Every feature, in depth.
Eight layers between a sportsbook's price and a bet you should take. We don't just find +EV — we route it, size it, score it, and track whether the close confirmed it.
Five ways to find true odds.
The price you see on a book is wrapped in the book's margin (“the vig”). Strip the vig, and the price reflects the market's true estimate of probability. Strip it five different ways and average the estimates, and you get a true line sharper than any single book can publish.
Whet runs five de-vig methods on every market: multiplicative, power, worst-case, Shin, and weighted-consensus. The five disagree on books with thin margins or asymmetric exposure — the disagreement itself is a signal. Where they agree, the consensus is the true odds we score every bet against.
25+ filters that reject bad spots before you click.
A +EV signal isn't enough. NBA second night of a back-to-back erodes points. NFL games in >15mph wind torch passing props. MLB hitters miss the lineup card and the prop voids — but only on some books. NHL goalie confirmation flips a +EV prop into a dead one in 30 seconds. We reject these spots before they get surfaced, per sport, per filter.
The 25+ filters cover B2B (NBA), wind / precip / game-total (NFL), lineup confirmation / BvP / park factor (MLB), goalie confirmation / PP-PK splits / SOG (NHL), and rotation / importance / form (soccer). A few examples of how the engine scores a filtered context, not just an unfiltered price:
- NBA back-to-back drops the score multiplier by ~10% for high-usage stars. Three games in four nights does worse.
- NFL wind > 15mph rejects passing-yard overs outright. Receiver props drop their score by 15-25% depending on team script.
- MLB BvP under 20 PA doesn't move the score; over 20 PA it's an input. Small samples don't pretend to be big ones.
Smart sizing that respects linked bets.
Kelly tells you what fraction of bankroll to risk given an edge. Full Kelly is too aggressive for variance; fractional Kelly (quarter, half) is the standard. Whet ships quarter-Kelly by default, capped at 3% per bet — math you can verify, not a black box.
The harder problem: linked bets. Tatum 25+ points AND Celtics over 116 are correlated — both pay if Boston has a high-pace offensive night. Stacked unconditionally, the combined stake is over-leveraged. Whet detects the correlation and de-rates the stakes so the linked exposure stays inside the single-bet cap.
Smart routing that respects your limits.
Sportsbooks rank from aggressive (Fanatics, DraftKings) to tolerant (BetMGM, Caesars) to sharp (Pinnacle). Aggressive books cut limits fast on sharp players. Tolerant books accept larger action on bigger edges without flagging. Sharp books don't flag at all but offer thinner margins.
Whet routes each bet to the right book given its EV and your history. Low-EV bets go to the aggressive book that's offering the best price for the smallest stake. High-EV bets route to the tolerant book that'll take real volume. Steam-chase bets route to the sharp book before the line moves everywhere.
One number, zero to one hundred.
Every prop gets a Whet Score: a single 0-100 ranking that fuses edge, confidence, market width, recent line movement, and the strategy-filter context. 87+ is fire; 60-86 is strong; 40-59 is okay; under 40 is a skip.
The score isn't a black box — every input is visible on the prop card. If you disagree with the math, you can see which input is driving the score. If you trust the math, the score tells you which one to take and which one to skip when you've got more +EV bets than bankroll for.
Closing-line tracking, every bet.
We log the closing line for every bet you place. If you bet Tatum 27.5 at -108 and the line closes at -120, you beat the close by 12 cents — the bet was sharp, regardless of whether it won. Over a sample, CLV converges to ROI faster than win-rate does, so it's the only honest performance metric for short windows.
Whet's CLV dashboard shows your closing-line value by sport, by stat type, by book, by time-of-day. If your CLV is positive but your ROI is negative, you're just running bad. If your CLV is negative, the system isn't finding what it claims to find — and you should pause.
Backtest the full pipeline, not just the picks.
Most “backtest” tools replay historical bets and tell you what your ROI would have been. That's table stakes. Whet's replay simulator runs the entire pipeline over historical odds: de-vig, filter, score, size, route — all of it, on the snapshot of books as they existed at scan time.
You can A/B-test strategy changes (turn off the NBA B2B filter, see what would have happened). You can simulate Kelly fractions (full vs. quarter vs. flat). You can replay a single bad week and see whether the system flagged the trap or fell into it. Sharp tier only — the compute cost makes it gated.
Cross-book mispricings, live.
When two books disagree on a market by more than the combined vig, you can lock in a guaranteed return by betting both sides — on paper. That's arbitrage. The catch: arb windows close in seconds, and arb-only players get limited fast.
Whet's arb scanner flags the windows it catches each scan (~every 5 minutes), computes the optimal split across books for a fixed bankroll, and tracks which arb routes have been used recently against your accounts so you don't burn books. Sharp tier only.
Want to see all eight in action?
The scanner is the front end. Each tier adds more of the eight — Sharp runs the whole system.