The auto-injected GITHUB_TOKEN defaulted to write-all in ci.yml, bench.yml
and release.yml (no top-level permissions block), and codeql.yml declared
its scopes only at job level. Add a top-level `permissions: contents: read`
to all four so the token starts read-only and only the jobs that genuinely
write through it raise the scope:
- release.yml: github-release keeps contents: write; node-/wasm-publish and
attestations keep their id-token / attestations: write blocks. The
cargo/python/node publish jobs push to crates.io/PyPI/npm via their own
registry secrets, not the GITHUB_TOKEN, so read-only is correct for them.
- codeql.yml: analyze keeps security-events: write (job level).
- ci.yml / bench.yml: no job writes back to the repo (coverage uploads via
CODECOV_TOKEN; bench only uploads an artifact), so no job override is needed.
sync-about.yml already had a top-level block but at contents: write; demote
the top level to read and move contents: write down to the single `sync` job
(the PR-head counter push is the only GITHUB_TOKEN write). The cross-repo
About/docs/webpage/org writes are unaffected — they run through the
fine-grained ABOUT_SYNC_TOKEN, which the permissions key does not govern.
Raises OpenSSF Scorecard Token-Permissions from 0 toward 10.
Minor release. The headline user-facing change is the Node binding now
rejecting invalid indicator periods instead of silently clamping period 0
to 1 (matches Python/WASM/core); plus per-ecosystem binding READMEs and a
corrected MSRV statement in CONTRIBUTING. See CHANGELOG [0.4.0].
- Cargo.toml (workspace.package + wickra-core dep) + Cargo.lock (6 members)
- bindings/python/pyproject.toml
- bindings/node/package.json (version + 6 optionalDependencies) + package-lock.json
- bindings/node/npm/*/package.json (6 platform subpackages)
- examples/node/package-lock.json (wickra-* platform pins)
- CHANGELOG: [Unreleased] -> [0.4.0] - 2026-05-31 + compare URL
No tag pushed (release publish is a separate, user-confirmed step).
Embed the brand banner at the top of the README (linked to wickra.org) and
drop the now-redundant "# Wickra" heading — the banner shows the wordmark.
The image is served from https://wickra.org/og-banner.webp, which the webpage
build regenerates on every deploy from the current indicator count (4K WebP),
so the README banner stays current with no committed binary in this repo and
renders on GitHub, crates.io and docs.rs alike.
Now that the wickra.org catch-all mailbox exists, move the project contact +
package-author email off the personal gmail to support@wickra.org across all
surfaces: CODE_OF_CONDUCT, SECURITY, CITATION.cff, Cargo.toml, the npm + PyPI
author fields, the release.yml npm author, and repo-metadata.toml. (The
package-author changes take effect on the next published release.)
repo-metadata.toml's [audit].forbidden still pins kingchencp@gmail.com (the
private commit email) as a banned substring — unchanged.
Also remove the FUNDING.yml custom "https://wickra.org/sponsor" entry: that
page 404s, so the Sponsor button linked to a dead URL. The GitHub Sponsors
entry (github: [kingchenc]) stays.
* ci(sync-about): auto-sync the published version to wickra-docs on release
* ci(sync-about): retarget indicator-count sync from wiki to wickra-docs + point About homepage at docs.wickra.org (P8.3)
* ci(sync-about): self-update the marketing site (webpage) count + version (P12.1)
* ci(sync-about): auto-sync the published version to wickra-docs on release
* ci(sync-about): retarget indicator-count sync from wiki to wickra-docs + point About homepage at docs.wickra.org (P8.3)
The documentation now lives in the wickra-lib/wickra-docs VitePress repo and
will deploy to docs.wickra.org. Rewrite every tracked, user-facing wiki link
in the main repo to the new canonical site:
- docs/README.md, CONTRIBUTING.md, PULL_REQUEST_TEMPLATE.md
- bindings/{node,python,wasm}/README.md, examples/wasm/README.md
- repo-metadata.toml: wiki_url/wiki_git -> docs_url/docs_git
Page paths map 1:1 to VitePress clean URLs (e.g. /wiki/Quickstart-Rust.md ->
docs.wickra.org/Quickstart-Rust). The sync-about.yml wiki-sync step is left
untouched on purpose: it stays until the docs site is live (tracked as P8.3).
The GitHub wiki itself is not deleted yet for the same reason.
* build(node): track the generated index.d.ts (P5.1)
index.js was committed but index.d.ts was gitignored, an inconsistency that
also contradicts CONTRIBUTING ('regenerate both .d.ts/.js when a binding API
changes'). Track index.d.ts too so the repo carries the TypeScript types as a
matched pair with index.js. Generated by napi build; ~214 indicator classes.
* fix(node): reject invalid periods instead of clamping them (P5.4)
The Node scalar-indicator macro clamped period 0 to 1 (via clamp_period + must)
and the multi-parameter constructors did the same, silently swallowing the
core's PeriodZero validation. The core rejects period 0 (Error::PeriodZero),
and the Python and WASM bindings already propagate it — Node was the outlier,
masking caller mistakes. Make the macro constructor fallible and let every
constructor propagate the core error via map_err, removing clamp_period/must.
Update the smoke test (period 0 now throws, matching core/Python/WASM).
* docs(changelog): note the Node period-validation change (P5.4)
Behavior change per CONTRIBUTING: Node constructors now reject invalid periods
instead of clamping. Add an [Unreleased] Changed entry.
* chore: drop accidentally committed scratch log
The .gitignore comment claimed package-lock.json is committed only under
bindings/node/, but examples/node/package-lock.json has also been tracked
since #80. Correct the comment and add a CONTRIBUTING 'Lockfile policy'
section spelling out every component: Cargo.lock + the two Node package-locks
are tracked; fuzz/Cargo.lock is ignored (cargo-fuzz default); the Python
package has no lockfile by PyO3 convention (pinned via Cargo.lock); the
ghost-ignored site keeps its lockfile local.
* examples(node): add RSI mean-reversion strategy
Node counterpart of strategy_rsi_mean_reversion.{py,rs}: RSI(14) < 30 long,
> 70 exit, 0.1% fees, hourly BTCUSDT. Output verified byte-identical to the
Rust reference (37 trades W24/L13, -17.84% return, 46.89% max drawdown).
* examples(node): add MACD + ADX trend-filter strategy
Node counterpart of strategy_macd_adx.{py,rs}: MACD(12,26,9) histogram
crossover entries gated by ADX(14) > 20, hourly BTCUSDT, 0.1% fees. Output
verified byte-identical to the Rust reference (246 trades W90/L156, -47.19%
return, 53.75% max drawdown).
* examples(node): add Bollinger-squeeze breakout strategy
Node counterpart of strategy_bollinger_squeeze.{py,rs}: enter on a fresh
180-bar Bollinger-bandwidth low + close above the upper band, exit on a
2*ATR(14) stop or upper-band collapse, daily BTCUSDT, 0.1% fees. Output
verified byte-identical to the Rust reference (1 trade, -7.82% return,
13.01% max drawdown).
* examples(wasm): add RSI mean-reversion strategy demo
Browser counterpart of strategy_rsi_mean_reversion.{py,js,rs}: RSI(14) < 30
long, > 70 exit, 0.1% fees, summary table. Same signal/fill/PnL/equity loop as
the runtime-verified Node example; loads via the established wickra_wasm.js
init + fetch-CSV pattern. (wasm32 build runs in CI.)
