Wednesday, June 28, 2006

Quote of the Week - Letting Go

"The difficulty lies, not in the new ideas, but escaping the old ones, which ramify, for those brought up as most of us have been, into every corner of our minds." -- John Maynard Keynes

This past weekend I wrapped up a new system I've been working on for several months now.  The sad part is it replaces all the current systems I trade.  So, I'm in the process of closing down my existing systems in order to begin trading this new one.

This kinda stuff is never easy.  One in particular has been very hard to let go.  It was the first system I developed back in 2001.  Named it after my daughter.  This new system has been named after my son.  Go figure.

One important change I have made is trading from an end of week basis to an end of month basis.  The backtesting has gone very well...but the forward testing is ongoing.  If this works out well...I may even push out to a quarterly basis.  Time will tell.

The interesting aspect of this system is it curtails nicely with the recent post by acrary here.  While I'm not anywhere close to what acrary has discovered...I too have found certain slices of the market where specific strategies work well.  And as embarassing as it is to say...all my systems I have built over the past five to six years are trying to capture the same market characteristic.  So, this monthly system really is just a simplification of all my weekly systems targetted at a very specific market slice.

What are my next goals?  Well, I have two...

1)  Figure out a strategy for the other side of the market coin.  The area I have yet to develop a viable system.  This should hopefully increase my rate of return while reducing my risk.  Heck, even if its a net loser...may still reduce my risk.

2)  Begin designing a backtesting engine.  I've done a lot of research over the past few days and had some help from a few technical gurus here at work.  I believe I've got a platform framework in mind.  Surprise, surprise...most of it will be done in Python.  Still much design work to do and testing.  Question for you Python guys and gals...any experience using Pytables?  That's what I'm considering for the time series data store.  Any feedback on Pytables would be much appreciated.

That's it here from a short-timer.  Only have a few days left at my current job before I move away from the great state of Texas.  There will be lots to miss but hopefully much to gain up in my new state of Missouri.

Later Trades,

MT

2 comments:

mike said...

I don't know if your model and backtesting uses technical data only or fundemental data also but I love AmiBroker for building and testing END of Day (or week or month) trading strategies.

It is very easy to build complex systems and you can grab data going back a long way for free.

Hope this helps.

Also, out of curiosity, I am fighting to decide if I want to focus on an intermediate strategy (monthly, weekly) or shorter term (daily, intraday). Can you share what sort of return you'd hope to make with your new monthly strategy while in the market?

Michael Taylor said...

My model only uses technical data.

Yes, I've worked with Amibroker in the past. Nice platform. But, it's been a few years since I've last checked it out.

Monthly/Weekly versus daily/intraday...whoa..that's tough. I'll just say I started out intraday and slowing made my way to monthly. And now, I'm beginning to see the value in trading across all time intervals. But, I'm still far from that goal.

Type of returns? Well, I've been very fortunate to hit 20%+ returns each year since system trading (without margin). But, this year, so far, is below par. And I do think those 20%+ returns are above par. I expect when all is said and done if things continue to go well...that I'll generate 15%+ returns annualized as a system trader. And I think that's something most investors/traders should strive for.

Of course, everything changes when you're dealing with margin. And that's something I have yet to try out.

Good luck with your systems!

MT