Skip to content

Latest commit

 

History

History
106 lines (80 loc) · 35.6 KB

catalog.md

File metadata and controls

106 lines (80 loc) · 35.6 KB

Preset catalog

Every preset is one versioned JSON query file in presets/ — an outcome, optional base conditions, parameters that expand into query fragments, and original definition prose. Run one with edgestats report <id> --symbol <SYM> (add --param key=value or --group <field> to reshape it), and the result arrives as the engine's standard envelope: sample size, Wilson 95% confidence interval, minimum-sample guards, a first-half/second-half stability split, per-year counts, and — where the outcome carries one — a value distribution. Every summary below states its denominator and numerator; these are historical conditional frequencies, not forecasts. The catalog is a directory, so extending it is a pull request, not a feature request.

Gaps

Preset Title What it measures Params Group by Beyond a fixed report
gap-and-go Gap and Go Of the sessions that gapped in the chosen direction, how often the gap NEVER filled AND the session closed beyond its own open in the gap's direction: the continuation measured directly, not inferred as the complement of the fill rate. dir (up/down/any, default any); minGapPct (number) gapBucket a real outcome, not the mirrored fill rate: no-fill AND directional close are both required, so a stalled gap can't masquerade as a go; grouped by gap size out of the box: the go rate is famously size-dependent, and the per-bucket N shows exactly where the sample thins out; a size floor turns 'gap and go' into a testable definition instead of a slogan; N + 95% CI + stability split on every number
gap-fill-by-direction Gap Fill by Direction The gap-fill rate with up-gaps and down-gaps side by side. minGapPct (number) gapDir both directions in one result, each with its own N and 95% CI; a size floor keeps micro-gaps from diluting the comparison; any further condition: weekday, events, volatility regime: applies to both rows at once
gap-fill-by-size Gap Fill by Gap Size The gap-fill rate split by gap-size bucket: xs under 0.15% of prior close, s 0.15-0.3%, m 0.3-0.75%, l 0.75-1.5%, xl at or above 1.5%. dir (up/down) gapBucket the buckets are a default view: absGapPct filters give any continuous cut of the same data; per-bucket N and 95% CI instead of one blended fill rate; time-to-fill distribution attached to every run
gap-fill-by-weekday Gap Fill by Weekday The gap-fill rate split by weekday. dir (up/down); minGapPct (number) dayOfWeek one grouped result instead of five separate weekday reports; per-weekday N and 95% CI expose which day-of-week differences are noise; direction and size filters stack on top of the weekday split; unlimited history: your data decides the depth
gap-fill-time Time to Gap Fill The same question as gap-fill: denominator: sessions that opened away from the prior close, roll days excluded; numerator: the ones that traded back to it: read through the minutes-to-fill distribution attached to the result. dir (up/down); minGapPct (number) — median and quantile fill times, not a single average; the distribution recomputes under any composed condition; fill rate and fill clock arrive in one envelope, so neither is quoted without the other
gap-fill Gap Fill Of the sessions that opened away from the prior session's close, how often price traded back to that close. minGapPct (number); maxGapPct (number); dir (up/down) — any composed condition (FVG state, event days, prior-day direction, volatility regime): not a fixed dropdown set; time-to-fill distribution alongside the fill rate; N + 95% CI + stability split on every number; unlimited history: your data decides the depth
gap-reversal Gap Fill and Reverse Of the sessions that filled their opening gap, how often the session went on to close beyond the prior close against the gap's direction: fill-and-reverse rather than fill-and-hold. dir (up/down); minGapPct (number) — an explicitly conditional denominator (filled gaps), stated rather than implied; direction and size parameters reshape the question without a new report; N + 95% CI + per-year counts on every number

