Chart Patterns (trading-patterns)
Transparent, parameterised detectors - every hit shows the measurements behind it.
python3 scripts/patterns.py --list # what each pattern means
python3 scripts/patterns.py NVDA # all patterns, one symbol
python3 scripts/patterns.py AAPL MSFT AMZN --patterns vcp,breakout --min-score 60
python3 scripts/patterns.py --universe assets/watchlists/global-large-caps.txt --min-score 70
python3 scripts/patterns.py --csv bars.csv --symbol XYZ --json
Output fields
| field | meaning |
|---|---|
score |
0-100 quality of the pattern by its own criteria (not a probability) |
pivot |
trigger price - the setup is "on" when price trades above it |
stop |
where the pattern is invalidated (use as a first stop candidate) |
to_pivot_% |
distance from the last close to the pivot |
risk_% |
pivot-to-stop distance; > 8-10% usually means the base is too loose |
details |
the measurements (contraction depths, pole gain, volume ratios...) |
Workflow
- Scan (this skill) -> 2. confirm trend/volume with trading-chart ->
- check the index with trading-market -> 4. build entry/stop/size with trading-plan ->
- sanity-check odds with trading-odds -> 6. trading-risk -> 7. trading-broker.
Notes
- VCP and cup-and-handle need ~1 year of daily bars (use
--period 2y). - Daily patterns are the default; intraday works with
--interval 15m --period 1mo(breakout, flag, inside bar, NR7). - Prefer setups with
risk_%small enough that a normal position size still fits the user's risk limit. - A pattern is a hypothesis, not a signal. Many fail; say so when presenting results.