The move, in one paragraph
A steam move is what happens when informed money hits a sportsbook market hard enough that the entire industry repositions in real time. It is not the slow drift of public action accumulating over a Wednesday-to-Sunday cycle. It is a 90-second cascade, usually triggered by a single originator placing significant volume at the wholesale market, with retail books following algorithmically within seconds. To the bettor watching a multi-book screen, it looks like the entire row of prices lighting up at once. That visual — the screen "steaming" — gave the move its name in the Don Best era of the early 1990s, and the terminology survived even as the underlying mechanics moved from phone calls to algorithms.
Steam is the cleanest real-time signal of informed money in the sports betting market. It is also one of the hardest signals to monetize, because by the time a retail bettor can see it, the price is gone.
Where steam originates
The map of who moves lines is small and well-known inside the industry. Three categories dominate.
Sharp syndicates — organized betting groups with capital, modeling, and market access. The lineage runs from the Computer Group of the 1980s (Billy Walters, Dr. Ivan Mindlin) through the modern era of quantitative funds operating out of Las Vegas, Pittsburgh, and offshore. These groups bet first at the highest-limit, lowest-vig venues to establish their position before the market moves against them.
Wholesale books — Pinnacle and BetCRIS in the offshore market, Circa Sports in the US legal market. These books accept large bets from sharps and move their lines based on the sharp action, which the rest of the market then follows. The wholesalers are not the originators of the information; they are the market-makers whose moves transmit the information to the rest of the industry.
Information leaks — injury news, weather updates, and lineup confirmations that reach a small population of sharps before the public. The classic example is a starting-quarterback scratch confirmed via a beat reporter's tweet 90 seconds before the public sees it. Sharps with monitors on the right Twitter accounts hit the books in the gap and the resulting move looks like steam even though no syndicate is involved.
Anatomy of a 90-second cascade
Walk through a typical NFL steam move on a Sunday morning. The market opens Thursday with the spread at Kansas City -3 across most US books. Public money trickles in on the favorite through Saturday, pushing the line to -3.5 at the retail books while Pinnacle and Circa hold at -3. At 9:47 AM Eastern on Sunday, a syndicate executes a series of bets totaling $400k on the dog at +3.5 at five offshore wholesale books and Circa simultaneously. Within 15 seconds, Pinnacle moves to +3 / -3. Within 45 seconds, Circa drops to -3 (-115). Within 90 seconds, the retail US books — running scrape-and-follow algorithms — have all moved to -3 / -3 (-115) / +3 (-105). The screen flashes red across the entire row. That is steam.
| Time (sec) | Event | Pinnacle line | Retail consensus |
|---|---|---|---|
| 0 | Pre-steam baseline | KC -3 / -110 | KC -3.5 / -110 |
| +8 | Syndicate bets hit Pinnacle | KC -3 / -115 | KC -3.5 / -110 |
| +22 | Pinnacle crosses the 3 | KC -2.5 / -120 | KC -3.5 / -110 |
| +40 | Circa follows | KC -2.5 / -120 | KC -3 / -110 (first movers) |
| +70 | Retail algorithmic cascade | KC -2.5 / -120 | KC -3 / -115 (most books) |
| +120 | Limits raised, market settled | KC -2.5 / -120 | KC -2.5 to -3 / -115 |
The retail bettor watching this on a multi-book monitor sees the cascade between +40 and +120 seconds. By the time they click to place a bet at the new -2.5 price, they are buying at a price the syndicate has already established as the fair number. The edge has been priced in.
The three signals that confirm real steam
Not every fast line move is steam. Air movement — algorithmic noise, single-book mistakes, manual re-positioning — can look like steam in its first 10 seconds and dissolve within minutes. Three signals separate the real article from the look-alike.
Speed and breadth: five or more books moving in the same direction within 120 seconds. A single-book move is not steam; a multi-book cascade is.
Price-tightening: when books move with steam, they often also tighten their vig on the side that just moved. A line that goes from -3 (-110) to -2.5 (-120) is more signal-rich than a line that goes from -3 (-110) to -2.5 (-110). The vig tightening signals that the book is confident in the new price as the new fair number.
Limit movement: wholesale books raise their maximum bet amounts immediately after steam, signaling that they consider the new price defensible. Public-driven moves trigger the opposite — books reduce limits and widen vig because they want to slow incoming bets while they reassess. A line that moves with rising limits is steam; a line that moves with falling limits is fear.
