Archive for May 21st, 2008

Just a reminder to review this post on VIXandMore on the VIX/VXV as a bear/sell indicator – it hit it right on the head.  Nice job Bill!  Now I just need to find some time to finish up the rest of the study I promised to do in the comments!

Read Full Post »

System design is very interesting and intellectually challenging.  But for me, personally, the biggest challenge comes when I actually put a system to work.  I think there are key steps that system designers need to take in order to have confidence in their system:

  • Test the system: While you’re testing the system, make sure you’re looking closely at MAE (Maximum Adverse Excursion) and MFE (Maximum Favorable Excursion).  These two statistics will give you a somewhat realistic view of how the trade will “feel” prior to going live.  A system, for instance, might have great returns, but if the MAE is going to make your stomach churn, then you can avoid the system or make changes to bring it into line.  The other stat I like to look at in this area is the Ulcer Index/Performance Index.  Lastly, I like to look at the Expectancy of the system very closely.
  • Paper Trade the system: Here’s the trick with paper trading – keeping yourself honest.  It is far too easy to do this by not recording your trades in some way.  You also need to do this in real-time.  Meaning, you shouldn’t be looking back at the end of the day and deciding what the trade was.  I also try and make my paper trading more realistic by assume a bad fill – so rather than assuming I manage on getting the bid price, I’ll often assume that I got the ask and then some.  Good trading platforms also allow you to program in slippage – make sure you’re using this feature.
  • Going live: Ok, you’ve done all the work, now you’re ready to go live.  I start with a very small amount of money for the strategy.  Options can sometimes be useful here because you can test for relatively small amounts of money, and you know what you’re likely to lose (assuming you’re buying an option and not selling one).  Obviously if you’re using options, the results will be different – so you can’t use it for the equity curve.
  • Review your results vs. your backtest: One thing that I find interesting to do, once you’ve got a week or so of actual trading under your belt, is to go back, run the backtester again and see how your results varied (or didn’t vary) from the backtest.  This gives me what the website C2 (http://www.collective2.com) calls a “realism” factor.

More than anything else – don’t get discouraged by poor results.  If you’ve done it right, you’ve only burned a small amount of capital while you tested the idea.  The way I look at this money is that it is your investment in “product design” – so I think it is a small price to pay for developing a great system.

Read Full Post »

MR1 Update

After days of being out of the market (due to the rally where it really doesn’t trade much), my MR1 strategy just went long XLF at the open this morning – I did not have a chance to enter the position at the open because of a meeting – but I’m taking the position now.  Obviously we’re down a bit on the position but it is important to realize that we’re counting on it mean-reverting.

Also note that I’m buying a call on XLF (in-the-money).  With InteractiveBroker’s support for penny pricing, it’s just great.

Read Full Post »

Need to thank Rob at Quantifiable Edge for mentioning this blog this morning.  And I also need to thank the Dog at Dogwood for a similar mention.  At this rate, I should have close to 3, count’em, 3 regular readers.  Roman, as they say, was not built in a day.

In an unrelated thought – I’ve been considering putting together a Quant trading meet-up in the MA/CT area.  We’d get together, have drinks and a dinner and get a chance to know each other.  If you’ve got some interest and are in the area – please let me know – I currently have two other people (other than myself) interested in doing it.

Read Full Post »