def di(plus_dm: NdType, minus_dm: NdType, atr: NdType, ma: tp.Any = 14) -> tp.Tuple[NdType, NdType]:
if isinstance(ma, int):
ma_period = ma
ma = lambda series: wilder_ma(series, ma_period)
plus_di = 100 * ma(plus_dm) / atr # round ?
minus_di = 100 * ma(minus_dm) / atr # round ?
return plus_di, minus_di
def rsi(
series: nda.NdType,
ma: tp.Any = 14 # lambda series: wilder_ma(series, 14)
) -> nda.NdType:
if isinstance(ma, int):
ma_period = ma
ma = lambda series: wilder_ma(series, ma_period)
ch = change(series)
up = np.maximum(ch, 0) # positive changes
down = -np.minimum(ch, 0) # negative changes (inverted)
up = ma(up)
down = ma(down)
rs = up / down
rsi = 100 * (1 - 1 / (1 + rs))
return rsi
def adx(dx: NdType, ma: tp.Any = 14) -> NdType:
if isinstance(ma, int):
ma_period = ma
ma = lambda series: wilder_ma(series, ma_period)
return ma(dx)
from qnt.data import load_data, load_assets, ds
from qnt.xr_talib import RSI
import time
assets = load_assets()
ids = [i['id'] for i in assets[0:2000]]
data = load_data(assets=ids,
dims=(ds.TIME, ds.ASSET, ds.FIELD),
forward_order=True).sel(field='close')
data_pd = data.to_pandas()
t1 = time.time()
rsi1 = RSI(data, 14)
t2 = time.time()
rsi2 = rsi(data, lambda series: wilder_ma(series, 14))
t3 = time.time()
rsi3 = rsi(data_pd, lambda series: wilder_ma(series, 14))
t4 = time.time()
print(t2 - t1, t3 - t2, t4 - t3, "relative delta =",
abs(rsi1 - rsi2).mean().values, "t(talib)/t(this) =",
(t2 - t1) / (t3 - t2))
import matplotlib.pyplot as plt
plt.plot(rsi1.coords[ds.TIME].values,
rsi1.sel(asset='NASDAQ:AAPL').values, 'r')
plt.plot(rsi2.coords[ds.TIME].values,
rsi2.sel(asset='NASDAQ:AAPL').values, 'g')
plt.show()
def atr(high: NdType, low: NdType, close: NdType, ma: tp = 14) -> NdType:
if isinstance(ma, int):
ma_period = ma
ma = lambda s: wilder_ma(s, ma_period)
return ma(tr(high, low, close))