* examples(wasm): add MACD + ADX trend-filter strategy demo
Browser counterpart of strategy_macd_adx.{py,js,rs}: MACD(12,26,9) histogram
crossover gated by ADX(14) > 20, hourly BTCUSDT, 0.1% fees. Logic identical to
the runtime-verified Node example; standard wickra_wasm.js init + fetch-CSV
loader. (wasm32 build runs in CI.)
* examples(wasm): add Bollinger-squeeze breakout strategy demo
Browser counterpart of strategy_bollinger_squeeze.{py,js,rs}: fresh 180-bar
Bollinger-bandwidth low + upper-band breakout, 2*ATR(14) stop, daily BTCUSDT,
0.1% fees. Logic identical to the runtime-verified Node example; standard
wickra_wasm.js init + fetch-CSV loader. (wasm32 build runs in CI.)
* ci: add examples syntax-smoke job (P2.3)
The Rust examples are built by 'cargo build -p wickra-examples --bins'; the
Node, browser-WASM and Python examples had no build gate. New job parse-checks
every examples/{node,wasm}/*.js, extracts and node --checks each WASM .html
module script, and python -m py_compile's every examples/python/*.py — so a
broken example edit fails CI instead of landing silently.
* docs(examples): list the new Node + WASM strategy examples
Add the three Node strategy scripts and three WASM strategy demos to the
examples README tables, bringing Node and WASM to parity with the existing
Rust and Python strategy rows.
* chore(examples): refresh examples/node lockfile for the linked wickra binding
npm install rewrote the file: dependency snapshot of the local wickra binding
that the examples link against (version 0.1.4 -> 0.3.1, license + engines
fields), which had gone stale in the committed lockfile.
* test(node): add input-validation suite
Node counterpart of bindings/python/tests/test_input_validation.py: invalid
constructor parameters (ATR zero period, MACD non-increasing fast/slow,
BollingerBands negative multiplier, PSAR step > max, ValueArea period/pct,
InitialBalance/OpeningRange zero period, Ichimoku non-increasing periods,
Ehlers-family ordering) and unequal-length candle/ValueArea batch inputs all
throw a JS Error. Validated against the built binding.
* test(node): add indicator completeness contract
Introspects every exported indicator class and asserts the full interface
(update / batch / reset / isReady / warmupPeriod) plus the pre-warmup contract
for zero-arg indicators, and guards that the full catalogue (>= 200 classes)
is exported. Catches a new indicator wired without the standard methods, or a
stale/partial native build dropping exports, with no per-indicator boilerplate.
* test(wasm): broaden scalar streaming-vs-batch coverage
Extend the inline wasm-bindgen-test suite with a streaming==batch check across
~70 scalar indicators spanning moving averages, momentum, volatility,
statistics/regression, Ehlers/cycle and risk/performance families (previously
only EMA + the candle-input group were covered per-indicator), plus four more
invalid-constructor assertions. Constructor args mirror the CI-passing Node
factories. Host-compiles (cargo test -p wickra-wasm --no-run); executed in CI
via wasm-pack test --node.
* bench(node): add indicator throughput benchmark
Node counterpart of the Rust criterion benches / Python compare_libraries:
measures streaming (per-tick update) and batch throughput in Mupd/s across a
representative indicator set over a synthetic OHLCV series (--bars, default
200k). Dependency-free; wired as 'npm run bench'.
* docs(wasm): list strategy demos + document the benchmark story
Add the three new strategy demos to the WASM examples table and a Performance
section: parallel_assets.html is the in-browser benchmark, with raw throughput
covered by the Rust criterion / Python / Node benchmarks (the WASM engine is the
same core compiled to wasm32).
Extend the count-sync workflow to drive three more surfaces from the same
single count, so the org page never drifts again:
- Repo About **homepage** URL — enforced unconditionally via
`gh repo edit --homepage` in the existing About step (fixes the stale
kingchenc URL and self-heals). Uses the same Administration-write
permission as --description, so NO extra token scope.
- Org **profile README** count (wickra-lib/.github, profile/README.md) —
clone + sed + bot commit/push, same pattern as the Sync Wiki step.
- Org **description** field ('… N indicators, install-free.') — gh api PATCH.
The two org steps need ABOUT_SYNC_TOKEN scope the main-repo syncs do not
(write on wickra-lib/.github; admin:org for the description PATCH). Both are
written to soft-skip with a ::warning:: and continue-on-error, so the run
stays green until the scope is granted — then they sync with no further
change. Homepage swap-point to a docs domain is a one-line change.
Refs findings P9 / P10.
* chore: remove ROADMAP.md from the public repo
ROADMAP is kept as a local-only draft (ghost-ignored via
.git/info/exclude); it is not part of the published package surface.
* release: bump 0.3.0 -> 0.3.1
CI-only patch: fixes the release.yml CycloneDX SBOM step (cargo-cyclonedx
has no -p flag, see #79) that skipped the GitHub Release attach-assets job
on 0.3.0. No library changes — republishes the same code with a working
release pipeline.
- Cargo.toml (workspace.package + wickra-core dep) + Cargo.lock
- bindings/python/pyproject.toml
- bindings/node/package.json (version + 6 optionalDependencies) + package-lock.json
- bindings/node/npm/*/package.json (6 platform subpackages)
- CHANGELOG: finalize [0.3.0] (was still under [Unreleased]), add [0.3.1]
* chore: track examples/node/package-lock.json
Since the global package-lock ignore rule was dropped (#68) this file was
left untracked. Commit it for reproducible example installs, consistent
with bindings/node (findings P4.1).
cargo-cyclonedx 0.5.9 walks the whole workspace in a single pass and
writes a <package>.cdx.json next to each member's Cargo.toml; it has no
-p/--package selector. The previous three 'cargo cyclonedx ... -p <crate>'
invocations aborted with 'error: unexpected argument -p found', failing
the cargo-publish job after the crates were already published and, in
turn, skipping the github-release attach-assets job (it needs all four
publish jobs). v0.3.0 published to every registry but got no GitHub
Release page or SBOM assets as a result.
Replace the three invalid calls with a single workspace pass and copy
the three crates.io crate SBOMs into the upload dir.
Minor bump (not patch) because the [Unreleased] section since 0.2.7 has
accumulated a sweep of additive changes that justify a new minor:
- Family 9-16 indicator catalogue expansion (Bands & Channels, Trailing
Stops, Volume, Statistics, Ehlers/Cycle DSP, Pivots, DeMark, Ichimoku,
Candlestick Patterns, Market Profile, Risk/Performance) — roughly
100+ new indicators since 0.2.7 across all four bindings.
- New `wickra_core::FAMILIES` const + family-taxonomy guard tests.
- GitHub org migration (kingchenc -> wickra-lib) and new maintainer
email (wickra.lib@gmail.com).
- New `repo-metadata.toml` + `sync-metadata.yml` audit workflow.
- WASM CI tests now run on every PR (existing tests had been
manually-only).
- CycloneDX SBOMs + npm provenance attestations attached to releases.
- Three end-to-end strategy examples.
- Governance polish: ARCHITECTURE / ROADMAP / CITATION / FUNDING /
.editorconfig.
