ta-indicators 0.1.0

Warmup-exact Rust port of TA-Lib batch technical indicators
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215

#!/usr/bin/env python3
"""Generate deterministic TA-Lib parity fixtures for `ta-indicators`.

This is the ground-truth contract for the Rust TA-Lib port. It builds a fixed
synthetic OHLCV series (seeded, reproducible) and records reference outputs from
the locally installed TA-Lib (`artifacts/venv-talib`) for every function we port.

The Rust integration tests under `tests/` load the emitted JSON and assert the
Rust implementations match within tolerance.

Run with the talib venv (it owns the compiled TA-Lib C extension):

    artifacts/venv-talib/bin/python scripts/gen_parity_fixtures.py

Output: tests/fixtures/talib_parity.json
"""

from __future__ import annotations

import json
import math
from pathlib import Path

import numpy as np
import talib

# Deterministic fixture size. Long enough for Hilbert warmup (~63 bars) plus
# slow MAs, short enough to keep the committed fixture small.
N = 400
SEED = 20260603
OUT_PATH = Path(__file__).resolve().parents[1] / "tests/fixtures/talib_parity.json"


def build_series() -> dict[str, np.ndarray]:
    """Seeded random-walk OHLCV plus a correlated second series for BETA/CORREL."""
    rng = np.random.default_rng(SEED)
    # Primary close: geometric random walk with mild drift.
    rets = rng.normal(0.0005, 0.012, N)
    close = 100.0 * np.exp(np.cumsum(rets))
    # Intrabar geometry derived from close so OHLC is internally consistent.
    spread = np.abs(rng.normal(0.0, 0.006, N)) * close
    open_ = close * (1.0 + rng.normal(0.0, 0.004, N))
    high = np.maximum(open_, close) + spread
    low = np.minimum(open_, close) - spread
    volume = rng.uniform(1.0e5, 5.0e5, N)
    # Second series: correlated market proxy for BETA / CORREL.
    market_rets = 0.7 * rets + 0.3 * rng.normal(0.0005, 0.012, N)
    close2 = 100.0 * np.exp(np.cumsum(market_rets))
    return {
        "open": open_,
        "high": high,
        "low": low,
        "close": close,
        "volume": volume,
        "close2": close2,
    }


def clean(arr: np.ndarray) -> list:
    """NaN/inf -> None so the JSON maps onto Rust Option<f64>/Option<i32>."""
    out: list = []
    for value in arr:
        if value is None or (isinstance(value, float) and not math.isfinite(value)):
            out.append(None)
        elif isinstance(value, (np.floating, float)):
            out.append(float(value))
        else:
            out.append(int(value))
    return out


def main() -> None:
    s = build_series()
    o, h, low, c, v, c2 = s["open"], s["high"], s["low"], s["close"], s["volume"], s["close2"]

    expected: dict[str, list] = {}

    # --- Cycle Indicators (Hilbert transform family) ---
    expected["ht_dcperiod"] = clean(talib.HT_DCPERIOD(c))
    expected["ht_dcphase"] = clean(talib.HT_DCPHASE(c))
    inphase, quadrature = talib.HT_PHASOR(c)
    expected["ht_phasor_inphase"] = clean(inphase)
    expected["ht_phasor_quadrature"] = clean(quadrature)
    sine, leadsine = talib.HT_SINE(c)
    expected["ht_sine"] = clean(sine)
    expected["ht_leadsine"] = clean(leadsine)
    expected["ht_trendmode"] = clean(talib.HT_TRENDMODE(c))
    expected["ht_trendline"] = clean(talib.HT_TRENDLINE(c))

    # --- Overlap Studies (Hilbert-adaptive + extended) ---
    mama, fama = talib.MAMA(c, fastlimit=0.5, slowlimit=0.05)
    expected["mama_0p5_0p05"] = clean(mama)
    expected["fama_0p5_0p05"] = clean(fama)
    expected["sarext_default"] = clean(
        talib.SAREXT(
            h,
            low,
            startvalue=0.0,
            offsetonreverse=0.0,
            accelerationinitlong=0.02,
            accelerationlong=0.02,
            accelerationmaxlong=0.2,
            accelerationinitshort=0.02,
            accelerationshort=0.02,
            accelerationmaxshort=0.2,
        )
    )

    # --- Momentum (extended MACD) ---
    macd, macdsignal, macdhist = talib.MACDEXT(
        c, fastperiod=12, fastmatype=0, slowperiod=26, slowmatype=0, signalperiod=9, signalmatype=0
    )
    expected["macdext_12_26_9"] = clean(macd)
    expected["macdext_signal_12_26_9"] = clean(macdsignal)
    expected["macdext_hist_12_26_9"] = clean(macdhist)

    # --- Statistic Functions (cross-series) ---
    expected["beta_5"] = clean(talib.BETA(c, c2, timeperiod=5))
    expected["correl_30"] = clean(talib.CORREL(c, c2, timeperiod=30))

    # --- Overlap: MAVP (variable-period MA) ---
    periods = np.full(N, 14.0)
    periods[N // 2 :] = 30.0
    expected["mavp_14_30"] = clean(talib.MAVP(c, periods, minperiod=2, maxperiod=30, matype=0))

    # --- Pattern Recognition (all CDL* candlestick functions) ---
    cdl_funcs = sorted(talib.get_function_groups()["Pattern Recognition"])
    for name in cdl_funcs:
        fn = getattr(talib, name)
        expected[name.lower()] = clean(fn(o, h, low, c))

