Monday, December 20, 2010

Some MarketSci Inspiration and a SPY Trading System

Michael Stokes over at MarketSci wrote a very cool, little post about his professional and educational background and how it got him to where he is today. It was titled "Be Unreasonble", which I found interesting because I can be unreasonable, for good and bad. It was nice to hear that this unreasonableness was a positive factor in his life and that it helped influence his current success. Even though Michael says the post was not meant to be motivational, it was for me.

Much of his story is similar to mine (had a good job, but quit to pursue his passion; late interest in grad school; swing [not day] trader.) Part of what Michael does is develop and sell market timing signals. With the help of TimerSeeds, he has created and marketing timing strategies that are available through a subscription or managed account (see his business webpage). I think this a cool way to do things, and I am very much trying to pursue that path. But I first need to develop a strategy, then create an audited track record (I'm a CPA nerd, so I understand the value of quality accounting) before I can even consider marketing it.

Below is a long/short strategy based around the SPY. The strategy code has fairly basic inputs that attempt to exploit short-term price inefficiencies (it's mean-reversion).  This system is not that crazy, which is good, because the simpler the better. This is the first run, so tweaks are sure to follow. This is an 'all in' trade: when a signal comes in, allocate 100% of your portfolio to the trade. It is from 1/1/2000 through 12/17/2010, starting with $10,000.

I haven't traded this strategy yet, but I have traded other systems that are very similar to this. I believe that this thing can work in real time. I will explain everything about the metrics above in another post.  There are many factors besides compound annual growth % when it comes to evaluating a system. I'll let you know what I'm looking at and my approach to this whole thing.

In the meantime, please comment on what you see so far. Would this be worth anyone's time? Let me know if I'm on the right track, or on crack. ;)



Anonymous said...

Hi Chris,

I just found your blog the other day. Looks like a lot of good ideas being brought to life. I see you are a CPA. I was too (are we ever not CPAs, even when we go on inactive status?).

A few observations:
> I like that you put the average gain %. I would like to have a sense of what you feel is an acceptable threshold for a given system and why (maybe low slippage is a known variable).
> You may want to take a look at David Aronson's book on "Evidenced Based Technical Analysis". I am still looking into practical ways to implement his findings. His most appropriate finding is that we need to take measures to ensure that systems have similar performance after testing as they do during testing (In sample vs Out of Sample, and he covers a whole lot more). It would be interesting to see the In sample/Out of Sample comparisons vs. the Full Periods. I find that a lot of great full periods systems are not the top ranked system in the In Sample periods - which raises the question, how would they be selected ex ante for use during the Out of Sample period. One last comment on this topic: If full period performance is used to pick a system going forward, then it makes sense that we could back test that selection technique. In other words, if you pick the best full period system using data that ended two years ago, how did it perform afterwards? Food for thought.

Keep up the good posts!

mr rolfsson said...

Hi Chris,

I started to follow your blog just recently. The system looks impressive but it would be interesting to see how commission and slippage affects the performance. Also how it performs on a few other index etfs.

But it looks highly interesting and promising to me.


Chris said...

I will have to check out Aronson's book. Commissions and slippage will be addressed soon. I just wanted to create the idea first. I am doing a follow up post. Coming shortly...

Michael S said...

Hello Chris - thanks for the kind words - I've been following you on my RSS reader and look fwd to big things from you in the future.

One off the cuff comment - would be nice to see those charts scaled logarithmically rather than linearly to give a more accurate view of performance (assuming you're compounding returns).


Chris said...


Thanks! I've got a lot to learn from you. There is new a post that has a log equity curve, and my comments about the system. If you have any other comments, I'm open ears.


Mike said...

Great tips . I always love reading post written by you. I read your post till end and found it very informative for me. I am beginner in trading and learned few things from your post. Thanks!

cfd trading training

Post a Comment