- Curated benchmark suite (~33 representative indicators).
- Three cold-path coverage fixes (mama, rsi, sine_wave).
- bindings/node/package-lock.json now committed.
Workspace + bindings (Rust crate, Python wheel, Node main + 6 platform
sub-packages, WASM) all step to 0.3.0. CHANGELOG opens the [0.3.0]
section dated 2026-05-28 with the full Changed / Added inventory.
Compare-URL block adds the v0.2.7...v0.3.0 line under [Unreleased] and
points [Unreleased] at v0.3.0...HEAD using the new wickra-lib org.
**Supersedes PR #61** (0.2.7 -> 0.2.8 patch bump). Close#61 when this
one merges. Merge ordering remains: #59 (org migration) + #60
(family-api) + the polish PRs first, rebase this PR on top of the new
main, then merge.
Tag-push `v0.3.0` is a SEPARATE, manual step after merge — it triggers
release.yml's irreversible publish to crates.io / PyPI / npm.
A full `cargo doc --workspace` (and docs.rs, which builds with
`-D rustdoc::all`) emitted five broken intra-doc links. docs.rs treats
these as hard errors, so any 0.2.x doc build was at risk of aborting.
- mama.rs: `[`Fama`]` -> `[`crate::Fama`]` (Fama lives in another module)
- standard_error.rs: `[`crate::Bollinger`]` -> `[`crate::BollingerBands`]`
(the public type is `BollingerBands`, not `Bollinger`)
- aggregator.rs: drop the link to the private `OpenBar::into_candle`,
keep it as plain code text
- resample.rs: drop the link to the private `RolledBar::into_candle`
- csv.rs: `CandleReader::with_timestamp_parser` never existed; reword to
state plainly that ISO/RFC timestamps must be converted to integers
Verified with `RUSTDOCFLAGS="-D rustdoc::broken_intra_doc_links \
-D rustdoc::private_intra_doc_links" cargo doc --workspace --no-deps`:
clean, zero warnings.
Until now `package-lock.json` was globally ignored. Two practical
consequences for the Node binding:
- A fresh `git clone && cd bindings/node && npm install` resolved
`@napi-rs/cli` and any transitive deps to whatever the npm registry
currently considered the latest matching the package.json semver
ranges. Contributors could get different dep graphs on different days.
- No protection against transitive-dep tampering at install time
(lockfile records resolved versions + integrity hashes, npm verifies
on subsequent installs).
Drop the global `package-lock.json` ignore and commit the freshly
generated `bindings/node/package-lock.json` (140 lines, only a couple
of direct deps because the binding is small). The `.gitignore` comment
notes that we still don't expect lockfiles at the workspace root.
No CI workflow changes — the existing `npm install` in the Node-test job
will now consume the committed lockfile, which is exactly what we want.
Two modern supply-chain-trust additions to the release pipeline,
neither of which changes what gets published — only adds verifiable
signals attached to existing releases.
1. **CycloneDX SBOMs.** `cargo-cyclonedx` is installed in the
cargo-publish job after the .crate files are built, and runs once
per published crate (`wickra-core`, `wickra-data`, `wickra`). The
resulting `*.cdx.json` files are uploaded as the `sboms` artifact,
then attached to the GitHub Release alongside the existing wheels,
tarballs, .node binaries and .crate files. Future security advisories
can answer "is my version of crate X transitive in wickra Y.Z?" by
reading the SBOM directly instead of resolving the lockfile.
2. **npm `--provenance` flag** on every npm publish call:
- main `wickra` package (node-publish, first + retry)
- per-platform `wickra-<triple>` subpackages (node-publish loop)
- `wickra-wasm` (wasm-publish)
Provenance attestations are generated server-side by npm from the
GitHub Actions OIDC token. The publishing jobs gain
`permissions: id-token: write` so the runner can exchange that
token. The npm page for each published version will then carry the
"Verified provenance" badge, which proves the tarball was built by
*this* workflow run and not by an arbitrary local laptop with the
NPM_TOKEN.
Skipped deliberately (to keep this PR focused, possibly follow-ups):
- Sigstore cosign signing of artefacts (different audit story; can be
layered on after npm-provenance lands).
- SLSA build-provenance attestations via `actions/attest-build-provenance`
(would target every artefact uniformly; the npm-provenance flag is
the more pragmatic first cut).
YAML structure validated (8 jobs intact). No production code touched.
This PR conflicts with PR #59 only in line-by-line URL substitutions
on release.yml — rebase after #59 should be clean.
Wires real indicators into complete signal -> fill -> PnL -> equity
loops over the checked-in BTCUSDT datasets, with per-trade Sharpe and
max-drawdown reported on stdout. Closes the gap where existing
examples showed only the mechanics of calling `update`/`batch` but not
how Wickra plugs into a trading-system shape.
Three strategies, each in Rust + Python (six files total):
- strategy_rsi_mean_reversion — RSI(14) thresholds (30/70) on 1h
BTCUSDT. Binary position, 0.1% per-trade fee.
- strategy_macd_adx — MACD crossover entries gated by ADX(14) > 20 on
1h BTCUSDT. Trend-follower demo of multi-indicator gating.
- strategy_bollinger_squeeze — Bollinger-bandwidth 180-day-low squeeze
+ upper-band breakout entry, ATR(14) * 2 stop. On 1d BTCUSDT for
interpretable lookback.
Each file is self-contained — print_summary is inlined per script so
the example stays a single-file read. Every script prints a
NOT-financial-advice notice next to its results.
examples/README.md updated to list the new bins/scripts.
* feat(wasm): wire wasm-bindgen-test into CI and expand binding coverage
The WASM binding already had 8 wasm_bindgen_test cases checked in but
they ran only when someone manually invoked `wasm-pack test` locally —
CI never executed them, so a breakage between the Rust core API and the
JS surface would only surface in user reports.
Two changes:
1. **New CI job `wasm-test`** in `.github/workflows/ci.yml` runs
`wasm-pack test --node bindings/wasm` on every push and pull-request.
Uses Node-runtime mode (no headless browser needed), pins
`taiki-e/install-action` for `wasm-pack`, installs the
`wasm32-unknown-unknown` target.
2. **10 new `#[wasm_bindgen_test]` cases** in `bindings/wasm/src/lib.rs`
covering families that the existing 8 tests did not touch:
- Family 4 — Macd multi-output batch shape (3-component packing)
- Family 5 — Bollinger {upper, middle, lower} ordering invariant
- Family 10 — FisherTransform streaming roundtrip (recursive DSP)
- Family 16 — SharpeRatio reset semantics, MaxDrawdown monotone-
uptrend invariant, ValueAtRisk constructor validation
- Cross-family — reset returns indicator to warmup (3 shapes),
warmup_period parity with wickra-core, NaN input rejection
does not advance warmup counter
No production code changed; additive tests only. Total wasm test count
goes from 8 to 18.
* fix(wasm-tests): WasmAtr's hand-coded wrapper has no is_ready accessor
The new `reset_returns_indicator_to_warmup` test in this branch called
`atr.is_ready()` but `WasmAtr` is hand-coded (not generated by the
`wasm_scalar_indicator!` macro) and does not expose `is_ready` on the
JS surface — that pattern is reserved for macro-generated scalar
wrappers.
