Pair Trading Strategy Journal

Quote from jonnysharp:

As per several requests Id thought Id do a performance update.

Start of journal: 18th August, 2008
Cut off date: 22nd April, 2009
Return: 93.73%
Annualized: 139.89%
Biggest Drawdown: 8.9%
SP500 Return: -33.4%
Outperformance: 127.13%

Using a generous risk free rate of return of 3% my sharpe ratio is 15.38 ((139.89-3)/8.9) long term stock market average is 0.40, hedge funds consider above 2.00 to be good and above 4.00 to be exceptional, so Im very happy with my risk-adjusted returns and my absolute return aswell.

Below is my a/c equity curve since starting, you can see a soft patch at the start when the market was very dislocated last year and when I didn't have my current trade filters, since then Ive improved my performance and conservatively increased my trade size.

<img src=http://www.elitetrader.com/vb/attachment.php?s=&postid=2398842>

Also attached is a pdf doc showing all my trades Ive taken exactly as Ive entered them in this journal in real time, no hindsight trades.

I really owe a big thank you to the team at pairtrade finder for sharing an incredible piece of software that actually has worked for me, it has definitely changed my life for the better.

Jonny.
More...


Hi Jonny,

thanks very much for sharing this info, very impressive ! I'm currently struggling a little with PTF at the moment due to inconsistent signals but I will persevere ...

Just wondering if you could share some more info regarding the filters you use. I know you've probably covered them in various places throughout your journal, but it would be nice to get a summary of what you now use

All the best
Ivan
 
Thanks jonny and for your comments re Portfolio Management.
I've had a reply from PTF about this issue and they say they are aware of it and should be upgrading it in V3 later this year.
My own trial is going well with my first trade closing yesterday at a good profit and far exceeding the current 'in trade' deficits.
I still find their interface troublesome and its difficult to find when to close trades if you have a large watchlist because the app doesn't seem to be aware of those trades you're 'in' and those you arent.
Still, as far as finding when to get into a trade it appears to be working OK and that's the nub of it really....
 
Quote from jonnysharp:

As per several requests Id thought Id do a performance update.

Start of journal: 18th August, 2008
Cut off date: 22nd April, 2009
Return: 93.73%
Annualized: 139.89%
Biggest Drawdown: 8.9%
SP500 Return: -33.4%
Outperformance: 127.13%

Using a generous risk free rate of return of 3% my sharpe ratio is 15.38 ((139.89-3)/8.9) long term stock market average is 0.40, hedge funds consider above 2.00 to be good and above 4.00 to be exceptional, so Im very happy with my risk-adjusted returns and my absolute return aswell.

Below is my a/c equity curve since starting, you can see a soft patch at the start when the market was very dislocated last year and when I didn't have my current trade filters, since then Ive improved my performance and conservatively increased my trade size.

<img src=http://www.elitetrader.com/vb/attachment.php?s=&postid=2398842>

Also attached is a pdf doc showing all my trades Ive taken exactly as Ive entered them in this journal in real time, no hindsight trades.

I really owe a big thank you to the team at pairtrade finder for sharing an incredible piece of software that actually has worked for me, it has definitely changed my life for the better.

Jonny.
More...

Jonny, fantastic. Well done.
Is it possible to post in Excel please? I was trying to do an equity curve to get an idea of the drawdown, but the numbers dont copy well from the pdf. I cant see the equity curve you posted for some reason (red cross).

When you say "I didnt have my current trade filters" are you referring to the >0.90 correlation, and one pair per industry filters?

Thanks for everything mate.

Rod
 
Quote from jonnysharp:

...Using a generous risk free rate of return of 3% my sharpe ratio is 15.38 ((139.89-3)/8.9) long term stock market average is 0.40, hedge funds consider above 2.00 to be good and above 4.00 to be exceptional, so Im very happy with my risk-adjusted returns and my absolute return aswell...
Jonny.
More...
FYI, the Sharpe is not a good mesure of performance. It is a well know fact in the performance of hedge funds literature. It only accounts for the two first moments (mean and standard deviation) and therefore assume that the returns are normal (gaussian). I'm sure there is a negative skewness (infrequent big losses) in the returns distribution and the Sharpe does not account for that and also the tail risk. Anyways, it's a first benchmark.
Good job BTW.
 
Filters are;

1) unique industry-only pairs
2) correlation above 80% and rising
3) non-trending ratio chart and good distance between ratio and avg ratio
4) spread near 50 day high if im bearish the spread and vice versa if im bullish

Equity curve attached.
 

Attachments

Quote from total_keops:

FYI, the Sharpe is not a good mesure of performance. It is a well know fact in the performance of hedge funds literature. It only accounts for the two first moments (mean and standard deviation) and therefore assume that the returns are normal (gaussian). I'm sure there is a negative skewness (infrequent big losses) in the returns distribution and the Sharpe does not account for that and also the tail risk. Anyways, it's a first benchmark.
Good job BTW.
More...

Not sure I understand what your saying here, if infrequent big losses were an issue it would show up in the drawdown thus lowering the sharpe ratio, otherwise if it doesn't cause a big drawdown what is the point of identifying it? I find it to be a good indication of risk-adjusted returns since it takes into account what one can make without risk, then the return relative to the volatility, if you know of a better measure for risk-adjusted returns im all ears.
 
Johnny, great job.

For what it's worth i can share some stats from someone else (me) who is trying to do this and is doing well...although not as well. Following more or less the same strategy from PTF (although usually picking different pairs for whatever reason) here is what I have:

Johnny has about a 64% winning percentage and his wins and losses both average around $800 per $10,000 based on around 101 trades (i did not look at the PDF but had created my own list of his trades while studying his approach). Great stats!

I have done 57 trades starting in January. I have a 74% winning percentage and the same loss amount, about $800, but my winners only average about $550 (per $10,000) vs his $800 per trade. I haven't taken the time to study yet why - but i don't filter for volatility and my guess is that i make a higher percentage of winning trades with a lower average win because i'm getting into trades that don't have a high enough volatility for Johnny (since I don't really look at that yet). For instance I've made an SPY/DIA trade that was almost a slam dunk but had a very low return.

So...so far i'm happy. Over 50% IRR in closed trades and should still be > 30% or so once I close one or two bad trades that are out there.

Things I'm going to do to try to improve:
1. Look into my selection to see if i can tell why my average win is so much smaller.
2. Consider not trading a stock if it has an earnings report coming out. One big loser that did not close is a big loser because of an earnings report.
3. Look at stops. I read somewhere that someone suggested an 11 day stop. I can say that almost every significant winner was done within 10 days or so...and my one big loser is bigger than it was after 11 days. I"m going to study to figure out if there is a stop (days or percentage) because right now i am not using stops.

Bottom line...if you read this journal and follow the rules...you should be able to be profitable.
 
Thanks for the comments guys, well done mlsignups on your results too.

Closed trade for profit

Sold ACGL @ 56.06
Covered ACE @ 43.99

New trades

Long MGG @ 18.85
Short TPP @ 26.12

Long TMK @ 26.28
Short MET @ 28.20
 

Attachments

Does anyone know where intra day data would be available for pairs and can be formatted in either price bars or candlesticks? If the source is free that's better.

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