Institutional-grade research on sports betting markets.
We study how betting markets price outcomes, and publish the results in full: the data, the method, the uncertainty, and the parts that did not work.
No picks. No tips. No promises about beating anyone.
The MLB moneyline is almost perfectly calibrated. Every bucket still lost money.
23,992 closing moneylines from 2021 to 2025. Every probability band’s calibration interval contains zero, and yet all ten price buckets lost money. The gap between those two facts is the vig.
- Return on risk
- -3.55%
- Across every price bucket
- Vig per side
- +1.87 pp
- Charged at the close
- Event-selections
- 23,992
- Six full seasons
Closing MLB moneylines, 2021–2025
What you can learn
Research notes
Long-form studies of market behaviour, each frozen against a versioned dataset so the numbers can be checked rather than taken on trust.
Browse the researchHistorical Explorer
Ask what actually happened when the market looked a particular way, across six seasons of MLB moneyline closing prices.
Open the ExplorerOpen data
Every published figure ships with machine-readable tables and a hash of the specification it was computed against.
See a published package
First Win Pro
Free is what we found. Pro is what to do with it.
- Free
- The finding, the conclusion and the evidence for it. Complete on its own, and staying that way.
- Pro
- What the result rules out, where it breaks down, and what is worth investigating next.
Pro does not exist because we hide the answer from free readers. Every public study is complete on its own. Pro is additive.
Billing is handled by Stripe. No picks, no tips, and no claim that any of this will make you money betting.
| Free | Pro | |
|---|---|---|
| Major public research studies | included | included |
| Main findings, full conclusion and core charts | included | included |
| Historical Explorer | included | included |
| Public methodology and frozen specs | included | included |
| Research newsletter | included | included |
| Pro Brief: implications, and what to investigate next | not included | included |
| Technical Research Pack behind each major study | not included | included |
| Complete supporting tables and additional cuts | not included | included |
| Robustness and sensitivity work | not included | included |
| Pro-only research notes | not included | included |
| Downloadable research assets, where licensing permits | not included | included |
| Early access to completed research | not included | included |
| Research queue participation | not included | included |
Why trust it
- Methods are frozen before results are computed.
- Every publication carries the SHA-256 of the specification it was run against, so a bucket cannot be moved after seeing what it produced.
- Uncertainty is reported, not hidden.
- Results that cannot be distinguished from chance are described that way, including when that makes the finding less interesting.
- The data is published.
- Tables and chart-ready series are downloadable alongside every note, and the site recomputes nothing it did not publish.
- Corrections are published too.
- When a method is found to be wrong it is versioned, explained, and the original is kept as provenance.
Get new sports-market research when we publish
Occasional email, only when a new research note goes out. No picks, no promotions. Everything on this site stays free to read without an account.
Institutional / Data Access
The corpus and research engine behind this site can support work beyond the published notes:
- Research and data licensing
- Historical sports market datasets
- Programmatic API and data access
- Custom research on a specific market question
- Institutional access for desks and research teams