Replace the second leg of the test with `WasmEma` so the test exercises
two macro-generated scalars whose `is_ready`/`reset` semantics are
guaranteed by the same code path. The candle-input lifecycle is
already covered by `candle_input_streaming_matches_batch_and_lifecycle`.
Workspace clippy locally green:
- cargo clippy -p wickra-wasm --all-targets -- -D warnings
- cargo check -p wickra-wasm --target wasm32-unknown-unknown --tests
* fix(wasm-tests): Bollinger batch layout is 4 floats per bar, not 3
The new `bollinger_batch_orders_upper_mid_lower` test indexed the batch
output as `[u0, m0, l0, u1, m1, l1, ...]` (3 floats per bar) but the
actual WasmBb::batch layout is `[u0, m0, l0, sd0, u1, m1, l1, sd1, ...]`
— four floats per bar including stddev. The assertion read an upper
band against a middle from the previous bar, which can violate the
expected upper >= middle ordering and made the test fail intermittently.
Switch the stride from `* 3` to `* 4` so we read upper / middle / lower
from the same bar. Also add an `is_finite()` precondition so the test
fails with a clear "warmup positions unexpectedly NaN" message rather
than a confusing NaN-comparison if the dataset ever shrinks.
* fix(wasm-tests): drop redundant wasm-test job; existing WASM build covers it
The CI workflow already had a `WASM build` job that runs
`wasm-pack test --node bindings/wasm` after building. The
`wasm-test` job I added in cc961b3 duplicated that step and was
the only one whose `taiki-e/install-action` SHA was wrong, so it
also kept failing on a non-resolvable action reference.
Removing the duplicate keeps the test coverage (it lives in the
existing WASM build job) and gets rid of the unresolvable SHA.
Previously every push of `cargo bench -p wickra` ran 114 indicators at
three workload sizes each (1k / 10k / 50k candles), which inflated bench
runtime past ten minutes for diminishing signal — most family members
are linear scalings of the same hot loop, so a regression in any one of
them shows up identically in the cheapest member.
This commit replaces the exhaustive list with a curated selection: the
cheapest baseline and the most expensive representative from each of
the sixteen families, totalling ~33 indicators across scalar, candle,
and multi-output APIs.
If you need to profile a specific indicator that is not in the curated
set, add it temporarily and run `cargo bench -- <name>` to target just
that bench; it does not need to be committed.
Fixed in passing:
- `Cci` is `Indicator<Input = Candle>`, not f64; corrected the bench
selector to `bench_candle_input`.
- `Psar::new` takes (af_start, af_step, af_max) — supplied all three.
- `TdSequential` uses `::classic` for the textbook (4, 9, 2, 13)
parameters; the old call was missing arguments.
Five governance + onboarding files that collectively close the
"no high-level documentation outside of README" gap.
- ARCHITECTURE.md: workspace layout, Indicator trait contract,
per-indicator file conventions, numerical-stability notes,
cross-crate flow diagram, navigation cheatsheet, deliberate
non-goals, performance characteristics, stability commitments.
- ROADMAP.md: north star, 0.3 / 0.4 / 0.5 release windows,
indicator-wishlist intake rules, explicit non-goals, versioning policy.
- CITATION.cff: GitHub-renders as "Cite this repository" button;
enables academic adoption.
- .github/FUNDING.yml: surfaces Sponsor button on the repo page.
- .editorconfig: normalises indent/EOL across IDEs (Rust/Python 4 spaces,
JS/TS/JSON/YAML/MD 2 spaces, Makefile tabs).
Touches no Rust source, no CI workflows, no behaviour. Additive only.
URLs in ROADMAP/CITATION already reference the post-transfer
`wickra-lib/wickra` org so the files stay valid once the org migration
in PR #59 lands; merge this PR only after #59 to keep main consistent.
* fix(bindings): clear clippy pedantic lints and lint bindings in CI
Resolve the pedantic lints that only surfaced under a full
`cargo clippy --workspace` (the CI clippy job covered only the core
crates, so the Python/Node bindings drifted):
- manual_midpoint: `(a + b) / 2.0` -> `f64::midpoint(a, b)` (node + python)
- new_without_default: add `Default` impls for the six no-arg Node nodes
- type_complexity: factor the pivot/Ichimoku return tuples into
`PivotLevels` / `WoodieLevels` / `IchimokuLines` aliases (python)
- many_single_char_names: allow at crate level — OHLCV batch helpers bind
the conventional o/h/l/c/v column names
Add a dedicated `clippy-bindings` CI job (ubuntu-only, with Python + Node
toolchains) so future binding lints fail CI instead of slipping through.
* ci: lint Python and Node bindings in a dedicated clippy job
The main `rust` job's clippy step only covers wickra-core/wickra/
wickra-data/wickra-wasm, so pedantic lints in the PyO3/napi bindings
slipped through. Add an ubuntu-only `clippy-bindings` job that
provisions Python + Node (needed by the build scripts) and runs
`cargo clippy -p wickra-node -p wickra-python --all-targets -- -D warnings`.
* feat(family-api): add FAMILIES const and family-taxonomy tests
Introduce `wickra_core::FAMILIES`, a `&'static [(&str, &[&str])]` mapping
every built-in indicator to one of 16 families (Moving Averages, Momentum
Oscillators, Trend & Directional, Price Oscillators, Volatility & Bands,
Bands & Channels, Trailing Stops, Volume, Price Statistics, Ehlers /
Cycle (DSP), Pivots & S/R, DeMark, Ichimoku & Charts, Candlestick
Patterns, Market Profile, Risk / Performance).
Two compile-time-anchored guards make sure the const stays trustworthy:
- `no_duplicates_across_families`: no indicator is listed in two families
- `total_count_matches_expected`: hard-coded total bumps in lockstep with
the indicator catalogue (214 today), so any new indicator added without
being filed under a family trips the test.
Also corrects the module doc comment which falsely claimed
indicators were grouped by category internally; the `mod` block is
alphabetical, and the canonical taxonomy now lives in `FAMILIES`.
* chore: sync indicator count to 214
---------
Co-authored-by: wickra-bot <wickra-bot@users.noreply.github.com>
Introduce repo-metadata.toml as single source of truth for repo identity
(org slug, maintainer email, canonical URLs) and add sync-metadata.yml
workflow with a Python audit script that fails CI if any tracked file
drifts back to pre-migration values.
Bulk-replace across 24 tracked files:
- kingchenc/wickra -> wickra-lib/wickra (URL segment)
- kingchencp@gmail.com -> wickra.lib@gmail.com (maintainer email)
- @kingchenc -> @wickra-lib (CODEOWNERS mention only)
Person-name credits are preserved: LICENSE copyright holder, Cargo.toml
authors handle, and CHANGELOG historical @kingchenc reference all remain
unchanged. Crate / PyPI / npm package names also untouched.
Merge this PR only after the kingchenc/wickra -> wickra-lib/wickra org
transfer has happened on the GitHub side, otherwise all badges and
repository links 404 until the transfer is performed.
* chore(sync-about): count public indicator types, not module files
The sync-about workflow counted `mod xxx;` lines in
crates/wickra-core/src/indicators/mod.rs to derive the indicator count
that gets propagated to README, the GitHub About description, and the
wiki. That under-reported by one because `vwap.rs` exports two public
types — `Vwap` (cumulative) and `RollingVwap` (finite window) — from
the same module. The bindings reach both, so users see 214 indicators
even though there are only 213 source files.
