ASX Stock Pairs Trade Journal

[tigger36]

Sorry I didn't see these until today. How many pairs do you have to get only 10 signals a month? Most of my pairs will produce anything like 10-15 signals a year, so with ~300 pairs in the data base I get 10+ signals per day.

My criteria are similar, except I am not that strict with correlations. To me the correlation has to go apart in order to have any divergence, so I am not that worried if correlation was 50-60% 2 months ago. Increasing correlation also means nothing to me. It's really not a share price chart and no reason to believe trends are likely to continue. The 20day SMA also feels unimportant... if 20day SMA is that useful as an indicator then you might as well do directional trades with that. Lastly I scrutinise the historical profit, and take out anything that's out of whack due to announcements, capital raising etc. Basically anything that's over 10% PnL are probably trades that shouldn't have been taken anyway.

The other trick is to close out the pairs and take profits quickly. Before I enter a pair I work out what's the potential profit if the pair goes back towards the mean straight away. And I will take profit if I can get 70-80% of that full potential amount. Reason being that the MA will move everyday and they generally move in the opposite direction of the ratio. So why wait for the exit signal for another 2 days when that signal will produce a profit similar to what you have today?



I know nothing about FTSE350 sectors so can't comment. But when I say I try to know my stocks intimately I mean really intimately. Two stocks might be both retailers but one has more sales in the US and is hurt by the higher $AUD. So unless I believe $AUD is due for a fall I am not going to long that stock straight away... wait a day or two before entering can often improve your entry, even when the std dev is no longer at 2.0.

And with mining companies they should mine the same stuff at a minimum...

skc,
your advice is much appreciated. I have spent some time re-evaluating my selection criteria and I estimate I should now get about 2-3 signals per day. I think the FTSE350 industries might be a bit more diverse than ASX as we have 36 sectors and when I split those down sorting out different miners etc.. as per your advice I have about 50 industries from which I take my pairs. So I suspect I will get fewer signals than your setup, but I am satisfied with 2-3 per day.

I also like your suggestion of taking profits quickly, I have added an algorithm like you say to my automated exit signaling, and initial backtesting shows it should achieve improvements.

I will let you know how my next month's trading goes...

Many thanks for your help

 

[tigger36]

Pair Trading not going to well at the moment, hopefully just one of those temporary drawdowns, had 2 big losers in a row...

One thing i've noticed is pairs where one of the pair is in a narrow trading range is generally not a good pair. The reasoning being, i've noticed if the market suddenly moves up, that narrow range can turn into a huge breakout, which the pair stock might not move in pace with. Also conversely, it can also mean if the market moves, that this stock stays in its narrow range, and the other stock follows the market... both ways lead to a loss!
Anyone else observed this?

 

[skc]

Pair Trading not going to well at the moment, hopefully just one of those temporary drawdowns, had 2 big losers in a row...

One thing i've noticed is pairs where one of the pair is in a narrow trading range is generally not a good pair. The reasoning being, i've noticed if the market suddenly moves up, that narrow range can turn into a huge breakout, which the pair stock might not move in pace with. Also conversely, it can also mean if the market moves, that this stock stays in its narrow range, and the other stock follows the market... both ways lead to a loss!
Anyone else observed this?

Actually narrow trading range is a great pair trading opportunity in my experience. Usually one leg of the pair have made a breakout, while the other is still in its narrow range. You short the breakout and long the consolidation.

If market keeps rising, the consolidation will end in a breakout while the breakout consolidates - a perfect trade.

If the market tanks, the consolidation won't breakout and may hold there abouts, while the breakout turn out to be a false break, other breakout traders got their stops hit and the share falls by 8% over 2 sessions - another perfect trade.

If the market doesn't do anything - then your pair may not do anything either.

A recent example comes to mind is short APN long WAN on ASX. APN went for a run Wed morning made a new high at 2.61 against WAN that's been flat around $8-$8.1 mark. The pair stretched to well above 2.5 std dev. Their prices today are $2.32 and $7.9. Anywhere between 6-10% gain depending on where you got in.

Of course you need to be sure that the breakout was not news related...

 

[tigger36]

Actually narrow trading range is a great pair trading opportunity in my experience. Usually one leg of the pair have made a breakout, while the other is still in its narrow range. You short the breakout and long the consolidation.

If market keeps rising, the consolidation will end in a breakout while the breakout consolidates - a perfect trade.

If the market tanks, the consolidation won't breakout and may hold there abouts, while the breakout turn out to be a false break, other breakout traders got their stops hit and the share falls by 8% over 2 sessions - another perfect trade.

If the market doesn't do anything - then your pair may not do anything either.

A recent example comes to mind is short APN long WAN on ASX. APN went for a run Wed morning made a new high at 2.61 against WAN that's been flat around $8-$8.1 mark. The pair stretched to well above 2.5 std dev. Their prices today are $2.32 and $7.9. Anywhere between 6-10% gain depending on where you got in.

Of course you need to be sure that the breakout was not news related...

Sorry I didn't explain very weel. What I meant to say is an entry signal has been generated (ie.2 stdev from mean etc..), but NEITHER has broken out of a range. One stock is in a range, the other stock has meandered downwards say to cause the signal. Hence we Short the stock in range, and Long the stock that has meandered downwards. Then the Market has a big surge upwards, the in range stock then decides to breakout upwards in your trade... big loss...!

I think having just re-read your post from "2nd Oct", you do agree, as you said exactly this: Quote:
"- I sometimes read the individual share charts before entering, so I will happily long a share that looks like ending a downward correction, or avoid shorting a share that is positioning for a potential breakout."

So thanks, your previous post i've just re-discovered is very useful!

 

[tigger36]

Actually narrow trading range is a great pair trading opportunity in my experience. Usually one leg of the pair have made a breakout, while the other is still in its narrow range. You short the breakout and long the consolidation.

If market keeps rising, the consolidation will end in a breakout while the breakout consolidates - a perfect trade.

If the market tanks, the consolidation won't breakout and may hold there abouts, while the breakout turn out to be a false break, other breakout traders got their stops hit and the share falls by 8% over 2 sessions - another perfect trade.

If the market doesn't do anything - then your pair may not do anything either.

A recent example comes to mind is short APN long WAN on ASX. APN went for a run Wed morning made a new high at 2.61 against WAN that's been flat around $8-$8.1 mark. The pair stretched to well above 2.5 std dev. Their prices today are $2.32 and $7.9. Anywhere between 6-10% gain depending on where you got in.

Of course you need to be sure that the breakout was not news related...

re. trading a breakout, yes i think these can be good trades, although i've discovered that timing the entry needs care, as yes the breakout was not news related, but driven by those investors chasing breakouts on volume..hence if you enter too soon the breakout may still have a long way to go, especially in a bullish market. I've discovered this to my loss on one occasion, I entered and that breakout stock just kept on going upwards....

 

[skc]

Sorry I didn't explain very weel. What I meant to say is an entry signal has been generated (ie.2 stdev from mean etc..), but NEITHER has broken out of a range. One stock is in a range, the other stock has meandered downwards say to cause the signal. Hence we Short the stock in range, and Long the stock that has meandered downwards. Then the Market has a big surge upwards, the in range stock then decides to breakout upwards in your trade... big loss...!

I think having just re-read your post from "2nd Oct", you do agree, as you said exactly this: Quote:
"- I sometimes read the individual share charts before entering, so I will happily long a share that looks like ending a downward correction, or avoid shorting a share that is positioning for a potential breakout."

So thanks, your previous post i've just re-discovered is very useful!

re. trading a breakout, yes i think these can be good trades, although i've discovered that timing the entry needs care, as yes the breakout was not news related, but driven by those investors chasing breakouts on volume..hence if you enter too soon the breakout may still have a long way to go, especially in a bullish market. I've discovered this to my loss on one occasion, I entered and that breakout stock just kept on going upwards....

All very true. I re-discover my own posts all the time!

Timing the breakouts are very difficult indeed. The APN example was a breakout that lasted only 2 hours, while others I can recall going for 2-3 days. I have entered the short leg way early on many occasions. Those trades are rarely profitable as a result.

In the end, however, breakout consideration is something that you can't get right all the time - otherwise you might as well just be a directional breakout trader

In situations like these some layering might help, although that's not a real part of my strategy yet.

 

[bochman]

would anyone consider long FXJ & short APN as a pair?
45 days seem to be reaching stddev 2. 60 days and 1 year suggest otherwise.

 

[skc]

would anyone consider long FXJ & short APN as a pair?
45 days seem to be reaching stddev 2. 60 days and 1 year suggest otherwise.

The short term signal to short APN / long FXJ is definitely on. Depending on your holding time horizon the 1 year data may or may not be relevant.

APN/FXJ has been a good consistent pair for me. I rarely hold more than a week and don't usually care about 1 year as a result.

 

[bochman]

what time span do you use normally? 45 days? 50? 60? thanks!

 

[skc]

what time span do you use normally? 45 days? 50? 60? thanks!

Std dev of last 20 days

MA of ratio of last 14 days

 

[n5032245]

skc,

the pair trade software comes installed with a lookback of 14 days for the std dev. what made you change to 20 days?

 

[skc]

skc,

the pair trade software comes installed with a lookback of 14 days for the std dev. what made you change to 20 days?

With longer std dev look back you filter out some signals that may occur due to correlation being too good for the pair over the short term. The std dev becomes so small that 2 std dev away from the mean may only be a deviation of 1-2% and may not be a profitable trade.

In earlier version of the software it wasn't as easy to calculate the potential profit from each signal, so I increased the std dev look back to reduce those "false positives". It is probably less of an issue now, but I left it as 20 days has worked well for me. It wasn't a number that I came up with through back test or anything like that.

 

[FXanyone]

Im curious to know if people are using commerical software for pairs trading or building their own speadsheets?

Thanks for your help

 

[dr3wfish]

Im curious to know if people are using commerical software for pairs trading or building their own speadsheets?

Thanks for your help

i'm using pairtrade finder however this is only a temporary measure while i finish off building my own software.

pairtrade finder pros:
- provides a good starting point for backtesting pairs
- once you have a basket of pairs, the console provides a nice simple overview of how each pair is tracking

pairtrade finder cons:
- horribly buggy (like incredibly bad)
- default yahoo data feed is inaccurate. can only be replaced with iq data feed and i don't believe they do australian equities
- does no data cleaning whatsoever ie you will find lots of stocks with missing price snapshots for a given day; this affects backtesting etc.
- to hard to edit stock lists, sectors, different exchanges, pairs.
- slow

i have found it useful getting my pair trading started but i've definitely outgrown it. in fact, grew so tired of the data problems that i wrote my own tool to replace the data it imports and now i pretty much use it just for the pretty graphs.

i need software that:
- uses reliable data feeds
- cleans the data
- allows manual import of data ie eod
- provides more options for backtesting ie instead of just exiting a position when the pairs revert to mean, add clause to exit when reverted to mean or 10 days have expired or maximum drawdown is 2% etc..
- provides a proper, usable UI.
- advanced alerting ie incorporate checking for news events when alerting

suggest you download the trial and test it out because it will help you on your journey, but i can't recommend buying it (thats just my personal take, others may disagree). build a spreadsheet or if you can code build, some software.

oh, before i forget, i looked at lots of so called "pairs trading" software out there and its mostly (it not all) rubbish. don't bother with it.

 

[skc]

i'm using pairtrade finder however this is only a temporary measure while i finish off building my own software.

pairtrade finder pros:
- provides a good starting point for backtesting pairs
- once you have a basket of pairs, the console provides a nice simple overview of how each pair is tracking

pairtrade finder cons:
- horribly buggy (like incredibly bad)
- default yahoo data feed is inaccurate. can only be replaced with iq data feed and i don't believe they do australian equities
- does no data cleaning whatsoever ie you will find lots of stocks with missing price snapshots for a given day; this affects backtesting etc.
- to hard to edit stock lists, sectors, different exchanges, pairs.
- slow

i have found it useful getting my pair trading started but i've definitely outgrown it. in fact, grew so tired of the data problems that i wrote my own tool to replace the data it imports and now i pretty much use it just for the pretty graphs.

I use the pair trade finder myself and agree with most if not all of the comments above. However, I still use it as my main tool because I don't have much coding skill to do a better job. But more importantly I think the smarts of the system lies with the trader and not in the software. I am not those fully mechanism / system / quants type traders so I am not really concerned with optimising the last parameter.

Having said that, if the software would crash less and allow for better data than it would be a huge positive.

 

[tigger36]

I use the pair trade finder myself and agree with most if not all of the comments above. However, I still use it as my main tool because I don't have much coding skill to do a better job. But more importantly I think the smarts of the system lies with the trader and not in the software. I am not those fully mechanism / system / quants type traders so I am not really concerned with optimising the last parameter.

Having said that, if the software would crash less and allow for better data than it would be a huge positive.

I agree also.. I trialed pair trade finder, just to see what it was like, then I wrote my own software that has alot of the extras pointed out above, earnings checking, advanced exit criteria (profit taking exit, exit slope off towards 10 days, etc..), and advanced back testing.

I also agree with skc's last comment, on the real difference comes with the trader. I'm learning alot from skc's comments, and am improving my pair selection and exit criteria, the software just gives me a cut down set to choose from to start with...

 

[keroppi]

I have been having a lot of problems with the pair trade finder software. (just sent you a message tigger36)

I find it extremely buggy and slow. The list of pairs does not even load up for me anymore.

I tried developing something basic on Excel but it takes too long to retrieve the data for one pair, let alone 50 pairs. If someone can send me a copy of their software that actually works I would really appreciate it. My email is bzltop@gmail.com

 

[tigger36]

I have been having a lot of problems with the pair trade finder software. (just sent you a message tigger36)

I find it extremely buggy and slow. The list of pairs does not even load up for me anymore.

I tried developing something basic on Excel but it takes too long to retrieve the data for one pair, let alone 50 pairs. If someone can send me a copy of their software that actually works I would really appreciate it. My email is bzltop@gmail.com

Hi, I have the advantage of being a Software Engineer by profession, and I know how early releases of software can be buggy, especially if not put through rigorous System Test. I would hope the next release of PTF which I think is due soon, if I read their web site correctly, should provide some improvement if they've put it through some more testing...
My program I have written in my spare time is fairly crude and has bugs in it too! (I know it does as I keep fixing them!), it is very specific to my requirements and my environment.

If you're logically minded and willing to put some time and effort in try taking a look at Sun's java tutorial, that will get you going in the world of Java programming. (http://java.sun.com/docs/books/tutorial/)

and lastly, I can "Program", I just wish I could "Trade" !! things not going too well on that front at the moment for me...

 

[Freddy]

Well, I'm now up 80% since August 2009 with a captial of $5000. Took a hit last week with WTF and I am still up. But HVN is giving me the problem now where I shorted it at 4.01.

Yes, the software is buggy and you get Enters such as IFL/PTM the other day and the next day it disappera off the screen.

Does anyone have a better price feed than Yahoo - it is terrible?

 

[skc]

Well, I'm now up 80% since August 2009 with a captial of $5000. Took a hit last week with WTF and I am still up. But HVN is giving me the problem now where I shorted it at 4.01.

Yes, the software is buggy and you get Enters such as IFL/PTM the other day and the next day it disappera off the screen.

Does anyone have a better price feed than Yahoo - it is terrible?

Great trading Freddy.

HVN last traded at $4.01 about a month ago. Consider using a time stop.

BTW, I would never short WTF. They will release updates on irregular basis and the results could swing the price a fair bit, with a bias to the upside because it's a market darling that actually has good fundamentals.