Why a similar historical situation is not a forecast

Updated 2026-10-09 Español·Deutsch·Русский·Italiano·Français·Türkçe·العربية

A historical analogue describes a past episode. To obtain a forecast, separately justify the relationship between the similarity feature and later outcomes. Visual or numerical resemblance alone cannot do this.

Similar shape

  • A
  • B
Similar shapeSimilarity measures shape, not the future. [[100, 105, 101, 108, 104], [100, 104, 102, 107, 104]]100108Time →
  • Similarity measures shape, not the future.
Teaching diagrams with hypothetical numbers. They explain concepts, not the next price move.

Frequency and probability

Seven rises among ten selected cases give an observed 70% frequency. That accurately describes those cases.

Claiming a 70% chance of a rise in the next case needs comparable conditions, defensible selection and validation of the estimate. Small samples and dependent observations increase uncertainty.

Probabilistic market models are possible; post-hoc case frequency simply does not become a reliable model automatically.

Three common problems

Dependence: overlapping episodes or several examples from one trend can repeat information.

Changing conditions: liquidity, volatility, participants and trading regimes vary; relationships may weaken or disappear.

Outcome-driven selection: adjusting length, instruments or similarity after examining outcomes risks finding convincing chance patterns.

Different outcomes

  • A
  • B
  • C
Different outcomesContinuation paths represent hypothetical completed past cases.. The share of past outcomes is not the next outcome’s probability. [[104, 110, 114], [104, 100, 96], [104, 106, 103]]96114Time →
  • Continuation paths represent hypothetical completed past cases.
  • The share of past outcomes is not the next outcome’s probability.
Teaching diagrams with hypothetical numbers. They explain concepts, not the next price move.

Independent validation

Fix selection and measurement before examining evaluation data, then test on an unused period. Respect information availability: a sample’s later continuation must not influence its selection or similarity.

Compare a simple baseline. If the market generally rose, many rising analogues do not establish added search value. Strategy evaluation also includes commissions, spread and slippage.

Legitimate uses

Analogues help explore diverse continuations, compare context and formulate testable hypotheses. Differences are informative; inspect the whole set and outcome dispersion.

Lo, Mamaysky and Wang (2000) study formal chart features under particular data and methods. Their findings do not validate every analogue search or tool.

Similarity supports investigation, not promised repetition or investment advice.

Related questions

In depth

See it on a live chart

The same markup on a prepared example: the chart, the objects it found and what followed.

Open the example