Windows (gaps) in Japanese analysis act as support when rising and resistance when falling. Island reversals, isolated by gaps on both sides, trap participants whose forced exits fuel the reversal. Nison's window rules applied.
These complete the multi-candle vocabulary by focusing on the role of gaps and price clusters.
| Pattern | Confirmation | Volume | Reliability | Common Failure Mode |
|---|---|---|---|---|
| Rising window / falling window | Subsequent sessions respect the gap as support (rising window) or resistance (falling window). Confirmation is structural — the window holds. | Heavy volume on the gap candle confirms the gap is meaningful rather than thin-trading noise. | Generally regarded as among the more reliable continuation signals when the window holds for several sessions. Becomes invalid when filled. | The window fills within a few sessions, especially common with low-volume gaps. |
| Island reversal | The second gap (against the prior trend) is the confirmation built into the pattern. Further follow-through strengthens. | Heavy volume on both gap candles is ideal. | Generally regarded as a high-conviction reversal signal when it occurs because two gaps on opposite sides of an isolated cluster is statistically uncommon. | The cluster's gaps are partial (shadows overlap) rather than true gaps, so what looks like an island is actually a different pattern. |
| Mat hold | The fifth candle. Must close beyond the first candle's close. | Same as rising three methods — heavy on the bookend candles, light on the middle. | Generally regarded as a strong continuation signal because the gap adds structural support beyond what rising three methods provides. | Gap fills during the pullback, removing the structural support that distinguishes mat hold from rising three methods. |
Rising window: QuantifiedStrategies reports the rising window pattern showed a high efficiency rating in 31.59% of occurrences when tested over a 5-candlestick period in the S&P 500 over a 20-year period. CandleScanner's research found that across 502 S&P 500 symbols over a 20-year date range, the rising window occurred in 5.73% of all candlestick occurrences, with average frequency of 61.1. Teaching point: The CandleScanner frequency statistic is useful for setting student expectations — windows appear roughly once every 60 trading sessions in their data, which is uncommon enough that students shouldn't expect to see them constantly but common enough to encounter regularly across multiple instruments. Falling window: No separate large-sample statistic widely cited; the structural logic and reliability are generally regarded as symmetric to the rising window. Island reversal: The pattern's reliability is primarily described qualitatively across sources — two true gaps on opposite sides of a price cluster is statistically uncommon under normal market conditions, which is the basis for the high-conviction assessment. Misidentification (shadow overlap disqualifying the true gap) is the dominant data-quality issue in any backtest of this pattern. Mat hold: The pattern is rare enough that sample sizes in public backtests are limited. PatternsWizard and thepatternsite.com note the pattern as a strong continuation signal when correctly formed, though frequency rank is low — similar to falling three methods. Multi-source pool (windows and islands): candlescanner.com, quantifiedstrategies.com, trendspider.com, luxalgo.com, synapsetrading.com, thepatternsite.com. Multi-source pool (mat hold): thepatternsite.com, patternswizard.com, therobusttrader.com.
Key Takeaways
A rising window formed two weeks ago. Price has since pulled back and closed within the window's range. What does this mean for the window's structural significance?