Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
re: Pair Trade using a Risk Factor Model - analysis of 260k pairs

This post examines the relationships between pair trading performance obtained from a window of data (t:t+80) versus the correlations (COR) and signal to noise ratios (SNR) of paired stocks (Vidyamurthy 2004) obtained from another window of data (t-281:t-1) in 12 sectors. This post and the corresponding code is an extension of David Edwards post.
- COR and SNR are obtained via a rolling window fashion, and their median (cor_m,snr_m) and standard deviation (cor_sd,snr_sd) are plotted against the sharpe ratio
- The stocks analyzed are those listed in NYSE.
- The factors used are:
YAHOO/INDEX_HUI, YAHOO/INDEX_VIX, YAHOO/INDEX_OSX, YAHOO/INDEX_XAU,
YAHOO/TSX_RTRE_TO,YAHOO/INDEX_WILREIT,YAHOO/INDEX_SML,
YAHOO/INDEX_N225,YAHOO/INDEX_W5KLCV,YAHOO/INDEX_HCX.
With the factors used, judging by the correlation coeffcient from cor_m and sharpe(strategy-equal weighted), there may be some edge trading pairs in the consumer durables sector (r=0.1).

1 response

This is exceptionally strong work. Well done -- I am glad to see you able to use the research platform to find these results. It looks like consumer durables have some tradeable pairs. I'd also look further into public utilities -- though there's no correlation, the histogram suggests that there are some high-sharpe possibilities. Have you considered taking the next step and turning this into an algorithm? Please feel free to reach out, either in the forums or to [email protected], if you need help at any step of the way.