What sharp money looks for in a steam target
Steam does not happen at random. Syndicates allocate capital to markets where their model edge is largest and where the public consensus has pushed the line furthest from fair. The three classic conditions: recreational lopsided action on one side moving the retail line away from the wholesale line; information asymmetry from injury or weather news not yet priced; and structural mispricing at key numbers where the book has been forced to take a position to keep the headline price intact. When all three converge — public hammers a favorite into a key number, key player questionable for game-time decision, retail line drifts 1.5 points off Pinnacle — the conditions for steam are in place.
How retail bettors should use the steam signal
The instinct of the new bettor is to follow steam — see the move, take the new price, ride the syndicate's coattails. This is almost always wrong. By the time the retail bettor sees the cascade complete at second +90, the price has moved through the value zone and the bettor is now competing at the post-move price, which the syndicate established because it was the new fair number, not a discounted one. Chasing steam systematically loses to taking the pre-steam price (which the bettor cannot reach because they are not the originator).
The correct retail use of steam is diagnostic. Track every steam move across a season — which side, what magnitude, what market — and compare against your own model. The pattern of where your bets agree with steam (and where they disagree) is one of the cleanest external calibration signals you can build. Bettors who consistently place bets in the pre-steam direction generate strong closing-line value. Bettors who chase steam after the move generate negative CLV and lose long-term.
Reverse steam and headfakes
The market sometimes moves in the direction opposite to where the smart money actually sits. Three causes account for most reverse-steam scenarios. Distributed positioning — a syndicate places its real position quietly across dozens of books over hours, then a recreational counter-move pushes the line to better closing prices for the position already built. Deliberate headfakes — a small originator places a one-book bet to trigger an algorithmic cascade, then takes the better price on the opposite side at the books that chased. Misread cascades — algorithmic scrapers see one book move on a mistake and chase it before the originating book corrects, creating a brief mis-priced consensus that mean-reverts within minutes. The cleanest defense against being faked is to watch the limits, not just the line. When books raise limits on the un-moved side, that is where the real money is.
The arms race: how books defend
Steam was a more profitable game for retail bettors in the 1990s and early 2000s, when line monitors were slower and human traders at retail books made manual repricing decisions. The modern US market is almost fully algorithmic. The defensive layers a sharp bettor encounters today include automated line freezes (a market pauses for 15-90 seconds when the algorithm detects a steam pattern), account flagging (bettors who consistently win the pre-steam window get tagged and limited within 20-50 bets), limit ladders (max bet sizes that drop dramatically for flagged accounts), and follow-the-originator scrapers that re-price retail lines within 3-5 seconds of Pinnacle or Circa moving. The result is that the time available for a retail sharp to act on steam has collapsed from minutes to seconds, and the bettor population able to act has narrowed to those with API-grade book access and dedicated infrastructure.
Steam as a marker of market efficiency
From a market-microstructure perspective, steam is the mechanism by which sportsbook prices converge to their efficient-market closing line. Academic research (Štrumbelj 2014, Snowberg & Wolfers 2010) finds that the closing line in major US sports markets is one of the most accurate publicly available probability estimates for game outcomes — more accurate than most analyst projections, more accurate than most model outputs. Steam is the transmission mechanism. Without sharp money repricing inefficient retail lines, the closing line would be a noisy consensus rather than the sharp price it has become. Bettors who beat the closing line systematically beat the most efficient probability estimate in the market — a high bar, but the only one that matters long-term.
A retail playbook for steam
- Do not chase steam blindly. The post-cascade price is rarely a buy.
- Track steam diagnostically — log every move, compare to your model, calibrate your edge.
- Focus on closing-line value, not on instantaneous steam-following. CLV is the long-run predictor.
- Watch limit movement, not just price. Limits raised confirm steam; limits cut signal fear.
- Recognize the speed signature — 5+ books in 120 seconds is steam; anything slower is public money.
- Accept the limit risk — if you do beat steam consistently, books will limit you. It is the structural cost of being sharp.
Sources & further reading
- Levitt, Steven D. "Why are gambling markets organised so differently from financial markets?" Economic Journal, 2004 — bookmaker pricing strategy and information asymmetry.
- Snowberg, Erik & Wolfers, Justin. "Explaining the favorite-longshot bias: Is it risk-love or misperceptions?" Journal of Political Economy, 2010 — efficiency of closing lines as probability estimates.
- Štrumbelj, Erik. "On determining probability forecasts from betting odds." International Journal of Forecasting, 2014 — how to extract efficient probabilities from market lines.
- Pinnacle Betting Resources — public documentation on market-making, limit structure, and the role of sharp action in line setting.
- Daniel Wallach & Action Network — modern reporting on steam-tracking infrastructure, Circa Sports' role in US wholesale pricing, and the algorithmic arms race in retail books.