Steam Moves: How to Verify a Rapid Odds Change

Define the rule before looking

Because "steam" is informal, an analysis needs its own operational definition. One illustrative rule might require:

• the same selection and settlement rules at three or more sources;
• a raw implied-probability change of at least 3 percentage points;
• the change to occur within 15 minutes; and
• a minimum executable stake at the new price.

Those values are research settings, not industry standards. Changing them after seeing a dramatic move creates selection bias.

Worked observation

Suppose three sources move from about 2.50 to 2.30 within ten minutes:

Under the stated rule, this qualifies. The data does not say whether the cause was news, copied prices, customer orders or an automated model update.

Data-quality checks

1. Confirm all sources refer to the same market and period.
2. Distinguish executable odds from cached comparison feeds.
3. Store the available amount where the platform provides it.
4. Record suspensions and missing snapshots rather than filling them silently.
5. Preserve moves that reverse, not only those that look meaningful after the match.

Betfair's exchange guide explains why displayed amount and matching status matter on an exchange.

Testing whether a rule adds information

A steam rule should be evaluated on future data with a benchmark and uncertainty, not by listing memorable winners. Probability-calibration guidance is relevant when the output is a probability. The historical football-market study by Goddard and Asimakopoulos also illustrates why conclusions belong to the sample and method actually tested.

Predeclare what counts as a rapid move

Because steam has no universal threshold, an analysis should define its own rule before examining outcomes. A record might require the same selection to shorten by at least a stated percentage across three named sources within ten minutes, with a minimum executable stake. Those numbers are research choices, not industry facts, and should be tested rather than presented as a standard.

A price move from 2.50 to 2.25 can be expressed as a relative price change of 2.25 / 2.50 - 1 = -10%, or as a raw reciprocal change from 40.00% to 44.44%, an increase of 4.44 percentage points. State which measure is used; they are not interchangeable.

Quality controls

• Match event, market, selection and settlement period across sources.
• Exclude stale, suspended or unavailable quotes under a declared rule.
• Retain moves that reverse as well as moves that persist.
• Compare against a predeclared benchmark and time window.
• Evaluate all detected cases, not only memorable winners.

The result should be a reproducible market observation. Without transaction or news evidence, the cause remains unknown; without out-of-sample evaluation, the label does not establish predictive usefulness.

Related resources

Compare ordinary odds movement and how to investigate an outlier price.

Continue learning

• Next guide: Even Money in Betting
• Related guide: Football Odds Comparison

Assumptions and limitations

The threshold and table are illustrative. A correlated move across sources may reflect shared suppliers rather than independent confirmation. A verified steam move is still only a price event; it is not a recommendation or evidence of future profit.