Fix the count by parsing the canonical `pub use indicators::{ ... }`
block in lib.rs, dropping the `FAMILIES` constant and any `*Output`
companion structs, and counting the remaining public types. Pure-shell
implementation so the workflow doesn't grow a python dependency.
Sync README to the corrected count (213 -> 214) in the same commit so
the PR validates cleanly through the workflow's PR-flow.
* docs(bindings): sync per-binding READMEs and docs/ pointer to 214 / 16
The bindings/{node,python,wasm}/README.md files (which ship to
npm / PyPI / npm again) and docs/README.md (the pointer to the
wiki) all carried the stale '71 streaming-first indicators across
eight families' header from before the family expansion. Pull the
canonical 16-family table out of the main README into all three
binding READMEs and update docs/README.md to list every family by
name, so a user landing on PyPI / npm sees the current catalogue
shape.
* feat(family-15): add 17 risk/performance metrics
Implements Family 15 pragmatically as standard `Indicator`s instead of a
separate `wickra-metrics` crate. Input is scalar `f64` per bar — period
return, equity sample, or per-trade P&L depending on the metric.
Scalar `Indicator<f64>` (14):
- SharpeRatio(period, risk_free)
- SortinoRatio(period, mar)
- CalmarRatio(period)
- OmegaRatio(period, threshold)
- MaxDrawdown(period) — rolling, peak-to-trough
- AverageDrawdown(period)
- DrawdownDuration — cumulative, bars under water (u32 output)
- PainIndex(period)
- ValueAtRisk(period, confidence)
- ConditionalValueAtRisk(period, confidence)
- ProfitFactor(period)
- GainLossRatio(period)
- RecoveryFactor — cumulative, net return / max drawdown
- KellyCriterion(period)
Two-series `Indicator<(f64, f64)>` for (asset, benchmark) returns (3):
- TreynorRatio(period, risk_free)
- InformationRatio(period)
- Alpha(period, risk_free) — Jensen / CAPM
Touchpoints:
- 17 new files under `crates/wickra-core/src/indicators/`.
- `mod.rs` + `lib.rs` re-exports.
- Python bindings (`bindings/python/src/lib.rs`, `__init__.py`).
- Node bindings (`bindings/node/src/lib.rs`, `index.js`).
- WASM bindings (`bindings/wasm/src/lib.rs`).
- Fuzz: scalar metrics appended to `indicator_update.rs`; new
`indicator_update_pair.rs` fuzz target for `(f64, f64)` indicators.
- Python tests: SCALAR + new PAIR parameter lists in `test_new_indicators.py`,
reference-value cases in `test_known_values.py`.
- Node tests: scalar factories + new pair-factory block in
`bindings/node/__tests__/indicators.test.js`.
- Benches: 5 Family-15 benches added in `crates/wickra/benches/indicators.rs`.
- Docs: README family-table row + counter (71 -> 88), CHANGELOG entry under
[Unreleased].
Note: Family 12 (statistik-regression, PR #51) introduces
`node_pair_indicator!` and `wasm_pair_indicator!` macros for Pearson /
Beta / Spearman. Family 15 needs the same pair-input pattern but Family 12
is not yet in main, so the three pair wrappers below are written by hand
in this PR. When PR #51 lands, the trivial merge-conflict is resolved by
keeping the macros from Family 12 and re-using them for Treynor / IR /
Alpha (drop the three handwritten wrappers).
cargo check --workspace --all-features: green.
* fix(family-15): satisfy clippy doc_markdown / if_not_else / digit_grouping
* fix(family-15): unused TreynorRatio import, duplicate pairFactories, _eq_nan inf handling
* fix(family-15): node eq() handles matching infinities for ratio indicators
* test(family-15): cover cold paths flagged by codecov patch
* feat(family-16): add ValueArea + InitialBalance + OpeningRange
Opens family #16 (Market Profile) with the three OHLCV-compatible scalar /
multi-output indicators:
- ValueArea(period, bin_count, value_area_pct) -> {poc, vah, val}.
Rolling bin-approximation volume profile over the last `period`
candles. Each candle's volume is spread uniformly across [low, high];
POC is the bin with highest cumulative volume; the value area expands
symmetrically from POC and always absorbs the higher-volume neighbour
next, until `value_area_pct` (default 0.70) of total volume is
enclosed. Defaults (20, 50, 0.70).
- InitialBalance(period) -> {high, low}. Tracks session-opening high
and low over the first `period` bars, then locks. Default period = 12
(one-hour IB on 5-minute bars for US equities). Callers MUST invoke
reset() at every session boundary, otherwise IB stays fixed for the
lifetime of the instance.
- OpeningRange(period) -> {high, low, breakout_distance}. Same
lock-after-N-bars semantics as IB with a shorter default period
(6 = 30 min on 5-minute bars) and a third output that tracks
close - or_mid (positive above the range mid, negative below).
Histogram-output Market Profile variants (Volume Profile, VPVR,
Composite Profile) are deferred because they need a new histogram
output API layer rather than fixed-arity scalars. Tick-data-only
variants (TPO Profile, Single Print, Order Flow Delta, Cumulative
Delta, Volume-Weighted Open) are out of scope because `wickra-data`
does not currently expose tick / L2 data.
All four bindings (Rust core, Python, Node, WASM) ship the new
indicators with parity tests; benches added; fuzz target extended.
Counter 71 -> 74 across 8 -> 9 families. cargo check --workspace
--all-features green.
* fix(family-16): cover cold paths in InitialBalance + ValueArea
InitialBalance::value() public getter had no test covering the post-update
Some(...) branch — extended accessors_and_metadata to call value() after one
update. ValueArea single-print bar path (c.high == c.low) was unreachable in
existing tests since the only single-print test used a uniform 100-price
window which exits early via the span == 0 guard; added a mixed-window test
that triggers the c.high <= c.low branch directly. The (None, None) arm of
the expansion match was by-construction unreachable (the loop condition
already requires at least one neighbour) and has been folded into an
if/else.
* feat(family-12): add 13 Statistik/Regression indicators
Brings the Price Statistics family to 20 indicators (7 → 20) and the
total catalogue to 84 (71 → 84). Every indicator ships in the Rust
core plus Python, Node, and WASM bindings with full streaming ↔ batch
parity, fuzz coverage, and benches.
Scalar (f64 → f64):
- Variance, CoefficientOfVariation: rolling population variance and
its dimensionless ratio with the mean. O(1) updates.
- Skewness, Kurtosis: rolling Pearson skewness and excess kurtosis,
derived from running sums of x, x², x³, x⁴ via the binomial
identities — also O(1) per bar.
- StandardError, DetrendedStdDev: standard error of estimate (n − 2)
and population StdDev (n) of OLS residuals, sharing the LinReg
O(1) sliding sums.
- RSquared: coefficient of determination of the rolling OLS fit; the
trend-quality filter, clamped to [0, 1].
- MedianAbsoluteDeviation: robust dispersion estimator; O(period log
period) per emission via two in-place sorts of a reusable scratch
buffer.
- Autocorrelation(period, lag): rolling lag-k Pearson autocorrelation.
