by bpleone
LIVE

Arbitrage, Middles & Scalps

Cross-book line discrepancies are mathematical free money — when you can find them. The math is wired; once a multi-book feed is connected (Pinnacle / DK / FanDuel / MGM), opportunities flow in here.

* AWAITING MULTI-BOOK FEED * — EdgeStat currently polls Bovada + DK only via The Odds API. Real arb requires 3+ books with overlapping markets. Tables below render empty until the feed is wired. Compute logic (correlations.py / arbitrage.py) is already implemented.
--Live Arbs
--Middle Opportunities
--Reduced-Juice Markets
--Quotes Scanned
2Books Polled
3x dailyRefresh Rate
★ ARBITRAGEguaranteed profit, both books must accept full size
MarketSide ASide BImplied SumProfit %Stake Split

A true arb pays a small but guaranteed profit. The "Stake Split" tells you the optimal % to put on each side so payouts equalize. Real-world snag: books cap bet size and ban frequent arbers. Use these sparingly, mix in -EV plays to look like a normal bettor.

◆ MIDDLErisk = juice paid, payoff = double win if game lands in the window
MarketTypeSide ASide BMiddle Window

A middle is when you can bet OVER at one line and UNDER at another. If the game total lands BETWEEN them, both bets win — a "middle hit." You risk a small loss (the combined juice) and win a payoff if it middles. In MLB, totals middles cash ~7% of the time historically, which is enough to make zero-juice middles +EV.

▾ LOW VIG (under 3%)cheapest 2-way pairings across books
MarketSide ASide BCombined Vig

Standard markets carry 4-5% vig. By splitting one side at book A and the other at book B, you can routinely find pairings with under 3% combined vig — meaning your model only needs a small edge to clear it. Pinnacle is usually one side of every low-vig pair.

🎓 The reality check

Arb betting looks too good to be true on paper. In practice: