ENH ht_trendline #759 MAINT general PERF wma mfi #731 DOC TST

.gitignore

@@ -139,9 +139,9 @@ data/SPY_D_TV3.csv
 data/TV_5min.csv
 data/tulip.csv
 examples/*.csv
-jnb/*
-notes/*
-csv/*
+jnb/
+notes/
+csv/
 
 *.txt
 df_file

Makefile

@@ -1,6 +1,6 @@
 .PHONY: tests
 tests:
-	pytest -vv -s -l tests
+	pytest -vvv -s -l tests
 
 caches:
 	find ./pandas_ta | grep -E "(__pycache__|\.pyc|\.pyo$\)"
@@ -12,16 +12,16 @@ init:
 	pip install -r requirements.txt
 
 test_metrics:
-	pytest -vv -s -l --cache-clear tests/test_metrics.py
+	pytest -vv -s -l -W ignore::DeprecationWarning --cache-clear tests/test_metrics.py
 
 test_numba:
-	pytest -vv -s -l --cache-clear tests/test_numba.py
+	pytest -vv -s -l -W ignore::DeprecationWarning --cache-clear tests/test_numba.py
 
 test_studies:
-	pytest -vv -s -l --cache-clear tests/test_studies.py
+	pytest -vv -s -l -W ignore::DeprecationWarning --cache-clear tests/test_studies.py
 
 test_ta:
-	pytest -vv -s -l --cache-clear tests/test_indicator_*.py
+	pytest -vv -s -l -W ignore::DeprecationWarning --cache-clear tests/test_indicator_*.py
 
 test_utils:
-	pytest -vv -s -l --cache-clear tests/test_utils.py
+	pytest -vv -s -l -W ignore::DeprecationWarning --cache-clear tests/test_utils.py

pandas_ta/_typing.py

@@ -59,4 +59,4 @@
 MaybeSeriesFrame = Union[T, Series, DataFrame]
 AnyArray = Union[Array, Series, DataFrame]
 AnyArray1d = Union[Array1d, Series]
-AnyArray2d = Union[Array2d, DataFrame]
+AnyArray2d = Union[Array2d, DataFrame]

pandas_ta/overlap/mama.py

@@ -11,44 +11,47 @@
 # http://traders.com/documentation/feedbk_docs/2014/01/traderstips.html
 @njit
 def np_mama(x, fastlimit, slowlimit, prenan):
-    a1, a2 = 0.0962, 0.5769
-    p_w, smp_w, smp_w_c = 0.2, 0.33, 0.67 # smp_w + smp_w_c = 1
-
-    sm = zeros_like(x)
-    dt, smp, q1, q2 = sm.copy(), sm.copy(), sm.copy(), sm.copy()
-    i1, i2, jI, jQ = sm.copy(), sm.copy(), sm.copy(), sm.copy()
-    re, im, alpha = sm.copy(), sm.copy(), sm.copy()
-    period, phase, mama, fama = sm.copy(), sm.copy(), sm.copy(), sm.copy()
-
-    # Ehler's starts from 6, TV-LB starts at 3, TALib 32
-    n = x.size
-    for i in range(3, n):
-        w_period = .075 * period[i - 1] + .54
-
-        # Smoother and Detrend the Smoother
-        sm[i] = 0.4 * x[i] + 0.3 * x[i - 1] + 0.2 * x[i - 2] + 0.1 * x[i - 3]
-        dt[i] = w_period * (a1 * sm[i] + a2 * sm[i - 2] - a2 * sm[i - 4] - a1 * sm[i - 6])
+    a, b, m = 0.0962, 0.5769, x.size
+    p_w, smp_w, smp_w_c = 0.2, 0.33, 0.67
+
+    wma4 = zeros_like(x)
+    dt, smp = zeros_like(x), zeros_like(x)
+    i1, i2 = zeros_like(x), zeros_like(x)
+    ji, jq = zeros_like(x), zeros_like(x)
+    q1, q2 = zeros_like(x), zeros_like(x)
+    re, im, alpha = zeros_like(x), zeros_like(x), zeros_like(x)
+    period, phase = zeros_like(x), zeros_like(x)
+    mama, fama = zeros_like(x), zeros_like(x)
+
+    # Ehler's starts from 6, TV-LB from 3, TALib from 32
+    for i in range(3, m):
+        adj_prev_period = 0.075 * period[i - 1] + 0.54
+
+        # WMA(x,4) & Detrended WMA(x,4)
+        wma4[i] = 0.4 * x[i] + 0.3 * x[i - 1] + 0.2 * x[i - 2] + 0.1 * x[i - 3]
+        dt[i] = adj_prev_period * (a * wma4[i] + b * wma4[i - 2] - b * wma4[i - 4] - a * wma4[i - 6])
 
         # Quadrature(Detrender) and In Phase Component
-        q1[i] = w_period * (a1 * dt[i] + a2 * dt[i - 2] - a2 * dt[i - 4] - a1 * dt[i - 6])
+        q1[i] = adj_prev_period * (a * dt[i] + b * dt[i - 2] - b * dt[i - 4] - a * dt[i - 6])
         i1[i] = dt[i - 3]
 
-        # Phase advance I1 and Q1 by 90 degrees
-        jI[i] = w_period * (a1 * i1[i] + a2 * i1[i - 2] - a2 * i1[i - 4] - a1 * i1[i - 6])
-        jQ[i] = w_period * (a1 * q1[i] + a2 * q1[i - 2] - a2 * q1[i - 4] - a1 * q1[i - 6])
+        # Phase Q1 and I1 by 90 degrees
+        ji[i] = adj_prev_period * (a * i1[i] + b * i1[i - 2] - b * i1[i - 4] - a * i1[i - 6])
+        jq[i] = adj_prev_period * (a * q1[i] + b * q1[i - 2] - b * q1[i - 4] - a * q1[i - 6])
 
         # Phasor Addition for 3 Bar Averaging
-        i2[i] = i1[i] - jQ[i]
-        q2[i] = q1[i] + jI[i]
+        i2[i] = i1[i] - jq[i]
+        q2[i] = q1[i] + ji[i]
 
-        # Smooth I and Q components
+        # Smooth I2 & Q2
         i2[i] = p_w * i2[i] + (1 - p_w) * i2[i - 1]
         q2[i] = p_w * q2[i] + (1 - p_w) * q2[i - 1]
 
         # Homodyne Discriminator
         re[i] = i2[i] * i2[i - 1] + q2[i] * q2[i - 1]
         im[i] = i2[i] * q2[i - 1] + q2[i] * i2[i - 1]
 
+        # Smooth Re & Im
         re[i] = p_w * re[i] + (1 - p_w) * re[i - 1]
         im[i] = p_w * im[i] + (1 - p_w) * im[i - 1]
 

pandas_ta/overlap/pivots.py

@@ -1,5 +1,5 @@
 # -*- coding: utf-8 -*-
-from numpy import NaN, greater, zeros_like
+from numpy import greater, nan, zeros_like
 from numba import njit
 from pandas import DataFrame, Series, Timedelta, infer_freq, to_datetime
 from pandas_ta._typing import Array, DictLike
@@ -197,7 +197,7 @@ def pivots(
 
     # Create nan arrays for "demark" and "fibonacci" pivots
     _nan_array = zeros_like(np_close)
-    _nan_array[:] = NaN
+    _nan_array[:] = nan
     tp = s1 = s2 = s3 = s4 = r1 = r2 = r3 = r4 = _nan_array
 
     # Calculate

pandas_ta/overlap/wma.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
-from numpy import arange, dot
+from numpy import arange, dot, float64, nan, zeros_like
+from numba import njit
 from pandas import Series
 from pandas_ta._typing import DictLike, Int
 from pandas_ta.maps import Imports
@@ -12,6 +13,25 @@
 )
 
 
+@njit
+def np_wma(x, n, asc, prenan):
+    m = x.size
+    w = arange(1, n + 1, dtype=float64)
+    result = zeros_like(x, dtype=float64)
+
+    if not asc:
+        w = w[::-1]
+
+    for i in range(n - 1, m):
+        result[i] = (w * x[i - n + 1:i + 1]).sum()
+    result *= 2 / (n * n + n)
+
+    if prenan:
+        result[:n - 1] = nan
+
+    return result
+
+
 def wma(
     close: Series, length: Int = None,
     asc: bool = None, talib: bool = None,
@@ -56,17 +76,9 @@ def wma(
         from talib import WMA
         wma = WMA(close, length)
     else:
-        total_weight = 0.5 * length * (length + 1)
-        weights_ = Series(arange(1, length + 1))
-        weights = weights_ if asc else weights_[::-1]
-
-        def linear(w):
-            def _compute(x):
-                return dot(x, w) / total_weight
-            return _compute
-
-        close_ = close.rolling(length, min_periods=length)
-        wma = close_.apply(linear(weights), raw=True)
+        np_close = close.values
+        wma_ = np_wma(np_close, length, asc, True)
+        wma = Series(wma_, index=close.index)
 
     # Offset
     if offset != 0:

pandas_ta/trend/alphatrend.py

@@ -1,5 +1,5 @@
 # -*- coding: utf-8 -*-
-from numpy import isnan, nan, zeros
+from numpy import isnan, nan, zeros_like
 from numba import njit
 from pandas import DataFrame, Series
 from pandas_ta._typing import Array, DictLike, Int, IntFloat
@@ -16,11 +16,10 @@
 )
 
 
-# Alphatrend alpha threshold calculation
 @njit
 def np_alpha(low_atr, high_atr, momo_threshold):
     m = momo_threshold.size
-    alpha = zeros(m)
+    alpha = zeros_like(low_atr)
 
     for i in range(1, m):
         if momo_threshold[i]: