@alphacharts/indicators 0.1.0 → 0.1.1

package/README.md

@@ -1,8 +1,8 @@
 # @alphacharts/indicators
 
-Technical studies for AlphaCharts: a typed study model, an incremental engine over the core's
+Technical studies for alphacharts: a typed study model, an incremental engine over the core's
 columnar data, a Web Worker runner, chart integration through the core plugin API, and
-210 studies (TA-Lib's catalogue plus TradingView's built-ins).
+211 studies (TA-Lib's catalogue plus TradingView's built-ins).
 
 ## Use
 
@@ -34,13 +34,23 @@ Subpaths: `.` (engine, runner, plugin), `./studies` (every study as a named expo
   (O(1) for recursive and running-sum nodes, O(period) for window scans), and a revision recomputes from committed
   state. Warm-up is NaN; absent values never become 0 (except where TA-Lib defines 0, noted in the study's reference).
   setData/prepend on the source recompute fully.
-- **Runner** (`src/runner.ts`): `createRunner(ids)` gives a `WorkerRunner` (arrays transferred, chunks of 65,536
+- **Runner** (`src/runner.ts`): `createRunner(ids, factory)` gives a `WorkerRunner` on the worker `factory` makes (the host's bundler spells the worker URL out; the package never guesses it) (arrays transferred, chunks of 65,536
   bars so a `cancel` is serviced within a chunk; protocol documented in the file header) or the `InlineRunner`
-  when Workers are unavailable. The worker returns every graph column; the main thread adopts them and keeps
+  when no factory is given or the worker would not start. The worker returns every graph column; the main thread adopts them and keeps
   streaming ticks. Studies not built in (defined at runtime) run inline (`canRun`).
 - **Chart integration** (`src/chart/studies.ts`): one core series per drawn plot (pane = the source's for overlays,
   the lowest free pane otherwise), the rest drawn on the overlay (fills, markers, ZigZag, lines shifted back in time
   or joined across gaps; these do not take part in autoscale). Layout JSON is keyed by study instance id.
+- **Outside series** (`src/engine/feed.ts`, `Study.setFeed` / `pushFeedBar`): a study can read bars the host gives it (another
+  instrument, another timeframe). Each is a graph of its own on its own bars; the study's graph reads it through a join by time
+  (`Graph.joined` / `external`): `"same"` is the bar with exactly the study bar's time, else NaN, `"closed"` the latest bar whose
+  open + length is not after the study bar's start (no value contains data from after its bar, and the live bar sees the last finished one).
+  Lookups walk forward from the last answer, so a run and a tick are O(1) amortised; `StudyEngine.pushFeed` re-evaluates the last bar only
+  when it sees a different row. Runners: `RunRequest.feeds`, `StudyRunner.feedBar`.
+- **Bar colours, background shading, per-point line colour, left axis** (`src/chart/studies.ts`): `barcolor` plots colour the source's candles
+  (core `Series.setBarColors`; several studies compose on one source, the latest wins, removal restores exactly), `bgcolor` plots draw through core's
+  `BarShading` primitive (no part in autoscale), a line plot's `colorBy: { col, color }` colours its segments per bar, `StudyDef.scale: "left"`
+  puts the series on the left axis.
 - **WASM** (`src/engine/kernels.ts`, `enableWasm()` from `src/wasm.ts`): optional batch kernels from `@alphacharts/wasm`
   for running sum, variance/stddev, rolling highest/lowest (+ bars) and linear regression, the kernels that won at 1M
   and 10M bars in its benchmark. They are bit-identical to the JS loops (test/wasm.test.ts runs every study both
@@ -49,7 +59,7 @@ Subpaths: `.` (engine, runner, plugin), `./studies` (every study as a named expo
 ## Verification
 
 Run `pnpm -C packages/indicators test`.
-- **TA-Lib** (`test/talib.test.ts`): fixtures generated offline by `scripts/gen_fixtures.py` (TA-Lib 0.8.2 in a
+- **TA-Lib** (`test/talib.test.ts`; `atr` and `natr` are not here, see below): fixtures generated offline by `scripts/gen_fixtures.py` (TA-Lib 0.8.2 in a
   Python venv, see below) from `fixtures/cases/*.json` on a seeded 400-bar OHLCV set (`fixtures/dataset.json`;
   3,000 bars of mixed candle shapes for the patterns). Compared at 1e-9 relative (+1e-12 absolute floor), NaN positions
   must be identical; patterns are integers compared exactly. Case flags: `lead` (our study has valid values earlier
@@ -61,6 +71,7 @@ Run `pnpm -C packages/indicators test`.
 - Runner protocol, worker/inline parity and cancellation: `test/runner.test.ts`. Browser: `bench/tests/indicators.spec.ts`.
 
