Cointegration is the key trigger in statistical arbitrage for correlated pair, measuring inconsistency and stability properties
Explanation of Homoscedasticity with normal diffused (white noise)
ADF_Cointegration.ipynbdiffusion convergence with drift term
KPSS_Cointegration.ipynbThesis: A Pairs Causality Research with ARIMA Forecasting
-
to be refined by Grid Search for time-window & stop-profit hyper-param, and calibrating divergent adjusting factor (Cocoercivity, a basin hopping parameter for long-term drift)
-
adding momentum rules(MA cross/ RSI/ tailormade Alpha) in practice (Hurst exponent>0.5: Long Vol by trending α momentum; Hurst exponent<0.5: Short Vol by Mean Reversion) (e.g. 5 lag=0.86 means more reverse in short term and 20 lags=0.91 means more trendy in longer term)
-
as Multi-asset Pricing Model by Second-Order Cone Programming (Solver for Linear/ Quadratic(non-linear) Gaussian simulation) with neural autograd