Pair Trading Strategy Journal

Thanks fredman007, if I plan to keep pair overnight I trade one leg at a time usually trade the short side first then buy long sometimes I can get a better price if I am right on the direction of the stock. When I do day trade I use combo stock/stock offered by IB, sometimes I do not get the order filled right away because IB does not guarantee the order but I don't have slippage.

PA

 

I have developed a quasi pair-trading system that I am papertrading.

Each Friday at the close, it:

1) shorts the 20 stocks that have both recently most-overperformed relative to their industry (after adjusting for volatility) but have tracked their respective industries reasonably well over the past 150 days. The other side of the pair is the average performance for the industry.

2) goes long the 20 stocks that have both recently most-underperformed relative to their industry (after adjusting for volatility) but have tracked their respective industries reasonably well over the past 150 days. The other side of the pair is the average performance for the industry.

Positions are closed at the next Friday close.

Last week's performance
average long 0.80% (mostly due to OCNW)
average short -0.84%
sum -0.04%

Although I use the % sign, this is only approximate due to the volatility adjustment. However, no long-term bias should be introduced by my methodology.

My definition of "have tracked their respective industries reasonably well" is not based on correlation. It is based on the price-adjusted and volatility-adjusted charts for the stock vs its associated industry crossing each other with regularity over the past 150 days.

 

New Trade:

Long XOP
Short IWM

what do you think?

 

You are betting that oil price is gonna stay flat for the holding period.
I'd wait till their ratio gets to 1.1x, (2st dev on Bollinger 50) before I trigger.

 

Dukatu --

Would you mind explaining how you got 1.1x as the trigger? I'm calculating the MA(50) as 1.56x and the lower 2 stdev band at 1.51, so I'm not sure what I'm missing.

Also, are you having good success using the 50-day moving average to generate triggers? I've been using the 252-day average, but I'm not getting great signals.

Much appreciated.

 

i think 252-days are to long as lookback-period for trading pairs; i use 14 days (or 30 days)

 

I did a preliminary analysis on stockcharts.

Check the blue line, Boll of 2 st dev.

Here is the theoretical background is normally use:
http://www.stanford.edu/~avinayan/Report445.pdf

 

Sold today.

MET 39.21 - 0.6%

UNM 22.15 6.7%


Holding period when little bit over, but what the heck. Net gain 6.1%

 

looks nice; how many pairs do you analyse on a daily basis?

 

3-5/day

Also, I’m a huge fan of 3-legged combos, 2 Sells+1Buy or 1S+2B. Usually the 3rd security is the industry ETF/SPDR. I have not done any official research on this; however noticed that they behave better in the co-integration analysis, providing ADF readings in excess of 5.