 Where a study is not TA-Lib's arithmetic it says so in its `reference` field. Known differences:
+`atr`/`natr` (and every ATR band: keltner, supertrend, chande-kroll-stop, median) are Wilder's: bar 0's true range is high - low and the first value is at bar `length - 1`, TA-Lib's skips bar 0 and starts one bar later (hand case and an independent loop in `test/atr.test.ts`; `trange` stays TA-Lib's); `rsi` (and `stochrsi`) report 100 where RS is 0/0, the first `length` changes all zero (TA-Lib: 0; charting packages: 100, `test/rsi-flat.test.ts`);
 MACD/APO/PPO seed like TA-Lib (fast EMA at the slow EMA's first bar), TradingView seeds at bar 0 (gap < 1e-9 after
 ~150 bars); `cmo` defaults to TradingView's window-sum form (`method: "wilder"` is TA-Lib's); the linear regression
 studies are TA-Lib's (checked against numpy polyfit); `ht-trendmode` reports TA-Lib's 0 inside its warm-up;
@@ -69,17 +80,17 @@ Not checked against TradingView itself (no access): the TradingView-only studies
 `reference` field; pivot formulas, Klinger, Schaff, the Relative Volatility form and Fractals tie rules were written
 from memory of TradingView's scripts and are verified only by the hand-computed tests.
 
-## Studies (210)
+## Studies (211)
 
 - **Overlap (TA-Lib + MAs, bands, SAR)** (17): `sma`, `ema`, `wma`, `dema`, `tema`, `trima`, `kama`, `t3`, `ma`, `bbands`, `bbpb`, `bbw`, `midpoint`, `midprice`, `mama`, `sar`, `sarext`
 - **Momentum (TA-Lib)** (30): `rsi`, `cmo`, `mom`, `roc`, `rocp`, `rocr`, `rocr100`, `trix`, `apo`, `ppo`, `macd`, `macdfix`, `macdext`, `stoch`, `stochf`, `stochrsi`, `willr`, `cci`, `bop`, `mfi`, `ultosc`, `aroon`, `aroonosc`, `adx`, `adxr`, `dx`, `plus-di`, `minus-di`, `plus-dm`, `minus-dm`
 - **Volume (TA-Lib)** (3): `ad`, `adosc`, `obv`
-- **Volatility (TA-Lib)** (3): `trange`, `atr`, `natr`
+- **Volatility** (3): `trange` (TA-Lib's), `atr`, `natr` (Wilder's, not TA-Lib's)
 - **Price transforms (TA-Lib)** (4): `avgprice`, `medprice`, `typprice`, `wclprice`
 - **Statistics (TA-Lib)** (9): `linearreg`, `linearreg-slope`, `linearreg-intercept`, `linearreg-angle`, `tsf`, `stddev`, `var`, `correl`, `beta`
 - **Math transforms and operators (TA-Lib)** (26): `acos`, `asin`, `atan`, `ceil`, `cos`, `cosh`, `exp`, `floor`, `ln`, `log10`, `sin`, `sinh`, `sqrt`, `tan`, `tanh`, `add`, `sub`, `mult`, `div`, `max`, `min`, `sum`, `maxindex`, `minindex`, `minmax`, `minmaxindex`
 - **Cycle / Hilbert (TA-Lib)** (6): `ht-dcperiod`, `ht-dcphase`, `ht-phasor`, `ht-sine`, `ht-trendline`, `ht-trendmode`
-- **TradingView-only** (51): `smma`, `vwma`, `hma`, `alma`, `lsma`, `mcginley`, `ma-ribbon`, `ma-cross`, `envelopes`, `keltner`, `donchian`, `price-channel`, `chande-kroll-stop`, `linreg-channel`, `median`, `ichimoku`, `supertrend`, `alligator`, `fractals`, `zigzag`, `fisher`, `kst`, `tsi`, `vortex`, `coppock`, `connors-rsi`, `choppiness`, `rvi-vigor`, `rvi-volatility`, `smi-ergodic`, `rank-corr`, `bull-bear-power`, `awesome`, `accelerator`, `mass-index`, `dpo`, `schaff-tc`, `stoch-tv`, `volume`, `volume-osc`, `pvt`, `net-volume`, `elder-force`, `eom`, `chaikin-mf`, `klinger`, `vwap`, `anchored-vwap`, `pivot-points`, `hist-vol`, `adr`
+- **TradingView-only** (52): `smma`, `vwma`, `hma`, `alma`, `lsma`, `mcginley`, `ma-ribbon`, `ma-cross`, `envelopes`, `keltner`, `donchian`, `price-channel`, `chande-kroll-stop`, `linreg-channel`, `median`, `ichimoku`, `supertrend`, `alligator`, `fractals`, `zigzag`, `fisher`, `kst`, `tsi`, `vortex`, `coppock`, `connors-rsi`, `choppiness`, `rvi-vigor`, `rvi-volatility`, `smi-ergodic`, `rank-corr`, `bull-bear-power`, `awesome`, `accelerator`, `mass-index`, `dpo`, `schaff-tc`, `stoch-tv`, `volume`, `volume-osc`, `pvt`, `net-volume`, `elder-force`, `eom`, `chaikin-mf`, `klinger`, `vwap` (Session VWAP), `anchored-vwap`, `pivot-points`, `hist-vol`, `adr`, `bars-per-year`
 - **Candlestick patterns (TA-Lib CDL*)** (61): `cdl-<name>` for every TA-Lib pattern, e.g. `cdl-2crows`, `cdl-3blackcrows`, `cdl-3inside`, `cdl-3linestrike`, `cdl-3outside`, ...
 