Opening range

Preset Title What it measures Params Group by Beyond a fixed report
orb-by-direction Opening Range Break Direction Of the sessions with a formed opening range, how often the FIRST break went the chosen way. window (duration, default 15m); dir (up/down, default up) dayOfWeek any opening-range window, not a fixed menu; up and down runs share one denominator, so the partition always adds up; per-weekday N and 95% CI on the directional rate
orb-false-break Opening Range False Break Of the sessions whose opening range FIRST broke in the given direction, how often that break failed: price closed back inside the range or beyond the opposite side. window (duration, default 15m); dir (up/down/any, default any) — the failure denominator is explicit: breaking sessions, not all sessions; window and direction are parameters, so the failure rate is checkable for your exact setup; compose with gap state, weekday, or event days to see when failures cluster
orb-target Opening Range Extension Targets Of the sessions whose opening range first broke in the given direction, how often price extended at least r times the range beyond the broken side. window (duration, default 15m); r (number, default 1); dir (up/down/any, default any) — r is continuous: the target ladder is yours, not a fixed 0.5/1/2 menu; the realized-extension distribution shows the whole curve in one query; N + 95% CI per rung keeps thin ladders honest
orb-time-to-break Time to Opening Range Break The opening-range break rate, read through the minutes-from-open-to-first-break distribution attached to the result. window (duration, default 15m); dir (up/down/any, default any) — median and quantile break times, not a single average; any derived window works: the clock is per-window, not one-size; rate and clock arrive in one envelope with the same N
orb Opening Range Breakout Of the sessions with a formed opening range, how often the range broke: in either or a chosen direction: plus time-to-break. window (duration, default 15m); dir (up/down/any, default any) — any opening-range window, not a fixed 5/15/30 menu; false-break and target hit-rate ladders as sibling presets on the same derived columns; compose with any condition: gap state, weekday, event days, volatility regime

Initial balance

Preset Title What it measures Params Group by Beyond a fixed report
ib-extension Initial Balance Extension Of the sessions that broke their initial balance in the given direction, how often price extended at least r times the IB range beyond the broken side. r (number, default 1); dir (up/down/any, default any) — a continuous extension ladder instead of fixed target lines; the conditional denominator (breaking sessions) is stated, not implied; N + 95% CI + stability split on every rung
initial-balance Initial Balance Break Types The green-close rate for each way a session resolved its initial balance — held inside (none), broke one side (single_up, single_down), or broke both (double). — ibBreakType the N column doubles as the break-type frequency table — two classic reports in one result; per-type 95% CI instead of one blended number; condition the whole table on anything: gap state, weekday, event days

Prior levels

Preset Title What it measures Params Group by Beyond a fixed report
prev-high-break-hold Prior High Break and Hold Of the sessions that traded at or above the prior session's high, how often they also CLOSED above it. — — touch, hold, and rejection all come from one stated denominator; compose with open location, gap state, or weekday to isolate the setup you actually trade; N + 95% CI + stability split on every number
prev-high-touch-by-open Prior High Touch by Open Location How often the session reached the prior session's high, given where it opened inside the prior range: by default, opens in the upper half (open position at least 0.5, where 0 is the prior low and 1 the prior high). minOpenPos (number, default 0.5); maxOpenPos (number) dayOfWeek open location is a continuous band, not a fixed upper-half/lower-half toggle; minutes-to-touch distribution attached to every run; per-weekday N and 95% CI on the same cut
prev-low-break-hold Prior Low Break and Hold Of the sessions that traded at or below the prior session's low, how often they also CLOSED below it. — — the mirror of prev-high-break-hold, on identical definitions: asymmetry becomes measurable; any registry condition narrows the denominator explicitly; N + 95% CI + per-year counts on every number
prev-low-touch Prior Low Touch by Open Location How often the session reached the prior session's low, given where it opened inside the prior range: by default, opens in the lower half (open position at most 0.5, where 0 is the prior low and 1 the prior high). maxOpenPos (number, default 0.5); minOpenPos (number) — continuous open-location bands instead of a halves toggle; minutes-to-touch distribution attached to every run; N + 95% CI + stability split on every number
prev-range Prior Session Levels How often the session trades back to the prior session's high (or low), conditioned on where it opened inside the prior range. minOpenPos (number); maxOpenPos (number) — generalizes beyond fixed 'open in upper/lower half' cuts: the open position is a continuous filter; minutes-to-touch distribution included; swap the outcome for touchPrevLow / breakHoldPrevHigh to get the full family from the same registry

