What is sharp money?

A sportsbook does not treat every customer the same. Some bettors get limited because the book has learned that their bets predict where the line is going. Money from those accounts is what people mean by sharp money. It is a real thing inside a book. It is not a real thing in any public data feed. No split, no odds screen and no alert can see which account placed a bet.

So when a site says a side is getting sharp money, it is making an inference from public numbers. The two numbers most often used are the gap between money share and ticket share, and the direction of line movement. Both are worth understanding on their own terms. Neither is a label that a book attached to a bet.

• Money outrunning tickets. If a side has 40% of tickets and 65% of money, the bets on that side are much larger on average. Large bets are a sign of a big bankroll, which is sometimes a sign of a professional.
• The line moving against the crowd. If most tickets are on one side and the number moves the other way, the book had a reason to move it that was not ticket count. Money is one possible reason.
• Timing. Moves in the first hours after a line opens, or in the last minutes before kickoff, are often read as respected money because books post limits low at the open and high at the close.
• Agreement across books. A move that shows up at several books within minutes is more likely to be a market-wide reaction than one book adjusting its own position.

Each of these is a reasonable thing to notice. Each also has ordinary explanations that have nothing to do with skill: an injury report, a weather change, a book balancing a lopsided position, or a single large recreational bet.

Giants at Rams on 9/21/2026 shows the first sign and not the second. Read at 8:14 PM ET on game day, a minute before kickoff, the Giants held 58% of spread tickets and 61% of spread money, so the money ran ahead of the tickets. But the spread did not move on game day: every consensus snapshot Locksmith tracked, from the morning to kickoff, had the Rams -6.5 and the total 47.5, with 60% of the total money on the over. Bigger bets on a side, and a book that saw no reason to reprice. One sign out of two. Those figures were read at 8:14 PM ET on game day; the game page shows the closing numbers, which can differ from a pre-kickoff snapshot by a point.

Buccaneers at Lions on 9/15/2024 shows both. The Buccaneers held 72% of spread tickets but only 46% of the money; the Lions held 28% of tickets and 54% of the money. The Lions opened -6 and closed -7.5, a 1.5-point move toward the side with fewer tickets and more money. That is the shape the label gets attached to. What the numbers establish is narrower: bettors on the Lions staked more per bet, and the market made them lay more points. Whether those bets came from respected accounts is not in the data. The Buccaneers won 20-16, which says nothing about the reading either way. One game never does.

Locksmith grades every NFL game since 2023 week 5 at its closing line. Betting the side whose money share was above its ticket share covered the spread 50.0% (384-384-13) and hit the total 50.8% (396-383-6). Break-even at -110 is 52.4%. Both readings sit under it. Those counts cover 822 NFL games, 2023 week 5 to 2026 week 2, Action Network closing splits, refreshed hourly from the engine.

A stricter version, the side holding 60% or more of the money, covered 49.7% of spreads, and fading that side returned -3.6% (as of 9/21/2026, 822 NFL games 2023 week 5 to 2026 week 2, Action Network closing splits). Following the heavy money lost. Fading it lost too. The vig ate both. Adding the second sign, the line moving toward the money side since the open, gives 50.7% (115-112-3) on the spread.

The two-leg version, both money sides of the same game as a parlay, hit 22.2% against a 27.8% break-even. You can check every one of these on the backtest page, with the confidence interval next to each hit rate.

• Who bet. Public splits cannot see accounts. Every sharp label on a public site is a guess about bet size and timing.
• Whether the price is still good. Respected money that moved a line from -6 to -7.5 was placed at -6. You are being offered -7.5.
• Whether it will cover. In this dataset, the readings people call sharp have covered at about the rate of a coin flip, before vig.
• Anything a book has not shown. Splits are a sample from participating books. Limits, buybacks and exchange volume are not in them.

Keep the term as shorthand for two measurable things: bigger bets on a side, and a line moving toward it. Read both against the opening number and the news, and expect them to describe the market rather than beat it. The market table shows the three reads for every NFL game and the record of past games that showed the same thing. Splits say where the money sits. They do not say what will happen.