2018issue C118
Evaluating double bottoms with a locked stochastic confirmation
A 2018 sample of 3388 double bottoms was resorted after crude filters were stripped away. A separate confirmation rule then keeps an overbought-oversold stochastic-oscillator print from changing meaning on its own.
- After crude filters were stripped away, pattern-height still sorted the historical double-bottom sample into short, middle, and tall groups.
- Bottom-to-bottom price variation and volume did not discriminate in that sample, so the retained checks compared trough order and a short bottom-volume-window instead of a hard breakout-day-volume floor.
- An overbought-oversold stochastic-oscillator print does not assign trade direction. The same daily oversold reading was described as a long-side setup and as a print that preceded a decline.
- Independent-confirmation from a 200-day simple moving average plus support and resistance, read with daily and weekly bars, keeps one oscillator message from changing meaning in silence.
A two-layer falsification drill
A double-bottom is a two-trough price structure whose intervening peak and later breakout define a testable reversal hypothesis rather than a finished forecast. A stochastic-oscillator is a bounded reading of where the latest close sits inside a chosen lookback range, computed on daily or weekly bars. Overbought-oversold is an extreme oscillator print that marks location inside that lookback range and does not, by itself, assign trade direction.
Editorial interpretation: run those definitions as two stacked checks. First ask which geometric traits of a double-bottom still sort a historical sample after crude filters are stripped away. Then lock each stochastic-oscillator reading to an independent-confirmation rule so the same overbought-oversold print cannot silently change meaning.
What still sorted the sample
A 2018 reader evaluation assembled 3388 manually identified double bottoms from 1080 stocks covering 1991 through 2018. Pattern-height, the distance from the trough structure to the breakout price, remained the attribute used to order short, middle, and tall groups. That evaluation used the median height-to-breakout price as the cutoff between short and tall double bottoms.
Bottom-to-bottom price variation and volume were treated as non-discriminating attributes in that sample. The same three height bands, 0-10 percent, 10-20 percent, and more than 20 percent, were kept as the comparison frame once other screens were removed.
Filters that were stripped away
A 100000-share floor on breakout-day-volume was omitted because it cut the usable sample from more than 2000 cases to 164. Breakout-day-volume is share turnover on the session that clears the intervening peak. A hard share-count floor can shrink the usable sample sharply.
Screens that dropped stocks priced under 5 or over 500 were also omitted so those three height bands could be compared without those price limits.
Rules that were retained
Two retained rule checks were a left bottom priced above the right bottom, and right-bottom volume, averaged over the five sessions from two days before through two days after the trough, exceeding left-bottom volume. That comparison uses a bottom-volume-window: a short symmetric window around each trough, such as two sessions before through two sessions after, used to compare activity at the right bottom with activity at the left.
When one oscillator print means two things
A contemporaneous letter observed that an oversold daily stochastic-oscillator during an uptrend was described both as a potential long-side setup and, on one illustrated advance, as a reading that preceded a decline. The same letter cited a weekly stochastic-oscillator that was overbought at the start of April and already trending lower, after which no upside reversal appeared.
The author's reply treated identical oscillator or pattern signals as messages that can differ by context and described a signal as a possibility rather than a guaranteed outcome. The reply specified a 200-day simple moving average plus support and resistance as extra confirmation and stated that weekly and daily stochastics should be read together with those tools. Independent-confirmation is a second, separately defined condition such as a long moving-average location or a support and resistance test that can ratify or reject the first signal.
All readings on this track · 32 readings
- 1988Reaction length as a trend integrity test
- 1991Sold-out double bottoms as a three-gate inventory test
- 1991A breadth classifier for V-bottoms and W-bottoms
- 1991Precomputed price-ratio clusters and double-top tests
- 1992Bond turning points as a regime check on equity double tops and breakouts
- 1992Commodity-bond ratio as an equity regime overlay
- 1992Gold lead confirmation for commodity-index turns
- 1994Constructing the thousand-line advance-decline indicator
- 1995Evaluating zero-line patterns on a breadth-price oscillator
- 1996Constructing double tops from a resistance retest to a trough break
- 1996Four-stage double-bottom construction
- 1998Double-bottom confirmation and stop placement
- 2000Two-bar reversal construction
- 2001Constructing double tops from failed resistance retests
- 2002Eve-Eve double bottoms: width, confirmation, and overhead resistance
- 2002Constructing Eve-and-Eve and classic double bottoms
- 2003Eve-Adam double bottoms as a two-step classroom test
- 2003Shape contrast then breakout confirmation in Adam and Eve double bottoms
- 2003Reading cyclical bottoms inside secular bear regimes
- 2004A case study of the shark-attack Fibonacci retracement
- 2004Confirming index turns with envelopes, divergence, and breadth
- 2005A five-wave euro/dollar case and the support that still had to fail
- 2007Constructing commodity seasonal indexes for regime context
- 2009Constructing rounded and double-top short setups
- 2010Hourly pattern entries, exits, and abstention as one playbook
- 2016Ugly double bottom after a yearly low
- 2016An unconfirmed stock double bottom next to a confirmed index
- 2016Constructing a range-midpoint moving average
- 2017Evaluating whole-dollar delays on pattern breakouts
- 2018Volume-confirmed bottoms and breakouts with moving averages
- 2018Evaluating double bottoms with a locked stochastic confirmation
- 2019Forex pairs as relative value: yield spreads, support, and a double bottom