Bets placed by professional or analytically-driven bettors known to historically win at scale.
Sportsbooks don't treat all money the same. Every major book maintains a tier system for its customers. Recreational accounts that lose over time are welcomed and their action does not move lines much. Sharp accounts — those producing positive closing line value over thousands of bets — are watched carefully. Their bets move lines even when the dollar amounts are modest, because the book's risk managers know the probability the line is wrong has just spiked.
The distinction matters because line moves driven by sharp money carry information that line moves driven by recreational money do not. If a soft book's line moves because $500,000 of recreational money piled onto a popular favorite, the line shift reflects bet flow, not new probability information. If a sharp shop's line moves because a known winner bet the underdog at $25,000, the shift reflects the market's updated belief about the true probability.
Following sharp money — or trying to — is one of the oldest strategies in retail betting. The practical problem is that sharp money is not labeled in any public feed. You can see line moves and approximate the public-money split, but you can't observe which accounts placed the bets. Inferences come from comparing book behaviors: if Pinnacle moves a line aggressively before the soft books catch up, the smart guess is that Pinnacle is responding to sharp action.
The honest framing is that sharp money is real but its signal-to-noise in public data is messy. Reverse line movement is the cleanest proxy. Steam moves — sudden, multi-book line shifts in the same direction within minutes — are another. Both are best used as confirmations layered on top of an independent model, not as standalone strategies.
Our process is independent of sharp-money tracking — we publish based on model edge, not on following sharps. We do, however, treat alignment with sharp action as positive confirmation and divergence as a flag to reaudit.