- HurstExponent(period, chunks): R/S-analysis trend-persistence
estimator clamped to [0, 1].
Pair indicators (Input = (f64, f64)):
- PearsonCorrelation: rolling cross-series Pearson, O(1).
- Beta: rolling OLS slope of asset vs. benchmark (CAPM).
- SpearmanCorrelation: rolling rank correlation with mid-rank tie
handling; O(period log period).
Touchpoints:
- crates/wickra-core: 13 new indicator modules + mod.rs / lib.rs
re-exports.
- bindings/python: pyclasses + add_class registration + __init__.py
import & __all__ updates. The pair indicators expose
update(x, y) and batch(x, y) over two equally-sized numpy arrays.
- bindings/node: scalar indicators via node_scalar_indicator! macro;
pair indicators via new node_pair_indicator! macro; explicit
structs for Autocorrelation and HurstExponent (two-arg ctors).
index.js extended with the new exports.
- bindings/wasm: scalar wrappers via wasm_scalar_indicator!; pair
wrappers via new wasm_pair_indicator! macro.
- fuzz: every scalar drove through the generic helper; pair
indicators stress-tested by pairing adjacent samples of the fuzz
input.
- Python tests (test_new_indicators.py): added to SCALAR
parametrisation, plus algebraic reference values
(variance of [2,4,6] = 8/3, MAD ignoring outlier = 0, monotone
non-linear Spearman = 1, two-to-one Beta = 2, etc.) and a
streaming-vs-batch test for the pair indicators.
- Node tests (indicators.test.js): extended the scalar factories
map and added a pair-indicator section with the same algebraic
reference values.
- crates/wickra/benches: bench_scalar entries for all 10 single-
input new indicators.
- README: counter 71 → 84; Price Statistics family-table row
expanded with the 13 new indicators.
- CHANGELOG: Unreleased section documents the family addition.
Wiki drafts (ghost-ignored, manual sync to wickra.wiki at release
time): indicator-ideas/families/wiki/family-12-statistik-regression/
contains 13 deep-dive pages plus _Sidebar / Indicators-Overview /
Warmup-Periods / Home fragments for the curator merge.
cargo check --workspace --all-features: clean.
* fix(family-12): remove unreachable defensive guards in hurst_exponent
The three guards (m < 2 continue, end > buf.len() break, denom == 0.0
return) are by-construction unreachable given the constructor invariant
period >= 2 * chunks: m = period / k for k in 1..=chunks always
satisfies m >= 2 and end = (c+1) * m <= k * m <= period = buf.len(),
and m_1 = period and m_2 = period / 2 are always distinct so the slope
denominator is strictly positive. Removing them brings codecov/patch
back to 100%.
Two new indicators in a brand-new "Ichimoku & alternative charts"
family:
- `Ichimoku` (Ichimoku Kinko Hyo): the full five-line cloud system
(Tenkan-sen, Kijun-sen, Senkou Span A/B, Chikou Span). Classic
(9, 26, 52, 26) defaults; configurable. Forward displacement is
handled in an O(1) ring buffer so the visible Senkou A/B at bar n
are the values computed at bar n-displacement.
- `HeikinAshi`: recursive candle smoothing transform emitting a
four-field synthetic candle. Seeds ha_open from (open+close)/2 on
the first bar.
Touchpoints: core + unit tests, mod.rs/lib.rs re-exports, Python +
Node + WASM bindings (multi-output via PyArray2 / interleaved Vec<f64>
/ Object+Float64Array), Python tests across smoke/new-indicators/
input-validation, Node parity tests, fuzz target (Candle), benches,
README family table + counter (71 -> 73, 8 -> 9 families), CHANGELOG.
Note: Renko, Kagi, and Point & Figure from the family-13 ideas list
are intentionally skipped. They are bar generators (the bar boundary
is defined by price moves, not by a fixed time interval) rather than
indicators that consume a candle stream, and belong in wickra-data
as candle/tick transforms alongside the existing tick-to-candle
aggregator and resampler.
Add tests that exercise the protective fallbacks reachable via flat or
zero-valued input:
- `CenterOfGravity::zero_window_uses_zero_fallback` — den == 0 branch.
- `EhlersStochastic::flat_window_emits_zero` — range == 0 branch.
- `Mama::flat_input_uses_phase_fallback` and the matching
`SineWave` variant — `i1` collapses to zero on a constant series.
- `Fama::new_with_valid_limits_constructs_via_mama` exercises the
`Ok(Self { inner: Mama::new(..)? })` arm that no other test reaches.
Three branches were genuinely unreachable by construction, so the dead
code is removed rather than masked with an attribute:
- `Mama` clamped `alpha > fast_limit` after the lower-bound clamp; the
upper bound is implied by `delta_phase >= 1` and `alpha = fast / delta_phase`.
- `CyberneticCycle` had a `0.0` fallback after the warmup gate that the
3-slot ring buffers preclude (`count >= 7` => all five `Some`s).
- `DecyclerOscillator` used a `let-else { return None }` over a pair of
`Decycler::update` calls that always emit `Some` from the first bar.
Implements Family 10 (Ehlers / Cycle) end-to-end across Rust core,
Python / Node / WASM bindings, fuzz, tests, benches and docs. This
is an entirely new family covering John Ehlers' digital-signal-
processing school of cycle analytics — a strong differentiator
versus TA-Lib and pandas-ta, which ship only fragments.
Indicators:
- MAMA (Mesa Adaptive MA) — multi-output { mama, fama }
- FAMA (Following Adaptive MA) — scalar wrapper around MAMA's slow line
- Fisher Transform — Gaussian-normalising price transform
- Inverse Fisher Transform — bounded oscillator (tanh-based)
- SuperSmoother — 2-pole Butterworth lowpass
- Roofing Filter — high-pass + SuperSmoother bandpass
- Decycler — price minus 2-pole high-pass (lag-free trend)
- Decycler Oscillator — fast / slow Decycler difference (MACD-like)
- Hilbert Dominant Cycle — phase-derived period estimator [6, 50]
- Sine Wave Indicator — sin(phase) with 45° lead companion
- Adaptive Cycle Indicator — half-period driver for adaptive oscillators
- Center of Gravity Oscillator — weighted-mass momentum
- Cybernetic Cycle Component — EasyLanguage classic
- Empirical Mode Decomposition — bandpass + envelope mean
- Ehlers Stochastic — Stochastic on Roofing Filter input, [-1, +1]
- Instantaneous Trendline — Ehlers 2-pole lag-free trend
Indicator count rises 71 -> 87 across nine families (was eight).
All sixteen pass batch == streaming equivalence, expose the standard
Indicator surface (update / batch / reset / is_ready / warmup_period
/ name), are fuzz-tested, benchmarked against the checked-in BTCUSDT
1-minute dataset and reach across all four bindings.
Wiki deep-dive drafts for every indicator + Sidebar / Overview /
Home / Warmup updates are staged under indicator-ideas/families/
wiki/family-10-ehlers-cycle/ in the main repo (ghost-ignored) for
the maintainer to publish to the wiki repo manually.
* feat(family-11): add DeMark suite (TD Setup, Sequential, DeMarker, REI, Pressure)
Family 11 (DeMark) was previously empty; this PR adds five
streaming-first DeMark indicators in one batch.
