Treat every paper umbrella as a timed experiment

Name the bar before you time it

A historical daily sample of 351,745 candlestick events, drawn from about eight years of KOSPI 200 members with roughly 1,758 bars per symbol, was used to study how price evolved after single-bar patterns. Trend context for a 1-to-16-day follow-through window was set with a 20-day moving average: 161,612 bars closed above it and 182,933 closed below it, after an upward-bias adjustment was subtracted from average holding results.

A paper umbrella required a small real body, a lower shadow at least twice that body, little or no upper shadow, and a high-low range greater than the 10-day average range. The same rules inverted defined the inverted umbrella. Under those rules a downtrend umbrella was labeled a hammer and an uptrend umbrella a hanging man. An inverted umbrella was an inverted hammer in a downtrend and a shooting star in an uptrend.

The labeled set contained 4,975 umbrellas (2,269 hammers and 2,674 hanging-man bars) and 8,310 inverted umbrellas (4,567 inverted hammers and 3,688 shooting stars).

What the return window showed

After umbrella events the sample’s average path continued lower, and hammers declined even more steeply. Hammers produced as many or more false bullish readings as correct ones, and both umbrella forms were read as selling-pressure events rather than automatic reversals.

After hanging-man bars the sample showed a short correction then a sharp upward turn around day eight. Inverted umbrellas generally headed higher, inverted hammers often reversed a prior downtrend, and shooting stars showed only a brief dip before the average path was higher by day 16.

Extra filters on white versus black real bodies and rising volume changed the paths only slightly and did not alter the general post-pattern dynamics.

Layer a 14-day RSI split at 50

A 14-day RSI split at 50 was layered on the same bars: 268 of 2,269 hammers and 2,404 of 2,674 hanging-man bars sat above 50, while 486 of 4,567 inverted hammers and 3,336 of 3,688 shooting stars sat above 50.

In that RSI overlay, inverted hammers with RSI above 50 in a downtrend showed a stronger average upward path for about 10 sessions, while shooting stars with RSI below 50, about one-tenth of shooting-star cases, were followed by an immediate downward exception.

Editorial scores: fade, delayed catch-up, or skip

This paragraph is editorial and is not an archive claim. The same paper-umbrella silhouette is scored as a fade when the 20-day label is a hammer and the average path, already below the moving average and usually below RSI 50, continued lower. It is scored as a delayed catch-up when a hanging-man bar, already above the moving average and usually above RSI 50, shows the short correction and the later turn around day eight.

This paragraph is editorial and is not an archive claim. An inverted hammer with RSI above 50 is scored as a delayed catch-up because that overlay showed a stronger average upward path for about 10 sessions. A shooting star is scored as a skip of a one-bar bearish reversal when RSI is above 50 and the average path is higher by day 16, and as a fade when RSI is below 50 and the immediate downward exception appears. White-versus-black body and rising-volume filters are scored as a skip for the general dynamic because they changed the paths only slightly.