Statistical Arbitrage Team Xing Liu, Leo, Ying, Yandong
Summary of Changes S-score calculation – Bug: Alpha = beta0 * 252 – Added k -> – Changed way to compute m: PNL calculation – Add transaction costs Opened the interface to integrate ETF in the code. Verified – Trading signals – PNL calculations – Way to hedge – just discussed with Professor Implemented modified returns according to daily trading volume.
Results Plot PNL – Transaction costs (with/without) Kappa & Mean of m ON / Volume OFF – Volume (with/without) Kappa & Mean of m ON / Transaction FeeOFF – Computing mean of m (with/without) Kappa ON / Volume & Transaction Fee OFF – Kappa (with/without) Mean of m ON / Volume & Transaction Fee OFF
With / Without transaction costs Without computing mean of m Actual time return K>8.4
With / Without computing mean of m Without transaction fee Actual time return K>8.4
trading time modified return/ actual time return Without transaction fee With computing mean of m K>8.4
With / Without kappa Without transaction fee With computing mean of m Actual time return
Plan Change the way of hedging. – Ensure beta neutral while hedging – Try to use SPY rather than eigenportfolio to hedge – Do all the simulations Integrate the ETF Verify the software to get the release version. Generate the final report.
Suggestions?
K<8.4 Fast mean-reversion requires that k> 252/30. If k<8.4, – 1) do not open trades – 2) close open trades