Transparent betting models for college football & MLB.
Every pick is published before first pitch or kickoff. Every result is graded from the final score. The track record on this page rebuilds itself every morning — nothing is edited after the fact.
Where it stands
Read this first. The season is only a few weeks old, so the win-loss and ROI numbers are small samples and will bounce around. Closing-line value — whether the market moves toward our number after we post — is the earliest honest signal that a model is doing something real, which is why it sits up here too.
How it works
Three things worth knowing before you read a single pick — what actually generates a number, what has to be true before that number becomes a published play, and how you check our work without taking our word for any of it. The longer, plain-language version is on the Methodology page.
The Models
Two purpose-built projection systems — not one template stretched across two sports.
MLB. Every game is projected from the matchup itself: how each lineup has been scoring, over the full season and lately; the starting pitchers' underlying skill, not just their ERA; each bullpen and how hard it has been worked; the ballpark; and the temperature and wind at first pitch. The moneyline, run line and total all come from that one projection, so they never contradict each other.
College Football. A projected margin and total for every game, built from team talent (recruiting, plus the players each program gains and loses through the transfer portal), established power ratings, play-by-play efficiency, scoring and returning production.
The Discipline
A model is only as good as the rules that keep it from lying to itself.
- Every published pick risks the same flat stake, every time — no size tells, no chasing the last loss.
- A pick publishes only if the model's edge over the market's de-vigged price clears a fixed bar, and only after the projection behind it passes a sanity check built to catch a bad park factor, a broken rate stat, or any other data problem before it becomes a “play.”
- When two available bets are really the same wager twice — a moneyline and its correlated run line on the same side of the same game — only the cleaner one publishes. Betting both would just double the stake on one outcome.
- A claimed edge large enough to look more like a data glitch than real value is logged and graded, never published — a rule we hold ourselves to specifically so a lucky one-off can never slip into the record as a repeatable “win.”
- Model changes are batched and re-tested against the full calibration history before they ship. Never a same-day reaction to one bad beat.
The Proof
Every claim on this page is checked against a public ledger, not our word for it.
- Every pick is logged before kickoff or first pitch with the exact line and price taken, then settled automatically from the final score — nothing is entered, edited or removed after the fact.
- Closing Line Value is tracked on every graded pick: did the market move toward our number between publish and close? It's the fastest honest read on real predictive skill, because it doesn't wait on how one season's games happen to fall.
- The “by claimed edge” tables on the Track Record page ask the question a model can't dodge: do the plays we said we were most confident in actually win the most? Every band is shown, gaps included, whether the answer flatters the model or not.
- The full record — every market, every month, every edge band, both sports — rebuilds itself from the ledger every morning. This page cannot show a curated version of the season.
What's inside every brief
The exact shape of a published pick — matchup, market, model edge, and the reasoning behind it. Illustrative example below, not a live pick — see the Track Record for actual graded results.
Why not just follow a tout?
A tout and a model you can actually check work differently in ways that matter once real money is on the line.
| Typical tout | Houdini Wagers | |
|---|---|---|
| Stake sizing | Arbitrary “confidence units,” picked after the fact | Flat stake, every pick, fixed before the result is known |
| Record | Screenshotted wins; losses often quietly dropped | Every pick logged before kickoff, graded automatically, nothing edited after |
| Method | “Trust my gut,” or a proprietary black box | Documented model — see the methodology |
| Self-check | Take their word for it | Full by-market, by-month and by-edge breakdown on the Track Record page |
Questions
How often do picks publish?
Every day there's a slate — college football in season and MLB through the end of its season. The brief lands before the day's first game.
What do you actually publish?
Moneyline, spread/run line and total picks, plus daily 3, 4 and 5-leg parlay cards — only when the model's edge clears a fixed bar. No edge, no pick; a quiet day is a valid day.
Is it free?
Yes, for now — see below to get it by e-mail.
Is this financial or betting advice?
No. This is an independent analytics project for informational and entertainment purposes only. See the disclaimer in the footer, and only ever wager what you can afford to lose.
Do you show the losses too?
Every one of them. The Track Record page is generated from the same ledger the model is graded against — it can't show a curated season.
How do I unsubscribe?
Every e-mail has an unsubscribe link. One click, no questions asked.
The daily brief
Every morning: the day's model card for CFB and MLB — picks, the number behind each one, and the running record. No hype, no locks.
Free for now. Unsubscribe any time. We never sell your address.