Market Timing Models

The recent sell-off and extremely high correlation in the overall market has given me a renewed interest in market timing models. When correlations are high, stat arb spreads have a tendency to do weird things. Similarly with vol arb strategies. William O’Neill and IBD were my first introduction into market-timing, aka trend-following the index. However IBD’s timing seemed flawed to me and the more I read and researched the less validity I assigned it. It wasn’t until I read Gil Morales/Chris Katcher book that I regained an interest in models to time the SPX. Even then after some research I was skeptical. Mark Minverini is a trader I respect and he has recently begun publishing a market timing model as well.

These models are trend following in nature and hence buying strength and unless I am mistaken ALL utilize market breadth to confirm an uptrend. Some utilize sell signals but most of the gains are from the long side. My first exposure to Market Breadth was in a great book “Winning on Wall Street” by Martin Zweig where he describes a Zweig Upthrust and the corresponding statistics that follow. The book is worth a read but the systems utilized are probably a little outdated.

So utilizing published buy and sell signals from Morales/Katcher, Minervini, etc I set about creating my own market timing model. My goal was just to get close, not straight replication, of their models in order to vary position sizing and exposure of various signals from my other models. If my hypothesis that quantitative systems tend to suffer in extreme volatility, this should act as risk reduction and lower volatility of the overall equity curve. And MAYBE, I could use it directionally as well? Here’s the results from the first pass through:

Looks pretty bad up until 2010 and then it really starts taking off. Good example of recency bias on my part. Nonetheless, it’s not a complete loss.If we look visually at the signals:

 

From the statistics, we have about ~40% percent probability and 1.65 avg win to avg loss. So we’re catching some nice trends with better then 1/3 chance. Not great but given the recent performance and with the notion that this model is intended for risk management and position sizing as opposed to straight alpha generation, it could be a useful tool. I’ll plug at it a little more this weekend and perhaps share signals if there is any interest. Incidentally the model is on a buy since 8/24 and about even on it right now.

Advertisements

~ by largecaptrader on August 28, 2011.

2 Responses to “Market Timing Models”

  1. […] Market Timing Model also started off slow with some big whipsaws but looks like performance is […]

  2. […] experimented with a few breadth models in the past. My hypothesis is that markets have become so efficient and correlated, that all […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

 
%d bloggers like this: