Pair Trading Strategy Journal

Quote from jonnysharp:

knocks, yes there have been some strange happenings in the pair trading world, these extraordinary events must be having an impact like you say, good opportunities for those who could identify them.

Nick, yes pairtrade finder has spread charts, however I don't use them.

Exited one trade today;

Sold HAL @ 30.29
Covered CNQ @ 64.10

One new trade today;

Long PHM @ 13.91
Short TOL @ 24.31
More...


Interesting Pair Johnny with the PHM and TOL. It is one I track as well. I had the reverse on today and did well with it.

Several weeks ago, I added a new screening criteria to my system. In addition to looking for pairs that have deviated more than 2x away from the mean, I also look for pairs who are trading within 0.25 stddev from the mean, 80% correlation, and whose 14 day RSI spread is greater than 7%. I also want the the average RSI to be above 40 and less than 60. I then pick the stronger stock to go long according to RSI.

For example: PHM/TOL
1. Correlation = 95%
2. PHM 14RSI = 41
3. TOL 14RSI = 50
4. RSI Spread = .09 or 9%
5. Average RSI = 46
6. Pair ratio within .25 STDdev of the mean. Mean = .57 and the pair ratio at yesterday's close was .569

I have only been trading this criteria for about three weeks, but I find I can open the trade in the morning and close for a nice profit by the day's end or hold it if I choose.

Just another way to look at pairs, but the strategy seems to work. I find mean reversion can take a long time so I wanted to find another way to identify opportunities.

***How about the craziness in goog today. I was one of the shareholders during the mayhem and almost had a heart attack as I was long at 415 and watched it hit 489 and then drop to pennies all within minutes. I got real lucky and put a market order to sell because I panicked, but got filled at 424.04 which the NAZ accepted as valid. A quick 6500 in profit, but the scariest profit I ever made. Another reason to stick with pairs. They helps you sleep at night.

Cheers.
 
Yobo, yes that was some crazy trading in GOOG this afternoon. Interesting how you use RSI extensively in your trading, nice method you got there.

Busy day in the pair trading world.

Closed 3 trades:

Sold BAX @ 65.5
Covered GENZ @ 77.67

Sold AKS @ 25.92
Covered MT @ 49.38

Sold APA @ 104.67
Covered DVN @ 92.11

One new trade today;

Long NWS-A @ 11.99
Short IACI @ 17.30
 

Attachments

Quote from yobo:

Interesting Pair Johnny with the PHM and TOL. It is one I track as well. I had the reverse on today and did well with it.

Several weeks ago, I added a new screening criteria to my system. In addition to looking for pairs that have deviated more than 2x away from the mean, I also look for pairs who are trading within 0.25 stddev from the mean, 80% correlation, and whose 14 day RSI spread is greater than 7%. I also want the the average RSI to be above 40 and less than 60. I then pick the stronger stock to go long according to RSI.

For example: PHM/TOL
1. Correlation = 95%
2. PHM 14RSI = 41
3. TOL 14RSI = 50
4. RSI Spread = .09 or 9%
5. Average RSI = 46
6. Pair ratio within .25 STDdev of the mean. Mean = .57 and the pair ratio at yesterday's close was .569

I have only been trading this criteria for about three weeks, but I find I can open the trade in the morning and close for a nice profit by the day's end or hold it if I choose.

Just another way to look at pairs, but the strategy seems to work. I find mean reversion can take a long time so I wanted to find another way to identify opportunities.

***How about the craziness in goog today. I was one of the shareholders during the mayhem and almost had a heart attack as I was long at 415 and watched it hit 489 and then drop to pennies all within minutes. I got real lucky and put a market order to sell because I panicked, but got filled at 424.04 which the NAZ accepted as valid. A quick 6500 in profit, but the scariest profit I ever made. Another reason to stick with pairs. They helps you sleep at night.

Cheers.
More...

Yobo,

This sounds more like a dispersion trade?
 
No alerts for this type of trade today based on yesterday's close available in my database. But if you have any questions about the strategy, I'd be happy to elaborate.

The concept is actually very simple. Buy correlated pairs when they are at the mean, but be long the stronger stock and short the weaker based on 14 day RSI. The RSI spread becomes important because when the market hits a volitle patch, investor sentiment will almost 100% of the time sell the weaker first creating a profit opportunity. The reverse is true in a strong market. Investors will buy the stronger first and the weaker will follow but not at the same rate.

The other piece of the puzzle is to make sure you avoide a short squeeze or over bought condition hence the reason for the average RSI being above 40 and less than 60.
 