- **TD Setup** (`TdSetup`): parameterised buy/sell setup counter.
Counts consecutive bars whose close is less-than (buy) or
greater-than (sell) the close `lookback` bars earlier, saturating
at `target`. Emits a signed `f64` so callers read direction from
the sign and run length from the magnitude. Classic config:
`lookback = 4`, `target = 9`.
- **TD Sequential** (`TdSequential`): the canonical Setup + Countdown
exhaustion pattern. Output struct `{ setup, countdown, direction }`
exposes both phase counts as signed numbers plus the active
countdown direction (+1 buy / -1 sell / 0 none). Countdown
activates when a setup completes and tracks the close-vs-high/low
comparison `countdown_lookback` bars back, capped at
`countdown_target`. Classic: 4/9/2/13.
- **TD DeMarker** (`TdDeMarker`): bounded [0, 1] oscillator from the
rolling average of upward high expansion (DeMax) and downward low
expansion (DeMin). Falls back to the neutral 0.5 on a flat market
(denominator zero).
- **TD REI** (`TdRei`): Range Expansion Index, bounded [-100, 100].
Per-bar numerator gated on a range-overlap condition vs the bars
5 and 6 back, normalised by a `period`-bar sum of absolute moves.
Classic period = 5. Saturates at +100 in a slow steady uptrend
and at -100 in the mirror downtrend; emits 0 on a flat market.
- **TD Pressure** (`TdPressure`): volume-weighted buying / selling
pressure normalised to [-100, 100]. Per-bar pressure is the
intra-bar close-vs-open ratio scaled by volume; the output is the
rolling mean divided by the rolling mean volume. Zero-range bars
contribute zero (avoid the undefined ratio) and a flat zero-volume
window falls back to 0.
Bindings: all five exposed in Python (`ta.TDSetup`, `ta.TDSequential`,
`ta.TDDeMarker`, `ta.TDREI`, `ta.TDPressure`), Node (`wickra.TDSetup`
etc.), and WASM. Multi-output classes (`TDSequential`) return either
a struct `{ setup, countdown, direction }` per bar (streaming) or a
flat interleaved Float64Array of length `3 * n` (batch).
Tests: 47 unit tests across the five new core files (pure-trend
saturation, flat-market neutral fallback, batch-equals-streaming,
zero-parameter rejection, reset semantics, accessors). Python
test_new_indicators.py picks up all five plus a multi-output TD
Sequential block. Node indicators.test.js picks up all five.
Reference values added to test_known_values.py.
Fuzz: candle fuzz target sweeps all five DeMark indicators with the
existing `Vec<f64>` -> `Vec<Candle>` driver.
Benches: BTCUSDT 1-minute dataset benches for each DeMark indicator
in `crates/wickra/benches/indicators.rs`.
Docs: README family table gains a "DeMark" row; indicator counter
bumped 71 -> 76. CHANGELOG entry added under [Unreleased]. Wiki
drafts (deep-dive pages + Sidebar / Overview / Warmup-Periods / Home
deltas) live under `indicator-ideas/families/wiki/family-11-demark/`
for manual merge into the wiki repo.
* feat(family-11): add 7 missing DeMark indicators
Complete the DeMark suite (family 11) with the seven indicators not
covered by the first commit: TD Combo, TD Countdown, TD Lines (TDST),
TD Range Projection, TD Differential, TD Open, and TD Risk Level.
- TdCombo: aggressive countdown variant with three strictness rules
on top of the classic close-vs-low/high lookback rule (monotone
low/high, monotone close vs prior bar).
- TdCountdown: standalone 13-bar countdown packaging only the signed
countdown count (the setup machine runs internally).
- TdLines: TDST horizontal support/resistance levels from the
highest-high / lowest-low bars of the most-recently-completed
setup, exposed as a multi-output struct.
- TdRangeProjection: DeMark X-projection of the next bar's high and
low from the current bar's OHLC via an open-vs-close-weighted
pivot (three branches: close<open, close>open, close==open).
- TdDifferential: two-bar buying-pressure vs selling-pressure
reversal pattern emitting +1/-1/0.
- TdOpen: gap-and-fade reversal pattern (open outside prior range
with subsequent recovery into it) emitting +1/-1/0.
- TdRiskLevel: protective stop levels derived from the setup
extreme bar +/- its true range.
All seven are wired through Rust core, Python, Node and WASM
bindings, registered in the candle-stream fuzz target, given
benchmark entries on the BTCUSDT 1-minute dataset, and covered by
streaming-vs-batch equivalence, reference-value, lifecycle and
input-validation tests on the Python and Node sides. README counter
moves 76 -> 83 and the CHANGELOG "family 11" entry is extended to
list all twelve indicators.
* fix(td_risk_level tests): check first emission at idx 12, not last bar
TdRiskLevel re-ratchets the sell-risk level on each subsequent setup
completion, so a strictly rising series produces 22.0 at idx 19 (latest
setup) rather than 15.0 (first setup). The test comment already named
idx 12 as the reference; switch the assertion from out[-1] to out[12]
to match the reference computation.
* test(family-11): cover buy-direction branches in TD indicators
Add downtrend tests to TdSequential, TdCombo and TdCountdown so the
buy-side countdown/combo increment branches are exercised; remove an
empty `if buy_countdown == target {}` block in TdSequential whose
behavior is already enforced by the outer strict `<` guard.
Closes codecov/patch gaps reported on PR #48 (10 missed lines across
the three files).
* feat(family-08): add Classic, Fibonacci, Camarilla, Woodie and DeMark pivots + Williams Fractals + ZigZag
Seven new indicators land the previously empty Pivots & S/R family
(family 08), each implemented in wickra-core with the full Indicator
trait surface (update / reset / warmup_period / is_ready / name),
exposed across Python (PyO3), Node (napi-rs) and WASM (wasm-bindgen)
with the standard streaming + batch APIs, and covered by Rust unit
tests, Python streaming-vs-batch + reference-value tests, Node
streaming-vs-batch tests, the candle-input fuzz target and Rust
microbenchmarks.
- ClassicPivots (7 levels): PP = (H+L+C)/3, three R/S tiers per the
floor-trader formulas.
- FibonacciPivots (7 levels): PP plus R/S spaced by 0.382 / 0.618 /
1.000 of the prior range.
- Camarilla (9 levels): Nick Stott's four-tier `C +/- (H - L) * 1.1 /
{12, 6, 4, 2}` levels.
- WoodiePivots (5 levels): close-weighted PP = (H + L + 2*C) / 4 plus
two R/S tiers.
- DemarkPivots (3 levels): conditional X sum based on the previous
bar's open-vs-close relationship.
- WilliamsFractals: five-bar swing detector emitting optional up/down
fractal prices at the centre of each window.
- ZigZag: percent-threshold swing tracker, non-repainting; emits the
just-completed extreme and direction on confirmed reversals only.
README family table updated to nine families / 78 indicators;
CHANGELOG records the family-08 addition under [Unreleased].
* fix(family-08 tests): unify MULTI dict to 3-tuple (factory, batch_call, k)
The HEAD-side family-08 test parametrised MULTI[name] as
`(factory, batch_call, output_arity)` so that pivots with arity 3/5/7/9
fit the same harness. Main's entries arrived as 2-tuples; convert them
all to the 3-tuple shape so `make, batch_call, k = MULTI[name]` unpacks
cleanly. Lifecycle test now indexes the tuple instead of destructuring.
