Time Series Analysis for Traders: Beyond Moving Averages

Moving averages are just the beginning. Here are more sophisticated time series techniques that can improve your trading analysis. Exponentially Weighted Statistics Standard moving averages weight all periods equally. Exponential weighting gives more importance to recent data: import numpy as np import pandas as pd def ewm_volatility ( prices , span = 20 ): returns = np . diff ( np . log ( prices )) ewm_var = pd . Series ( returns ). ewm ( span = span ). var () return np . sqrt ( ewm_var * 252 ) # Annualized def ewm_correlation ( series1 , series2 , span = 30 ): r1 = np . diff ( np . log ( series1 )) r2 = np . diff ( np . log ( series2 )) return pd . Series ( r1 ). ewm ( span = span ). corr ( pd . Series ( r2 )) Hurst Exponent: Trending or Mean-Reverting? def hurst_exponent ( prices , max_lag = 100 ): """ H > 0.5: trending (momentum strategies work) H = 0.5: random walk (no edge) H < 0.5: mean-reverting (reversion strategies work) """ lags = range ( 2 , max_lag ) tau = [] for lag in la