    # --- Existing-coverage audit: every already-ported public crate function ---
    # Trend / moving averages
    expected["ema_14"] = clean(talib.EMA(c, timeperiod=14))
    expected["dema_14"] = clean(talib.DEMA(c, timeperiod=14))
    expected["tema_14"] = clean(talib.TEMA(c, timeperiod=14))
    expected["t3_5_0p7"] = clean(talib.T3(c, timeperiod=5, vfactor=0.7))
    expected["kama_30"] = clean(talib.KAMA(c, timeperiod=30))
    expected["trix_15"] = clean(talib.TRIX(c, timeperiod=15))

    # Momentum
    expected["rsi_14"] = clean(talib.RSI(c, timeperiod=14))
    expected["cmo_14"] = clean(talib.CMO(c, timeperiod=14))
    # Crate apo/ppo are EMA-based, so request matype=1 (EMA) for a fair parity check.
    expected["apo_12_26"] = clean(talib.APO(c, fastperiod=12, slowperiod=26, matype=1))
    expected["ppo_12_26"] = clean(talib.PPO(c, fastperiod=12, slowperiod=26, matype=1))
    expected["ultosc_7_14_28"] = clean(talib.ULTOSC(h, low, c, timeperiod1=7, timeperiod2=14, timeperiod3=28))
    expected["bop"] = clean(talib.BOP(o, h, low, c))
    macd_m, macd_s, macd_h = talib.MACD(c, fastperiod=12, slowperiod=26, signalperiod=9)
    expected["macd_12_26_9"] = clean(macd_m)
    expected["macd_signal_12_26_9"] = clean(macd_s)
    expected["macd_hist_12_26_9"] = clean(macd_h)
    # NOTE: TA-Lib's MACDFIX emits a MACD line that is not EMA12-EMA26 and never
    # converges to it (verified out to the full series), so it is intentionally
    # excluded from the parity contract. The crate's `macdfix` aliases the
    # canonical `macd(.., 12, 26, signal)`, which is the correct definition.
    srsi_k, srsi_d = talib.STOCHRSI(c, timeperiod=14, fastk_period=5, fastd_period=3, fastd_matype=0)
    expected["stochrsi_k_14_5_3"] = clean(srsi_k)
    expected["stochrsi_d_14_5_3"] = clean(srsi_d)

    # ADX / directional movement family
    expected["adx_14"] = clean(talib.ADX(h, low, c, timeperiod=14))
    expected["adxr_14"] = clean(talib.ADXR(h, low, c, timeperiod=14))
    expected["plus_di_14"] = clean(talib.PLUS_DI(h, low, c, timeperiod=14))
    expected["minus_di_14"] = clean(talib.MINUS_DI(h, low, c, timeperiod=14))
    expected["dx_14"] = clean(talib.DX(h, low, c, timeperiod=14))
    expected["plus_dm_1"] = clean(talib.PLUS_DM(h, low, timeperiod=1))
    expected["minus_dm_1"] = clean(talib.MINUS_DM(h, low, timeperiod=1))

    # Bands / regression
    bb_u, bb_m, bb_l = talib.BBANDS(c, timeperiod=20, nbdevup=2.0, nbdevdn=2.0, matype=0)
    expected["bb_upper_20_2"] = clean(bb_u)
    expected["bb_middle_20_2"] = clean(bb_m)
    expected["bb_lower_20_2"] = clean(bb_l)
    expected["linearreg_14"] = clean(talib.LINEARREG(c, timeperiod=14))
    expected["linearreg_slope_14"] = clean(talib.LINEARREG_SLOPE(c, timeperiod=14))
    expected["linearreg_angle_14"] = clean(talib.LINEARREG_ANGLE(c, timeperiod=14))
    expected["linearreg_intercept_14"] = clean(talib.LINEARREG_INTERCEPT(c, timeperiod=14))
    expected["tsf_14"] = clean(talib.TSF(c, timeperiod=14))

    # Aroon
    aroon_down, aroon_up = talib.AROON(h, low, timeperiod=14)
    expected["aroon_up_14"] = clean(aroon_up)
    expected["aroon_down_14"] = clean(aroon_down)
    expected["aroon_osc_14"] = clean(talib.AROONOSC(h, low, timeperiod=14))

    # Volume / volatility / transforms / math
    expected["ad"] = clean(talib.AD(h, low, c, v))
    expected["adosc_3_10"] = clean(talib.ADOSC(h, low, c, v, fastperiod=3, slowperiod=10))
    expected["sar_0p02_0p2"] = clean(talib.SAR(h, low, acceleration=0.02, maximum=0.2))
    expected["sum_10"] = clean(talib.SUM(c, timeperiod=10))
    expected["avgprice"] = clean(talib.AVGPRICE(o, h, low, c))
    expected["medprice"] = clean(talib.MEDPRICE(h, low))
    expected["typprice"] = clean(talib.TYPPRICE(h, low, c))
    expected["wclprice"] = clean(talib.WCLPRICE(h, low, c))

    fixture = {
        "meta": {
            "seed": SEED,
            "n": N,
            "talib_version": talib.__version__,
            "description": "Deterministic TA-Lib parity reference for ta-indicators.",
        },
        "input": {key: clean(value) for key, value in s.items()},
        "expected": expected,
    }

    OUT_PATH.parent.mkdir(parents=True, exist_ok=True)
    OUT_PATH.write_text(json.dumps(fixture, indent=1))
    print(f"wrote {OUT_PATH} ({OUT_PATH.stat().st_size / 1024:.0f} KiB)")
    print(f"functions: {len(expected)} (incl. {len(cdl_funcs)} CDL patterns)")


if __name__ == "__main__":
    main()