* test(zig_zag): tighten flat-oscillation test (drop dead counter branch)
The previous version of `small_oscillations_yield_no_swings` counted
emitted swings, but the assertion proves the counter never increments
so codecov flagged `emitted += 1` as uncovered. Switch to a per-bar
`assert!(...is_none())` — same coverage of the no-swing path, no dead
branch.
* feat(family-09): add 7 trailing stops (HiLo, Volty, Yo-Yo, Donchian, Pct, Step, Renko)
Rounds out the Trailing Stops family from 5 to 12 indicators:
- HiLoActivator (Crabel): SMA-of-high/SMA-of-low trail with a one-bar
lag; emits the opposite-side SMA as the trailing stop.
- VoltyStop (Cynthia Kase): ATR trail anchored on the extreme close
since the trade was opened — tighter than AtrTrailingStop on
pullbacks.
- YoyoExit: long-only ATR trail with an explicit re-entry trigger at
trail + multiplier*ATR; exposes an in_trade flag.
- DonchianStop (Turtle): lowest low / highest high over the window;
multi-output {stop_long, stop_short}.
- PercentageTrailingStop: fixed-percent trail that scales across
instruments without per-asset tuning.
- StepTrailingStop: snaps to a step_size-aligned grid; mirrors
discretionary stop-by-hand workflow.
- RenkoTrailingStop: block-anchored trail; only moves on full-block
advances, ignores intra-block noise.
All seven are wired into wickra-core, the Python / Node / WASM
bindings, the indicator_update + indicator_update_candle fuzz targets,
the wickra bench harness, and the Python + Node test suites. README
counter bumps from 71 to 78; CHANGELOG entry under [Unreleased].
* fix(family-09): satisfy pedantic clippy lints
- hilo_activator: rewrite match-Some/None as if-let-else (single_match_else),
add backticks around the HiLo identifier in module/struct doc (doc_markdown).
- percentage / step / renko trailing stop tests: use f64::from(i32) instead
of `as f64` (cast_lossless).
- bench `benches()` is now >100 lines after Family 09 was wired in; allow
too_many_lines (matches the python pymodule fn).
* feat(kvo): add Klinger Volume Oscillator
Stephen J. Klinger's trend-aware volume-force MACD. Each bar produces a 'volume force' (vf) signed by the local trend (+1 / -1 / carry) and scaled by the ratio of the current accumulation horizon to its previous trend. KVO = EMA(vf, fast) - EMA(vf, slow), classic (34, 55).
Rust core (Kvo) with 7 unit tests (rejects zero / fast>=slow, accessors, constant series collapses to 0, warmup lands at slow+1, batch == streaming, reset clears state), plus Python (PyKvo + KVO export), Node (KvoNode), and WASM (WasmKvo) bindings. Fuzz target adds Kvo to the candle-input sweep, bench adds the candle-input KVO benchmark, README counter 71 -> 72 + family table row, CHANGELOG [Unreleased].
* feat(volume-oscillator): add Volume Oscillator (VO)
Percent difference between a fast and a slow SMA of the bar volume: 100 * (SMA(vol, fast) - SMA(vol, slow)) / SMA(vol, slow). Default (14, 28). The line stays near zero in stable conditions; positive readings show rising short-term participation, negative readings show waning interest.
Rust core (VolumeOscillator) with 8 unit tests (period validation, accessors, constant volume == 0, zero-volume window defensive branch, two reference values verified algebraically, batch == streaming, reset), plus Python (PyVolumeOscillator + VolumeOscillator export), Node (VolumeOscillatorNode), and WASM (WasmVolumeOscillator) bindings. Fuzz target adds VolumeOscillator to the candle-input sweep, bench adds the volume_oscillator benchmark, README counter 72 -> 73 + family table row, CHANGELOG [Unreleased].
* feat(nvi-pvi): add Negative & Positive Volume Index
Paul Dysart's cumulative volume-flow indices, popularised by Norman Fosback in 'Stock Market Logic'. Both run from a 1000.0 baseline and only update on a specific direction of volume change:
- NVI updates on volume-contraction bars (volume_t < volume_{t-1}), absorbing the percent close change. Tracks the 'smart money' leg per Fosback.
- PVI updates on volume-expansion bars (volume_t > volume_{t-1}). Tracks the 'crowd' leg.
Both expose with_baseline(f64) for custom starting indexes. The NVI/PVI pair is listed as a single line in indicator-ideas/families/07-volume.md and shares the same lifecycle/test/binding surface, so they ship as one commit.
Rust core (Nvi, Pvi) with 9 unit tests each (accessors, baseline seed, volume direction branches, zero-prev-close guard, custom baseline, batch == streaming, reset), plus Python (PyNvi/PyPvi + NVI/PVI exports), Node (NviNode/PviNode), and WASM (WasmNvi/WasmPvi) bindings. Fuzz target adds Nvi+Pvi to the candle-input sweep, bench adds nvi+pvi entries, README counter 73 -> 75 + family table row, CHANGELOG [Unreleased].
* feat(family-07): add Williams A/D, Anchored VWAP, Demand Index, TSV, VZO, Market Facilitation Index
Finishes the volume-flow family with the remaining (new) entries from
indicator-ideas/families/07-volume.md.
Indicators added:
- Williams A/D (`WilliamsAD`): Larry Williams' volume-less cumulative
accumulation/distribution line. Anchors each bar's contribution to
the previous close via true-high/true-low (gap-aware).
- Anchored VWAP (`AnchoredVwap`): cumulative VWAP whose accumulation
starts at a user-chosen anchor bar. Exposes `set_anchor()` (queued
to the next `update`) for click-to-anchor workflows. Reset clears
both state and pending-anchor flag.
- Demand Index (`DemandIndex`): James Sibbet's smoothed buying-vs-
selling pressure, in the streaming-friendly textbook form
`EMA(volume * close-return * (1 + range/close), period)`.
- Time Segmented Volume (`Tsv`): Don Worden's rolling window-sum of
`(close_t - close_{t-1}) * volume_t`. Default `period = 18`.
- Volume Zone Oscillator (`Vzo`): Walid Khalil's normalised volume-flow
oscillator bounded in `[-100, +100]`, defined as
`100 * EMA(signed_volume) / EMA(volume)`.
- Market Facilitation Index (`MarketFacilitationIndex`): Bill Williams'
per-bar `(high - low) / volume`. Returns `None` on zero-volume bars.
All six indicators ship with unit tests (`rejects_zero_period` where
applicable, `accessors_and_metadata`, constant-series behaviour,
batch == streaming equivalence, reset semantics, and reference-value
or saturation-extreme tests), Python / Node / WASM bindings, fuzz
coverage in `indicator_update_candle`, a `bench_candle_input` line per
indicator, README + CHANGELOG entries, and Python reference-value
tests in `test_new_indicators.py`.
The README indicator counter advances 75 -> 81.
* test(family-07): cover defensive cold paths + Default impls
- ad_oscillator: exercise `value()` after first emission.
- kvo: cover the `cm == 0.0` zero-OHLC defensive branch.
- nvi / pvi: exercise the Default impls.
