect where we have measured the most persistent edges in MLB betting markets.

We do not chase closing line value as the sole signal. We track it, but we also care about whether picks win. The performance page shows both the season record and historical accuracy. We publish what the models say, including days when the slate is thin and the high-confidence picks are scarce.

Track record

We track every published pick and its outcome on the model performance page. No cherry-picking, no retroactive edits. If a pick loses, it stays in the record. Verified, timestamped, public. The performance page shows season W-L, accuracy by model, and a historical pick log going back to when we started publishing officially.

What this site is not

To be explicit about what thecapper.io is not:

• Not a sportsbook or betting operator. We do not accept wagers.
• Not a tout service. We do not sell picks or take payment for picks.
• Not financial advice. Nothing published here should be construed as a recommendation to place any bet or make any financial decision.
• Not a data broker. We do not sell user data.

See our Terms of Service, Disclaimer, and Responsible Gambling pages for the full legal context and gambling risk disclosures.

Contact

Questions, corrections, or feedback: Contact page or email hello@thecapper.io.