We play every FBS matchup out thousands of times, snap by snap, and set that range of outcomes against four models and 404 proven signals. Every spread, total, and moneyline gets a grade — A+ down to the games we pass on — and every call we've ever made stays on the public record, losses included.
Three seasons of history, and every variable we could measure — kickoff window, day of the week, conference, week of the season, line size, how far our models sat from the market — crossed with the situational spots bettors argue about: rest, travel, rankings, coaching changes, streaks, revenge games. For spreads and totals that made 31,000 candidate signals. For the moneyline, where who wins is a different question than who covers, the search ran past a million candidate patterns over five million graded outcomes. Every one was graded point-in-time, at the line as it stood, then tested against random chance.
Almost everything died. 253 spread-and-total signals and 151 moneyline patterns survived — 404 in all. Those are the voters: a pick's grade, C up to A+, is how many of them agree on it once the ones pointing the other way are subtracted out. One voice alone doesn't get your money — that grades No Edge. And underneath, four models rebuilt from scratch — opponent-adjusted, zero fudge factors, every backtest that looked too good torn down until we found the bug. Newest of all, the engine: a play-by-play simulation that plays each game thousands of times and gets the last word on the spread. Agree with the signals and the bet is A+. Disagree and it comes off the board.
One layer sits on top and never votes: the streak marker. When a team's recent form against our projections adds an extra scenario that aligns with the standard signals, the pick wears a small bolt — a marker on the grade, never part of it. A second marker, the chain link, belongs to the Chain: a fifth model on trial this season, spreads only, that looks at each game once and marks a pick only when it lands on the same side. Its own fires are graded in the open, agree or disagree.
That's what a grade is made of. What it earns gets settled in public.
The backbone of the product. Every graded bet — every spread, total, and moneyline — gets a row like this; the rest of the site is built on top of them. Here's the spread on Oklahoma @ Michigan, and how to read it.
This is one row from the Board — a single bet. Every graded spread, total, and moneyline gets its own. Click any row to expand the journal, the model reads, and the tier rationale.
Same picks, same grades, same numbers everywhere — each surface is a different job you'd hire them for. The last one is the only page here that tracks your money instead of our record.
Every bet on the board — spread, total, and moneyline — tier-graded, best book named, re-graded every 30 minutes. The "what's the bet right now" view.
The whole slate ahead — every game a card, its grades on the front — then, once games kick, a live scoreboard with every pick tracked ▲ or ▼ against the score in real time. The "what's coming, what's playing" view.
Stack legs straight off the board — every one priced at the live market, our combined hit probability against the ticket's own break-even. The "stack the edges" view.
Give it a bankroll and it splits the whole stack — heavier where the edge is bigger, and at one sportsbook's lines if you only play one. The "how much on each" view.
Every FBS team, ranked by the engine — the margin it expects against an average team over thousands of played-out games. See what the system believes before you see what it bets. The "who's actually good" view.
Live Lines for finished games — every pick graded at the number we released, linked to the full breakdown. The "hold us to it" view.
Every pick opens into its game page — the engine's range of outcomes with the market line and every model's number on it, the game path minute by minute, the signals and the engine's word behind the grade, the full journal, the team stats, and live scoring through the final whistle. The "under the hood" view.
Your own bets, on your own numbers — log a side, a book, a price and an amount in a few taps, singles or parlays, or send the whole Allocator sheet over at once. Private to your login. It runs live during games so every ticket carries its standing at a glance, and settles at the line you got, not the one we published. The "how am I actually doing" view.
A record only means something if you can't move the goalposts. Here's the whole lifecycle of every pick.
We grade ourselves on the number we gave you — not the one that would've looked best.
SP+ — Bill Connelly's efficiency model. Adjusts for opponent strength.
PPA — Expected points per play. The drive-by-drive read.
Advanced — Success rate, explosiveness, finishing drives. The "how" of efficiency.
Pace+ — Our in-house model. Possession elasticity weighted by opponent strength.
The engine — Our play-by-play simulation. Plays each game thousands of times from the same team numbers, and gets the last word on the spread.
A pick doesn't start because the models average out a certain way. It starts when a model's number lands in a proven spot — a specific combination of model, market, and line range that survived the three-season test — and usually several signals fire on the same side at once. The grade counts that agreement across the 404 survivors. The four models carry the signals; the engine checks them — on the same side, the bet is A+; on the other side, it comes off the board. And when nothing proven lines up, we don't manufacture a lean — that market publishes as No Edge.
These are release prices. They lock for the life of your subscription. Resubscribers pay current pricing.