Five ways to analyze a bet. We use all five.
Fundamental & Situational
Matchups, injuries, quarterback and backup tier, weather, rest, scheme, motivation — the why behind a number. Most “situational angles” (revenge, letdown, lookahead) are myths the market already priced, so we test them and refuse to sell the ones that don’t hold. The ones that do — QB value, wind on totals — we surface.
Quantitative & Model-Based
A team-strength and prop model that owes nothing to the betting line — power ratings from game results only, and a prop score fit on real graded outcomes. Market-independent on purpose: it’s the one number you can honestly hold against the market instead of just echoing it.
Market & Line Value
The closing line is the best single estimate of a game that exists — so we track where the number moves, shop for the best price, and measure closing-line value. When our model and the market disagree, that gap is the only honest definition of an edge. We never claim to see “sharp money” we can’t actually observe.
Statistical, Trends & Regression
What’s sticky and what isn’t. Records built on turnover margin, one-score games and red-zone luck regress; streaks and “hot hands” are usually already in the price. We show the trend — who’s covering, who’s due to fall — but label it as context, and we live by one rule: out-of-sample or it doesn’t count.
Risk & Bankroll Management
This one doesn’t pick winners — it makes an edge survive variance. Flat unit sizing, expected value over hit rate, and the guardrails that stop the bets that quietly bleed you: long-odds overs, all-over parlays, single-book prices. The best handicap in the world loses if the staking blows up the roll on a normal cold streak.
How we weight them
They aren’t equal. We anchor on the market (03) because it’s the sharpest estimate available, and hold everything to the discipline of (04) — nothing counts until it survives out-of-sample. Our own model (02) is a market-independent second opinion, fundamentals (01) are the context that explains the number, and risk management (05) is baked into every guardrail so a good read isn’t undone by a bad stake.
That’s why you’ll never see a single magic “1–100” score here. Different sports and different angles carry different amounts of evidence, so we report them as what they are — a validated edge, a model lean, or a situational spot — never flattened into one number that hides how much is actually known.
The honesty layer on top of all five
- Validated edge vs. model lean vs. situational spot — every play is tiered by how much evidence backs it.
- “Assumes independence” on every combined-probability number — parlays compound the vig, and we say so.
- Availability states (Verified / Partial / Pending / Stale) so missing context never reads as neutral.
- No “sharp,” “steam,” or “lock” language. We don’t claim to see money we can’t.
- Out-of-sample or it doesn’t count. The rules that looked unbeatable in-sample and reversed taught us this the expensive way.