Research
What happened after similar market patterns?
Choose a market and a time window. Pattern Lab finds historically similar periods and shows what happened next.
Historical similarity is not a prediction or trading signal. These are descriptions of what happened after comparable periods in the past.
S&P 500 · 2026-10-09 · current regime: up trend, normal volatility
Similar periods rose 57.5% of the time
Over the following 20 days, 23 of 40 similar historical periods ended higher.
Matched comparison
59.7%
Same market, similar volatility and momentum
Difference
−2.2 percentage points
That is −2.2 percentage points lower.
Historical matches
40
Non-overlapping similar periods found
What happened next
Returns over the 20 days after historically similar periods, against a matched comparison sample from the same market.
What does this mean?
Historical periods with a similar market shape were followed by positive 20-day returns less often than the matched comparison sample. This describes historical outcomes and does not predict a specific future price.
Advanced statistics
These statistics are provided for readers who want to inspect the underlying analysis in more detail.
| Statistic | Value | What it is |
|---|---|---|
| Similar periods (N) | 40 | Non-overlapping analogues found |
| Probability of a positive return | 57.5% | Share of analogues that ended higher |
| Matched probability | 59.7% | Same measure in the matched control sample |
| Lift | −2.2 percentage points | Analogue rate minus matched rate |
| Permutation p-value | 0.329 | How unusual the observed difference is under the analysis procedure. Not a probability that the market will rise. |
| Median return | +0.70% | Analogues |
| 10th percentile | −4.74% | Analogues |
| 90th percentile | +2.71% | Analogues |
| Unconditional probability | 60.1% | All 1,238 observations in the same market |
| Mean DTW distance | 0.333 | Average shape distance of the analogues |
Important limitations
Pattern similarity is based on historical price behaviour. Markets can change structurally over time, and similar historical shapes can lead to very different outcomes. Results are descriptive research rather than forecasts or investment advice.
Analysis generated 2026-10-11 18:40 UTC · methodology version 2026-08-12.1.
Validation: walk-forward backtest · Validation: multivariate test · Raw analysis JSON