Session shape

Preset Title What it measures Params Group by Beyond a fixed report
doji-follow-through Doji Follow-Through Of the sessions that closed as a session-level doji: close within 10% of the session range from the open: how often the NEXT session closed green. — — a stated, reproducible doji definition instead of an eyeballed one; next-session return distribution attached alongside the rate; compose with gap state or weekday to test the pattern in context
engulfing-bear Bearish Engulfing Follow-Through Of the sessions whose red body engulfed the prior green session's body: a session-level bearish engulfing candle: how often the NEXT session closed green. — — the exact mirror of engulfing-bull, on one definition, so the asymmetry is measurable; minimum-N guards and CIs keep rare-pattern statistics honest; compose with trend or streak conditions to test the context the textbooks assume
engulfing-bull Bullish Engulfing Follow-Through Of the sessions whose green body engulfed the prior red session's body: a session-level bullish engulfing candle: how often the NEXT session closed green. — — an explicit engulfing definition (body engulfs prior red body) instead of chart-by-chart judgment; minimum-N guards and CIs keep rare-pattern statistics honest; next-session return distribution attached alongside the rate
inside-day Inside Day Follow-Through Of the sessions whose entire range sat inside the prior session's range, how often the NEXT session closed green. — — next-session open-to-close return distribution attached, so direction and size are both visible; compose with NR4/NR7 or volume conditions to sharpen the compression definition; N + 95% CI + stability split on every number
nr7-expansion NR7 Range vs ATR Among NR7 sessions: the narrowest range of the trailing seven, a flag the session earns itself: how often that same session's range still reached at least r times the prior 14-session ATR. r (number, default 1) — the full range-vs-ATR distribution of NR7 sessions, not just a threshold rate; r is a parameter, so 'compressed' is your definition rather than a preset author's; N + 95% CI + per-year counts on every number
outside-day Outside Day Follow-Through Of the sessions that traded beyond BOTH the prior session's high and low, how often the NEXT session closed green. — — the exact complement design of inside-day, so the two read on one scale; next-session return distribution attached alongside the rate; minimum-N guards refuse to dress up a thin sample

Streaks

Preset Title What it measures Params Group by Beyond a fixed report
streak-continuation Green Streak Continuation After at least n consecutive green sessions, how often the next one closed green as well: the streak extending one more session. n (number, default 3) — one parameterized query replaces a table of fixed streak lengths; the same definition drives reversion and continuation, so the two are directly comparable; N + 95% CI + stability split on every number
streak-reversion Red Streak Reversion After at least n consecutive red sessions, how often the next one closed green. n (number, default 3) — n is a parameter: every streak length is one run, not one report; open-to-close return distribution shows size, not just direction; N + 95% CI + guards make thin long-streak samples visibly thin

Seasonality

Preset Title What it measures Params Group by Beyond a fixed report
day-of-week Green Rate by Weekday The share of sessions closing above their open, split by weekday. — dayOfWeek per-weekday N and 95% CI instead of five bare percentages; return distribution alongside the rate separates frequency from magnitude; condition the whole split on anything: gaps, events, volatility regime
green-rate Baseline Green Rate The unconditional share of sessions that closed above their open: the baseline every conditional preset in this catalog should be compared against. — — the anchor number for every other preset, computed on identical data with identical guards; per-year counts show drift in the baseline itself; open-to-close return distribution attached
month-of-year Green Rate by Month The share of sessions closing above their open, split by calendar month (1-12). — month monthly split over your full history, with N and 95% CI per month; swap the grouping to year to see whether the pattern is stable or a two-year artifact; any condition narrows the split without a new report