Quote from knocks420:

Curious if anyone noticed the response today to stat arb intraday?

Volatility at the open was followed by an immediate lull as everyone waited for the Bail Out decision and created low volatility but trending (downward) prices. Many of the intraday spreads I trade widened significantly and held until after the decision whereupon there were a number of opportunities.

Reminds me of something written on the downturn in stat arb during 01-03, theory was proposed that it was due to lack of institutional involvement and program/algo trading that caused spreads to widen and delayed reversion causing opportunity cost, perhaps today was a microcosm of that argument??
More...

Yesterday was a fine day but today was sh*t again, very low intra-stock volatility. Spreads of stocks i watch widened in the am and oscillated at new levels-never really reverting. I'm sitting on unrealized losses which is not the norm but felt reluctant to take the loss since the individual spreads were at extreme levels on lower frequencies.

As I wrote above, this day fit the thesis that returns are lower with lack of institutional participation and algo trading. It seemed that the index moved quite a bit but few people trading stocks, combine with low volume due to the holidays and continued apprehension going into Senate meeting and you get a sh*t day.

Other research i've seen on simple overbought/oversold fading strategies also seemed to underperform during economic stress or uncertainty. My 2 cents is to reduce size when forecasting decreased intra-stock vol. I wonder what the thoughts are on this strategy going forward?
 
Quote from knocks420:

Yesterday was a fine day but today was sh*t again, very low intra-stock volatility. Spreads of stocks i watch widened in the am and oscillated at new levels-never really reverting. I'm sitting on unrealized losses which is not the norm but felt reluctant to take the loss since the individual spreads were at extreme levels on lower frequencies.

As I wrote above, this day fit the thesis that returns are lower with lack of institutional participation and algo trading. It seemed that the index moved quite a bit but few people trading stocks, combine with low volume due to the holidays and continued apprehension going into Senate meeting and you get a sh*t day.

Other research i've seen on simple overbought/oversold fading strategies also seemed to underperform during economic stress or uncertainty. My 2 cents is to reduce size when forecasting decreased intra-stock vol. I wonder what the thoughts are on this strategy going forward?
More...

I got caught in several pairs like that today, it was frustrating to say the least, but tight spreads allow you to trade the stocks of the pair individually and weight yourself to the movements of the market. As the market began to trend upwards I was able to get extra long several pairs, some of which i took profits on my long positions and others which i waited and then shorted giving me a great spread price. Bkc/Mcd200 and Cop/Cvx finally turned for me the last hour or so of the day. With the $VIX this high I look to run and gun rather than build positions, tending to daytrade rather than carry overnight pairs positions. Tighter, higher correlated spreads have offered great oppurtunities with high volatility like pbr/pbr.a and rio/rio.pr. The tight spreads offer some safety if you get whipsawed.
 
Quote from samueldoernte:

I got caught in several pairs like that today, it was frustrating to say the least, but tight spreads allow you to trade the stocks of the pair individually and weight yourself to the movements of the market. As the market began to trend upwards I was able to get extra long several pairs, some of which i took profits on my long positions and others which i waited and then shorted giving me a great spread price. Bkc/Mcd200 and Cop/Cvx finally turned for me the last hour or so of the day. With the $VIX this high I look to run and gun rather than build positions, tending to daytrade rather than carry overnight pairs positions. Tighter, higher correlated spreads have offered great oppurtunities with high volatility like pbr/pbr.a and rio/rio.pr. The tight spreads offer some safety if you get whipsawed.
More...

Interesting. I too am sticking with intraday trading reversion but I will keep overnight and possibly add to positions if it aligns on a longer timeframe.

It seems that you convert your pair into a directional/dispersion trade should the market start making a move? I think I understand, if you'll humor me:

Say your short XYZ/long ABC, you see the market breaking out, will you add to ABC OR perhaps cover XYZ and re-short at a higher price? Or some other combination? Thanks in advance.
 
johnnysharp,

Thanks for the journal, and best wishes for continued success! I am currently paper trading the signals from Pairtrade Finder, and I wonder if I could bother you for some software details. Have you changed any of the default Preferences settings in the software? Specifically, I'm wondering if you have changed the correlation and standard deviation lookback settings from their default of 100 days? The reason I ask is that I am not getting some of the signals you have posted, although I seem to be looking at some of the same pairs. One more: What criteria did you use when you were deciding what symbols to enter into the system in the beginning? Thanks in advance for any suggestions you might have.

mousejockey
 
You must log in or register to reply here.
Back
Top