Research
PatternX research
Status
No studies have published results yet. 6 studies are pre-registered below, with their full methodology.
PatternX research follows four rules:
- Real data only. Results come from PatternX’s own detectors replayed over historical market data, or from aggregated, anonymised platform records — never from invented or illustrative numbers.
- Method first. Each study’s question, success definition, criteria and exclusions are published before the analysis, so the method can’t be bent to fit the result.
- Full disclosure. Published results state the sample size, period, markets, timeframes, costs assumed, confidence intervals, exclusions and limitations.
- Reproducible. Every study records the detector version and parameters it used.
Studies without results are kept out of search engines until they have them. The detection rules the studies use are documented in how PatternX detects chart patterns and the pattern library.
Pre-registered studies
- Bull Flag Performance Study
When PatternX's production bull flag detector is replayed over historical candles using only completed bars, what share of its detections reach the measured-move target (one pole height above the breakout) before the flag-low stop within 40 bars, what distribution of net R-multiples follows after costs, and do both differ from matched random long entries with the same stop and target distances?
Status: pre-registered — not yet analysed
- Cup and Handle Across Crypto Markets
Do the outcomes of PatternX's production cup and handle detections, replayed on completed bars and measured as the share that reach the cup-depth target before the handle-low stop and as net R-multiples, differ between crypto market segments: BTC and ETH versus other actively traded coins, one exchange versus another for the same coins, and perpetual futures versus spot markets?
Status: pre-registered — not yet analysed
- Chart Pattern Performance by Timeframe
How do the detection frequency, target-before-stop rate and net R-multiple of PatternX's 20 production pattern detectors change across its eight timeframes, from 1m to 1d, when replayed on completed bars over one shared period with identical cost assumptions, and on which timeframes, if any, does mean net R differ from that of matched random entries?
Status: pre-registered — not yet analysed
- Pattern Reliability During High Volatility
Are PatternX's production pattern detections, replayed on completed bars, more or less likely to reach their targets before their stops, and do their net R-multiples differ, when the instrument's volatility at the signal bar is high relative to its own recent history rather than normal or low, once matched random entries made in the same regime are taken into account?
Status: pre-registered — not yet analysed
- Momentum Breakouts by Market Capitalization
For US stocks that trigger Momentum Lab's Premarket Momentum Breakout under default settings, does the share of triggers that gain 3% before losing 3% within 60 minutes, and their net R-multiple, differ across market-cap bands of $100M to $500M, $500M to $2B, and $2B or more, when market cap is measured as of the trigger date rather than today?
Status: pre-registered — not yet analysed
- Premarket Momentum and Relative Volume
Among US stocks that trigger Momentum Lab's Premarket Momentum Breakout under default settings, are triggers with higher relative volume at the trigger minute more likely to gain 3% before losing 3% within 60 minutes, and do they earn higher net R, when relative volume is measured both as production defines it (today's volume divided by average daily volume) and adjusted for the time of day?
Status: pre-registered — not yet analysed