Pairs Trading

[TraDaToR]

Quote from stock_trad3r:

The problem is everyone is already doing this

A system works great...when there aren't 1000 other people copying it

Whenever there is a quite profitable, logical strategy around, you will always have people saying that it is dead because of the number of guys already doing it.

Pair trading has always been around and of course the biggest winning pairs have been exhausted or arbed so much that it is not possible anymore for a retail to be profitable trading it.

But if you search deeper, there is always some unconventional pairs to trade. And it can be rewarding.


Itmediaco, on the level of correlation needed to create a pair strategy, I think the problem in stocks is that the correlation is quite unstable. Futures correlation are more stable. There are pairs that have 90 % correl for a long time. That is just an impression, no real study.

 

[Yisterwald]

Quote from TraDaToR:

Pair trading has always been around....

Heh -- indeed. When my partner and I started doing it we thought we had discovered fire. There was no popular literature on it, no web sites dedicated to it, no free, quickie stock correlation utilities available on the net. We thought we had cracked some secret code, and we held tight to every tidbit of information about what we were up to. We were shocked -- shocked, I say! -- when eventually we began to overhear other traders in the elevators at the CME talking about some of the same trades we were in. Eventually we figured out that we hadn't discovered fire after all, that the strategy was actually growing rapidly in popularity. A couple of changes in the marketplace made it quite accessible -- the growing amount of free stock research available on the net, and the growing availability of significant overnight leverage. A strategy that once had some barriers to entry was now a piece of cake.

But if you search deeper, there is always some unconventional pairs to trade. And it can be rewarding.

Yep. Don Bright seems quite impressed with the PairCo guys maintaining data on 1000 or so pairs, but that really isa piece of cake, assuming you're willing to put in the time to find them, and if you're not then you probably aren't willing to do the other work trading requires anyway. It's self-sorting. Furthermore, you need to have that many at your disposal once you start filtering out unattractive trades based on your fundamental and technical criteria.

Itmediaco, on the level of correlation needed to create a pair strategy, I think the problem in stocks is that the correlation is quite unstable. Futures correlation are more stable. There are pairs that have 90 % correl for a long time. That is just an impression, no real study.

Agreed here also. Stocks correlate until they don't. Analysis of past correlation doesn't replace loss control.

 

[jrlvnv]

Quote from gkishot:

Does anyone know of a good free software for charting pair spreads?

Anyone have the answer for this?

 

[Don Bright]

Quote from jrlvnv:

Anyone have the answer for this?

It depends on what you call "tracking" ....I could give you an excel spreadsheet that works with a dde link that our guys use to monitor buy and sell points (what we call Portfolio Watcher Basic), in which you enter the spread ratios and it keeps a constant update of spread prices.

Our Redi platform can be programmed to back test ratio spreads.

I'm told that even Yahoo finance has a somewhat limited, but useable feature as well (I have not seen it myself, however).

I will check with some of my guys for you all, and get back to you.

Don

 

[stereo70]

Don Bright wrote (and I edited w/ bold):

By having pretty elaborate spread sheets,and deep, deep pockets an individual can actually manage over 100 pairs, as my brother does.

 

[Don Bright]

Quote from stock_trad3r:

The problem is everyone is already doing this

A system works great...when there aren't 1000 other people copying it

Again, this depends on what you call "everybody" ...it takes a few $million to carry 40 or 50 pairs and to trade them effectively. I have been told that the Opening Only strategy will become dilluted for decades, but since retail traders cannot engage in it, the overall effectiveness has stayed intact.

There are 1000's of pairs, and several methods involved, and we still find that we are at the tip of the iceburg, so to speak.

All I can do is relflect what my family overall, my brother especially, and our PairCo guys are doing.

But, as always, this strategy is not for everyone, and unfortunately (or fortunately, depending on which side of the fence your're on, LOL), most don't get into the details enough to engage in the strategy properly.

Don

 

[Yisterwald]

Quote from gkishot:

Does anyone know of a good free software for charting pair spreads?

For a low-cost solution you're probably best off with Excel. I haven't heard of another good, cheap solution for charting pairs, but that doesn't mean there isn't one out there. Perhaps someone will chime in.

If price is no object, CQG does a bang-up job of charting stock pairs along with whatever technicals you'd like to apply to them. The last time I used it for that purpose was well over a year ago, and at that time it was most convenient to use line charts. But CQG is always improving their product and bar charts may be easier to use with pairs now. With CQG you can also set alerts for your pairs at price levels that merit your attention.

It's pricey, but I wouldn't want to trade pairs without it. Don Bright might be able to point you to alternatives.

 

[Yisterwald]

Quote from Don Bright:

I have been told that the Opening Only strategy will become dilluted for decades, but since retail traders cannot engage in it, the overall effectiveness has stayed intact.

I'm told it has become diluted, though not rendered entirely ineffective. Isn't it fair to say that today the Opening Only strategy doesn't produce the returns, with the ease, that it did even 5 years ago?

 

[FullyArticulate]

Quote from gkishot:

Does anyone know of a good free software for charting pair spreads?

IB's TraderWorkstation can do this, although that's not really free.

stockcharts.com claims they might do it in the future.

http://stockcharts.com/help/doku.php?id=support:working_with_sharpch#can_i_create_spread_

Here's how I chart the pairs. I generally look at them as line charts:


Although I can also plot them as bar charts to look at suitability for intraday trading and reasonable stops:


The bars are generated by looking at every tick for both stocks as far back as DTN's feed goes and correlating them for time. (The empty bars are beyond the limit of DTN's feed)

As I mentioned earlier, note the Feb 27th day is not really much of a change in the spread. The move starting on December 1st was at least partly due to BAC's CFO resigning and rumors Citigroup was going to make massive changes to restructure. The move lasted nearly a month, and almost entirely in a straight line. These sort of trends are common in spreads and the breakout was clear and obvious.

Yes, you could have just bought Citigroup and not messed around with spreading it against BAC, but the profit would have been identical, and spreading it against BAC gives you protection if the market as a whole goes down. Buying C at the start of a broad market decline might have been a loser. Buying it and selling BAC makes broad market moves irrelevant.

 

[GGSAE]

Quote from Yisterwald:




Yep. Don Bright seems quite impressed with the PairCo guys maintaining data on 1000 or so pairs, but that really isa piece of cake, assuming you're willing to put in the time to find them, and if you're not then you probably aren't willing to do the other work trading requires anyway. It's self-sorting. Furthermore, you need to have that many at your disposal once you start filtering out unattractive trades based on your fundamental and technical criteria.


Agreed here also. Stocks correlate until they don't. Analysis of past correlation doesn't replace loss control.

You make some very good points, i'm glad you've contributed to this thread.

Even with 1400 pairs in the pairco database, i've researched a couple on my own that i'm interested in trading that i've recently asked to have added...i hate to sound like a broken record, but if you want to succeed at this you really have to work at it. I'm continually evaulating my portfolio of pairs, making additions/substitutions as the fundamentals change in the stocks of each pair.

As it was touched upon when building positions, winners/losers, ect. Actually implementing the trades and trading the pairs that are going against you is the biggest challenge in pairs. There is a lot of subjectivity involved, you should be flexible in adding/removing capital and adjusting trade size and also scaling in/out. I don't think i agree with the 80-90% win ratio minimum to succeed, i think it depends a lot more on the individual skill in handling those rare events that throw you a curve ball.