Volume, acceleration, and candle filters on a completed double bottom

The finished double bottom is the starting sample

A programmed double-bottom scan was applied to S&P 500 and Russell 3000 constituents from 2001 through the start of 2018. A double bottom is a two-trough reversal whose completion is marked by a break above the intervening peak. The breakout is the session that first clears that mid-pattern peak and thereby completes the structure.

High volume, a shorter final ascent, and bullish breakout-day candle structure were examined as independent combination tests stacked onto the same completed double bottom. The two-trough picture was the shared starting sample, not the last word on the tape.

Volume-price filters locate participation

A volume-price filter is a participation test that requires unusually heavy volume either somewhere inside the whole pattern or only on the late advance into the breakout. One version requires some session inside the entire pattern to print at least twice the prior 14-session average volume and also the highest volume of the prior 60 sessions.

A narrower volume-price filter confines that extreme-volume session to the interval between the right trough and the breakout. That version ties volume to the late advance rather than to either trough.

Acceleration requires a shorter final ascent

The final ascent is the rally from the right trough to the breakout. An acceleration filter requires that rise to last fewer sessions than the preceding drop from the mid-pattern peak to that right trough. The comparison is a session count, not a redrawing of the same two-trough outline.

Breakout-candle filters test acceptance

A three-higher-closes overlay requires three consecutive higher closes ending on the breakout session. A long-body candle filter flags a session whose real body exceeds the bodies of the three prior sessions, applied either between the right trough and the breakout or on the breakout session itself.

A bullish breakout candle was defined as a session whose close is nearly equal to the high, whose open is nearly equal to the low, and whose upper and lower shadows are each no longer than $0.05, with no constraint on total candle length. A second breakout-candle filter requires the upper shadow to be shorter than the real body so the session does not resemble a shooting-star rejection.

Tighter stacks leave fewer events

Added filters reduced event counts relative to the unfiltered pattern. The large-cap sample fell from 943 events to 75 under the tightest bullish-candle breakout rule, and the smaller samples were treated as statistically less reliable.