Pair Trading Strategy Journal

Quote from jonnysharp:

OIH is dramatically oversold at the moment, popping up on my radar everywhere as a buy, probably due to the gulf oil spill, fading these types of situations usually pays off in the long run.

My favourite right now, currently 2.95 standard deviations and 13.71% below its mean, huge divergence;

Long OIH @ 89.48
Short XOP @ 39.43
More...

Exited this trade on Friday;

Sold OIH @ 96.54
Covered XOP @ 38.26

10.8% return on capital invested, had larger size on since my conviction on this trade was high. Love the ETF pair trades, makes me sleep easier at night knowing I'm not going to wake up to a takeover or bankruptcy 30% move associated with stocks.
 
Quote from jonnysharp:

Exited this trade on Friday;

Sold OIH @ 96.54
Covered XOP @ 38.26

10.8% return on capital invested, had larger size on since my conviction on this trade was high. Love the ETF pair trades, makes me sleep easier at night knowing I'm not going to wake up to a takeover or bankruptcy 30% move associated with stocks.
More...

You made 7.9% on the long side and 3% on the short side. Congratulations on that, but surely your return on capital invested is only 5.4% (average of the two).

Had you used 100% leverage in the trade, then your return on equity would be 10.8%
 
Jonny,
Very happy to see you are still well and trading.
Bought Pair trade finder a few weeks ago, still paper trading and learning.
Thanks for all of your very instructive posts.
 
Quote from waditude:

For those of you who have pairstraded for some years, what type of returns have you achieved and how have these returns varied year-by-year?

I am considering getting into this game. No, I have not read all 300+ pages of this journal, but I certainly will if the light at the end of the tunnel is bright enough. If one is intelligent, uses proper money management, and does one's homework, what may he expect?

Thank you.
More...

I think if you put your homework in and strictly follow a method 30-50% returns per year on a retail size account is not out of the question with pair trading. Of course some years will be better than others, for example 2008 was a cracker of a year for pair trading, I personally did over 100% return as documented in this journal and yes it is worth the read, there are quite a few experienced pair traders who have shared their nuggets of gold in this thread, I still go back and read through it every now and then.
 
Do industry groups revert to the mean?

I used the 30-group Fama-French data to answer this question. (I discarded the "other" group and was left with 29 groups.)

I used data from 2005 to the present. Although older data was available, I chose not to include it, as these sorts of relationships can change with time. I regressed (linear regression) each week's percentage price change for each group against:

1. the previous weeks percentage price change
slope= -0.0787
2. the previous 2 weeks percentage price change
slope= +0.0122
3. the previous 4 weeks percentage price change
slope= +0.0147
4. the previous 8 weeks percentage price change
slope= +0.0090


I then looked at the data another way. For each week I asked what the performance would be if I bought the 5 groups with the best or worst performance the previous week.

The best performing 5 groups returned +0.18% on average the following week.
The worst performing 5 groups returned +0.33% on average the following week.
The average of all 29 groups returned +0.26% the following week.

Conclusion: Although there is a week-to-week industry-group reversion to the mean effect, it is quite tiny and probably should not influence one's trading decisions.
 
Quote from kuzco:

Bit of a silly question but where do you put the stop loss?
More...

I don't know what you mean. This is not a trading system - I just looked at industry group reversion-to-the-mean to see if there was an edge here that one could use.


Quote from konviction:

Is it possible to scan for those correlations on a weekly time frame?
More...

These are not correlations. These slopes are regression coefficients and were measured over a 5 1/2 year period. An additional few weeks of data would not change anything.
 
You must log in or register to reply here.
Back
Top