 ## Regenerating the TA-Lib fixtures

package/dist/{all-DeXJJd7D.js → all-_xzUPrL7.js}

@@ -1,21 +1,21 @@
-import { $n as e, M as t, Mn as n, Mr as r, Q as i, ar as a, cn as o, d as s, dr as c, fn as l, hr as u, ni as d, un as f } from "./tv-volume-7hh6ajKt.js";
+import { $ as e, N as t, Nn as n, Nr as r, dn as i, er as a, f as o, fr as s, gr as c, ln as l, or as u, pn as d, ri as f } from "./tv-volume-BYSbfrVk.js";
 //#region src/studies/all.ts
 var p = (e) => typeof e == "object" && !!e && "build" in e && "columns" in e && "plots" in e, m = [
-	d,
+	f,
 	r,
-	u,
 	c,
+	s,
+	u,
 	a,
-	e,
 	n,
-	l,
-	f,
-	o,
+	d,
 	i,
+	l,
+	e,
 	t,
-	s
+	o
 ].flatMap((e) => Object.values(e).filter(p));
 //#endregion
 export { m as t };
 
-//# sourceMappingURL=all-DeXJJd7D.js.map
+//# sourceMappingURL=all-_xzUPrL7.js.map

package/dist/{all-DeXJJd7D.js.map → all-_xzUPrL7.js.map}

@@ -1 +1 @@
-{"version":3,"file":"all-DeXJJd7D.js","names":["overlap","momentum","volume","volatility","price","statistic","math","cycle","sar","patterns","tvTrend","tvOsc","tvVolume","statistic_exports","tv_osc_exports","math_exports","momentum_exports","tv_trend_exports","price_exports","patterns_exports","tv_volume_exports","volatility_exports","cycle_exports","volume_exports","overlap_exports","sar_exports","isStudy","x","ALL_STUDIES","m"],"sources":["../src/studies/all.ts"],"sourcesContent":["// Every built-in study, collected from the group modules. Importing this pulls\n// them all in; import single studies from their group module to stay small.\n\nimport type { StudyDef } from \"../types.js\";\nimport * as overlap from \"./overlap.js\";\nimport * as momentum from \"./momentum.js\";\nimport * as volume from \"./volume.js\";\nimport * as volatility from \"./volatility.js\";\nimport * as price from \"./price.js\";\nimport * as statistic from \"./statistic.js\";\nimport * as math from \"./math.js\";\nimport * as cycle from \"./cycle.js\";\nimport * as sar from \"./sar.js\";\nimport * as patterns from \"./patterns.js\";\nimport * as tvTrend from \"./tv-trend.js\";\nimport * as tvOsc from \"./tv-osc.js\";\nimport * as tvVolume from \"./tv-volume.js\";\n\nconst isStudy = (x: unknown): x is StudyDef => typeof x === \"object\" && x !== null && \"build\" in x && \"columns\" in x && \"plots\" in x;\n\nexport const ALL_STUDIES: readonly StudyDef[] = [overlap, momentum, volume, volatility, price, statistic, math, cycle, sar, patterns, tvTrend, tvOsc, tvVolume].flatMap((m) =>\n  Object.values(m).filter(isStudy),\n);\n"],"mappings":";;AAkBA,IAAM0B,KAAW,MAA8B,OAAOC,KAAM,cAAYA,KAAc,WAAWA,KAAK,aAAaA,KAAK,WAAWA,GAEtHC,IAAmC;CAAC5B;CAASC;CAAUC;CAAQC;CAAYC;CAAOC;CAAWC;CAAMC;CAAOC;CAAKC;CAAUC;CAASC;CAAOC;AAAQ,CAAC,CAAC,SAAS,MACvK,OAAO,OAAOiB,CAAC,CAAC,CAAC,OAAOH,CAAO,CACjC"}
