03 Configuration and Tuning

Configuration and Tuning

All thresholds are module-level constants at the top of scan.py. Two of them (MIN_SCORE, MAX_BREAKOUT_AGE) can also be set per run from the CLI; everything else is edited in the file so the rules stay auditable in one place. There are no environment variables and no API keys.

CLI

Flag	Default	Effect
--tickers A,B,C	full S&P 500	scan only these symbols (upper-cased)
--csv path	pinned GitHub dataset	local constituents CSV; Symbol column or first column
--period	2y	yfinance period; 2y gives about 500 daily bars, enough for SMA200 plus a 250-bar cup
--min-score	60	overrides MIN_SCORE for the run and is echoed in meta.min_score
--max-age	3	overrides MAX_BREAKOUT_AGE; the effective per-pattern limits are in meta.max_breakout_age_by_pattern
--out-dir	output	destination for signals.json and report.md
-v	off	DEBUG logging, including per-symbol last-bar detail

Global gates

Constant	Default	What it controls	Loosen →	Tighten →
PIVOT_ORDER	5	bars on each side needed to call a swing high/low	fewer bars: more (noisier) pivots, patterns visible sooner	fewer, cleaner pivots; H&S/Wolfe seen later; BREAKOUT_AGE_LAG follows it automatically
ATR_LEN	14	ATR window used for stops and the head/overshoot tests
MIN_SCORE	60	minimum quality score reported	more marginal setups	only the cleanest geometry
MAX_BREAKOUT_AGE	3	max bars since the confirming close (Cup); +PIVOT_ORDER for H&S and Wolfe	older breakouts reported	only fresh breakouts
BREAKOUT_AGE_LAG	Cup 0, H&S 5, Wolfe 5	extra age tolerated because the last pivot lags
MAX_RUNAWAY	0.05	close more than this above the trigger = chasing, dropped
WATCH_PROXIMITY	0.05	setups whose close is within this of the trigger → watchlist (was 0.03 until 2026-09-05; at 3 % a normal down day cleared most of the list)	longer watchlist	shorter watchlist
LAST_BAR_MIN_FRACTION	0.5	share of symbols that must have a complete bar for it to be meta.last_bar
FILL_CLOSE_MIN_AGE	1 h	how old the last trade must be to count as the closing print
CONSTITUENTS_COMMIT	2026-08-20 hash	pinned upstream commit of the constituent CSV

Trend gate

Constant	Default	Meaning	Tuning note
TREND_STRONG_DOWN	0.90	close below this fraction of a falling SMA200 = strong down-trend; reversal patterns are vetoed	1.0 would veto any close below a falling SMA200
TREND_SLOPE_LOOKBACK	40	bars back used to decide whether the SMA200 is falling	shorter reacts faster to a roll-over, longer ignores wobble
TREND_STRONG_DOWN_SMA50	0.85	the same test against SMA50 when fewer than 200 bars exist, or when the SMA200's slope cannot be judged yet (200–239 bars)

The up-trend requirement for the cup (close above SMA200) is not a constant.

Cup & Handle

Constant	Default	Meaning	Tuning note
CUP_MIN_LEN / CUP_MAX_LEN	30 / 250	rim-to-rim width in bars	O'Neil: 7 weeks to a year; 30 bars is already short
CUP_MIN_DEPTH / CUP_MAX_DEPTH	0.12 / 0.50	depth as a fraction of the left rim	the score peaks at 25 %
CUP_RIM_TOL	0.05	right rim within 5 % of the left rim	wider tolerance admits ascending/descending cups
CUP_PRIOR_ADVANCE	0.25	required rise into the left rim (120-bar look-back)	the main "it must be continuing something" filter
CUP_MIN_ROUNDNESS	0.60	R² of the convex quadratic fit of cup lows	rejects ragged bases; on its own it would pass a clean V (R² ≈ 0.93)
CUP_MAX_V_ADVANTAGE	0.0	how much the best two-legged V fit's R² may exceed the parabola's	0 = the U must explain the lows at least as well as a V; negative values demand a clear U; ~0.07 would re-admit clean Vs
HANDLE_MIN_LEN / HANDLE_MAX_LEN	5 / 40	handle length in bars	below 5 bars a close above the running high is treated as the handle still forming
HANDLE_MAX_DEPTH	0.12	handle pull-back vs. the right rim	O'Neil's 12 %
HANDLE_MAX_FRACTION_OF_CUP	0.50	handle depth vs. cup depth

Not configurable but relevant: the cup bottom must sit in the middle 60 % of the cup, and the handle low must stay in the upper half of the cup.

Inverse Head & Shoulders

Constant	Default	Meaning	Tuning note
IHS_MIN_LEN / IHS_MAX_LEN	20 / 200	shoulder-to-shoulder width
IHS_MIN_HEAD_ATR	1.0	head at least this many ATR below both shoulders	the scale-free "is there really a head" test
IHS_SHOULDER_SYM	0.50	shoulder gap as a fraction of the shallower shoulder depth
IHS_TIME_SYM	2.5	max ratio of left-half to right-half duration
IHS_MAX_NECK_SLOPE	0.15	neckline change over the width, as a fraction of the head price	beyond this it is a trend line
IHS_PRIOR_DECLINE	0.10	required decline into the left shoulder as a share of the 60-bar high

Bullish Wolfe Wave

Constant	Default	Meaning	Tuning note
WW_MIN_LEN / WW_MAX_LEN	15 / 200	point-1 to point-5 width
WW_MAX_OVERSHOOT_ATR	2.0	max undercut of line 1-3 by point 5	more = accepts deeper false breakdowns; the −0.5 ATR "must reach the line" floor is fixed
WW_MAX_BARS_SINCE_P5	25	point 5 must be within the last 25 bars
WW_MAX_ETA_BARS	250	lines 1-3 and 2-4 must meet within this many bars after point 5 for a target to be reported	guards against near-parallel lines projecting absurd targets

False-positive filters, in the order they act

1. Trend gate — the cheapest filter; runs before any pivot search. Cups need an up-trend; reversal patterns are vetoed in a strong down-trend.
2. Geometry — the width/depth/symmetry/slope rules above. Each is a hard reject.
3. Prior move — a cup must continue a ≥ 25 % advance, an inverse H&S must reverse a ≥ 10 % decline.
4. Roundness (cup only) — R² ≥ 0.60 and the parabola must fit at least as well as a V.
5. Confirmation state — stale (> age limit), failed (closed back below the trigger) and runaway (> 5 % above) breakouts are dropped; setups more than 5 % below the trigger are dropped.
6. Risk — stop above entry or risk > 15 % is dropped.
7. Score — everything surviving with a score below MIN_SCORE is dropped.
8. De-duplication — only the best-scoring signal per (ticker, pattern, status) is kept, so overlapping pivot combinations never inflate the count.

Measuring the effect of a change

The random-walk sweep in test_scan.py::test_random_walk_false_positive_rate prints the share of 200 synthetic 500-bar series on which any detector fires (baseline 1.5 %, hard limit 5 %):

python -m pytest test_scan.py -k random_walk -q -s

For sensitivity, run the negative-control mutations (test_patterns.py -k violations) after loosening a threshold: each mutation names the rule it violates, so a newly passing mutation tells you which rule you have effectively removed.

python -m pytest test_patterns.py -k "violations or controls" -q

Suggested procedure for a real-data tuning pass: run the scan on a date range with --tickers on a small set, compare notes against the chart, adjust one constant, re-run both commands above, and only then re-scan.