Events

Preset Title What it measures Params Group by Beyond a fixed report
day-after-event Day After an Event Of the sessions whose PREVIOUS trading session was an event day in the named calendar, how often the session closed above its open. event (string, default OPEX) — one parameterized preset covers the entire post-event family; holiday-aware adjacency instead of calendar-day arithmetic; open-to-close return distribution attached for the size of the after-move
day-before-event Day Before an Event Of the sessions whose NEXT trading session is an event day in the named calendar, how often the session closed above its open. event (string, default OPEX) — one parameterized preset covers the entire pre-event family; holiday-aware adjacency instead of calendar-day arithmetic; user-supplied event files (earnings, expiries, your own day tags) work identically
event-day-cpi CPI Day Green Rate Of the sessions whose trade date is in the CPI calendar, how often the session closed above its open. — — event days as a first-class filter, usable on every preset in the catalog; exclusion form (NOT eventDay) is the same one keystroke away; N + 95% CI on a sample that is small by nature: stated, not hidden
event-day-fomc FOMC Day Green Rate Of the sessions whose trade date is in the FOMC calendar, how often the session closed above its open. — — the same flag works in reverse. NOT eventDay('FOMC') cleans any other preset of FOMC noise; trading-calendar-aware day-before and day-after siblings; per-year counts show whether the event's behavior drifted
event-day-nfp NFP Day Green Rate Of the sessions whose trade date is in the NFP calendar, how often the session closed above its open. — — one predicate covers the whole event family: swap the name, keep the statistics; compose with gap or range conditions to ask sharper event-day questions; per-year counts expose regime changes around the release
event-day-opex OPEX Day Green Rate Of the sessions falling on monthly options expiration, how often the session closed above its open. — — expiration dates are versioned data with a stated rule, not a hand-kept list; the same flag filters any other preset (gap fill on OPEX, ORB on OPEX, ...); N + 95% CI + per-year counts on every number

Time of day

Preset Title What it measures Params Group by Beyond a fixed report
high-time Time of Session High How often the session's high printed within the first m minutes (default: the first hour). within (duration, default 60m) — the full time-of-high distribution, not a single early-or-late percentage; the cutoff is a parameter: first 15, 30, 60, 90 minutes are one preset; condition on gap or event days to see when the high clock shifts
low-time Time of Session Low How often the session's low printed within the first m minutes (default: the first hour). within (duration, default 60m) — the full time-of-low distribution alongside the threshold rate; a movable cutoff replaces a fixed early/late split; N + 95% CI + stability split on every number

Volatility

Preset Title What it measures Params Group by Beyond a fixed report
quiet-then-wild Narrow Flag, Wide Session Among NR4 sessions: the narrowest range of the trailing four, a flag the session earns itself: how often that same session's range nevertheless reached r times the prior 14-session ATR. r (number, default 1) — makes the relative-vs-absolute volatility disagreement measurable instead of anecdotal; the range-vs-ATR distribution of flagged sessions is attached in full; N + 95% CI + guards on a condition that thins the sample fast
range-vs-atr Range vs ATR How often the session's high-to-low range reached at least r times the prior 14-session average true range. r (number, default 1) — r is continuous: expansion is your threshold, not a preset author's; the ratio distribution shows the whole profile in one run; warm-up sessions are excluded from the denominator, not silently zeroed

Fair value gaps

Preset Title What it measures Params Group by Beyond a fixed report
fvg-below-green Close Direction with an FVG Below Of the sessions that started with an unfilled session-level fair value gap entirely below the open: zones form on prior sessions only, so there is no lookahead: how often the session closed above its open. — — a testable FVG claim with a stated denominator, instead of chart anecdotes; swap the side or the outcome to build the full zone-context family; N + 95% CI + stability split on every number
fvg-magnet Gap Fill Toward an FVG The gap-fill rate when an unfilled session-level fair value gap sat on a chosen side of the open at session start. dir (up/down, default up); side (below/above, default below) — a compositional question: gap state x FVG state: that fixed single-topic reports cannot ask; aligned and opposing configurations are the same preset with flipped parameters; time-to-fill distribution shows whether the zone changed the clock, not just the rate

Total: 42 presets across 11 categories.

Expanded variants: 121. Counted structurally from these files and the registry: each preset counts once, plus one variant per allowed value of each enum parameter, plus one per registry-enumerated value of its default grouping field. Numeric, duration, and string parameters — and the unlimited ad-hoc conditions every preset accepts — multiply the real space far beyond this number, while some counted combinations can be empty on a given dataset; treat it as a size of the named surface, not a claim about your data.

This page is generated by scripts/generate-catalog.ts from presets/ and the registry — do not edit it by hand. Regenerate with pnpm run gen:catalog.