+{"version":3,"file":"all-_xzUPrL7.js","names":["overlap","momentum","volume","volatility","price","statistic","math","cycle","sar","patterns","tvTrend","tvOsc","tvVolume","tv_trend_exports","tv_osc_exports","math_exports","momentum_exports","sar_exports","statistic_exports","tv_volume_exports","volatility_exports","volume_exports","patterns_exports","price_exports","cycle_exports","overlap_exports","isStudy","x","ALL_STUDIES","m"],"sources":["../src/studies/all.ts"],"sourcesContent":["// Every built-in study, collected from the group modules. Importing this pulls\n// them all in; import single studies from their group module to stay small.\n\nimport type { StudyDef } from \"../types.js\";\nimport * as overlap from \"./overlap.js\";\nimport * as momentum from \"./momentum.js\";\nimport * as volume from \"./volume.js\";\nimport * as volatility from \"./volatility.js\";\nimport * as price from \"./price.js\";\nimport * as statistic from \"./statistic.js\";\nimport * as math from \"./math.js\";\nimport * as cycle from \"./cycle.js\";\nimport * as sar from \"./sar.js\";\nimport * as patterns from \"./patterns.js\";\nimport * as tvTrend from \"./tv-trend.js\";\nimport * as tvOsc from \"./tv-osc.js\";\nimport * as tvVolume from \"./tv-volume.js\";\n\nconst isStudy = (x: unknown): x is StudyDef => typeof x === \"object\" && x !== null && \"build\" in x && \"columns\" in x && \"plots\" in x;\n\nexport const ALL_STUDIES: readonly StudyDef[] = [overlap, momentum, volume, volatility, price, statistic, math, cycle, sar, patterns, tvTrend, tvOsc, tvVolume].flatMap((m) =>\n  Object.values(m).filter(isStudy),\n);\n"],"mappings":";;AAkBA,IAAM0B,KAAW,MAA8B,OAAOC,KAAM,cAAYA,KAAc,WAAWA,KAAK,aAAaA,KAAK,WAAWA,GAEtHC,IAAmC;CAAC5B;CAASC;CAAUC;CAAQC;CAAYC;CAAOC;CAAWC;CAAMC;CAAOC;CAAKC;CAAUC;CAASC;CAAOC;AAAQ,CAAC,CAAC,SAAS,MACvK,OAAO,OAAOiB,CAAC,CAAC,CAAC,OAAOH,CAAO,CACjC"}

package/dist/all.js

@@ -1,5 +1,5 @@
-import { r as e } from "./registry-CYZuqhXX.js";
-import { t } from "./all-DeXJJd7D.js";
+import { r as e } from "./registry-DdYYyZN_.js";
+import { t } from "./all-_xzUPrL7.js";
 //#region src/all.ts
 function n() {
 	e(t);