Anchoring bias is the tendency for an initial value to influence a later numerical estimate or adjustment, even when the anchor is not sufficiently informative. The foundational heuristics-and-biases paper includes anchoring and insufficient adjustment. In betting, test the idea by saving an estimate before seeing the possible anchor; simply noticing that two prices differ is not evidence of bias.
What counts as an anchor
| Possible anchor | Later judgement it might influence | Evidence needed for an audit |
|---|---|---|
| First displayed odds | Fair-price or probability estimate | Estimate saved before and after display |
| Opening line | Interpretation of a later line | Independent model output and timestamped revisions |
| Pundit percentage | Personal forecast | Forecast made before hearing the percentage |
| Previous match score | Estimate for a new match | Defined model showing whether the score is a valid input |
| Round number such as 50% | Confidence report | Elicitation without the number for comparison |
An anchor can be relevant information. The bias claim concerns disproportionate influence, not the mere presence of a reference point. A market price can contain information, so an analyst must define whether the task is an independent forecast, an update from market evidence, or a comparison of executable prices.
Illustrative price audit
Suppose a forecast saved before viewing odds assigns a 45% probability to a clearly defined event. Its model-implied fair decimal price is 1 / 0.45 = 2.222... before costs or uncertainty, using the reciprocal relationship described in the expected-value and probability framework.
The first displayed price is 2.50. A second price is 2.30. The second price can feel unattractive relative to 2.50, but that comparison does not answer whether 2.30 is favourable under the forecast. At p = 0.45, expected net return per GBP 1 at decimal 2.30 is:
0.45 * (2.30 - 1) + 0.55 * (-1) = 0.035, or GBP 0.035.
Expected value is a probability-weighted average, not a promised result. The calculation depends entirely on the 45% estimate being valid and the price being accepted. It does not prove an edge or prove that anchoring occurred.
Audit the revision path
- Define the event and information cutoff.
- Save the probability and method before seeing the possible anchor.
- Record the anchor, source, and time.
- List new evidence learned after the anchor.
- Recalculate from the evidence rather than adjusting by feel.
- Save the revised estimate and reason for every change.
- Compare many similarly designed decisions before drawing a conclusion.
If a forecast cannot be produced without market information, call it a market-informed forecast. Do not pretend it is independent.
What anchoring is not
- A shorter later price is not automatically poor value.
- A longer later price is not automatically good value.
- Refusing to ignore useful market information is not necessarily anchoring.
- One changed estimate does not establish a systematic bias.
- Removing an anchor does not make the remaining model accurate.
Next step
Use Cognitive Biases Betting for the next part of this topic.
Continue learning
- Next guide: Confirmation Bias in Betting
- Related guide: Dunning-Kruger Effect in Betting
Assumptions and limitations
The probability and prices are illustrative. Anchoring research does not show that every displayed price causes a predictable adjustment in every person. A pre-price audit can expose a revision path, but alternative explanations and model error remain. Bias awareness does not make gambling safe or profitable.
GambleAware
GamCare
Gambling Therapy
GAMSTOP