Tuesday, December 22, 2009

KD's Year In Review: Part 1: Isaac Newton, Mean Reversion, and Momentum

I can't stop thinking about how Newton's First Law applies to the big momentum move we've seen in the market this year:

"An object at rest tends to stay at rest and an object in motion tends to stay in motion with the same speed and in the same direction unless acted upon by an unbalanced force."

I worked on the sell side of Wall Street for 5 years, culminating in my shift to the buy side in May, 2005.  Although I traded, literally, tens of billions of shares of stock on the sell side, it was a presentation early on in my stint on the buy side that made the most lasting impression on me.  (quick side note:  sell side = Wall Street broker/dealers, buy side = money managers: sometimes internalized at those same Wall Street broker/dealers, like in my case)

The presentation was from Jeff Degraaf, who was at that time the head technical analyst at Lehman Brothers.  While my sell side trading desk had basically made the bulk of our profits off of mean reversion - exploiting small inefficiencies in the market that resulted in deviations from the norm, Degraaf explained the key difference between mean reversion strategies (think: buy low, sell high) and momentum strategies (think: buy stocks making new highs, sell stocks making new lows).  If a stock has already rallied strongly, we fear we might have missed the move already, and are generally pre-disposed to want to sell it.

First of all, it's important to realize that by nature, almost EVERYONE is a mean reverter - we're hard wired to want to buy bargains, such as stocks that have been walloped by the market, and sell spikes - stocks that have recently rallied heartily.  Now, don't be intimidated by mathematical terms, none of this is rocket science, but the conclusion is enlightening.

In a basic mean reversion strategy, one would sell stocks that have rallied a given amount in a certain time frame - say, stocks that have rallied 5% in the last 30 days.  In a momentum based strategy, on the other hand, your model would have you buy stocks that are already moving up - it might not be buying the same stocks that the mean reversion strategy tells you to sell, but you're not buying stocks that are falling - you're buying stocks that are already showing upward momentum (and selling stocks that are falling, of course).

Now, Degraaf's presentation illustrated distribution curves showing return profiles for each strategy.  Each curve looked nearly normal in shape (a bell curve), but there were two key differences.  First - in the mean reversion strategy, the mode (most common result - the peak of the bell curve) of the return distribution was slightly positive - mostly your trades make a little bit of money.   On the momentum strategy, the mode of the return distribution was slightly negative - the bell curve was centered slightly to the left of zero - most of your trades actually lose a little bit of money.

The key, however, came in the tails of the distribution - the multi-sigma events - the Black Swans so to speak.  In the mean reversion method, the upside tails were basically nonexistant, which is understandable.  If you're selling stocks that rally, and buying stocks that are falling, the only way you have a massive positive return is when you make the one trade at the turning point of a bull or bear market:  ie, when you short at the top, or buy at the bottom.  Unfortunately, these trades only happen once - you only buy the bottom once, and in the mean reversion strategy, you probably sell far too early, once you've captured some kind of rally - because you expect a reversion to the mean! You expect the rally you've captured to end.

The momentum strategy, on the other hand, has fat positive tails - there is a much larger than expected frequency of large positive returns than you'd expect under a normal distribution.  Any time there is a big trend in the market, you capture the vast majority of it with the momentum model.  Most of the time, when the market churns around (as it has for the last several weeks), you're losing money, slowly, on each and every trade - but when the market establishes a trend (like it did for 6 months earlier this year) you're on board and profiting from the entire move.

That's it - it's not a difficult concept, but it goes against most people's trading instincts - we don't like to buy stocks that have already rallied, and we don't like to sell stocks that have already fallen - our instinct is to do the opposite - buy low, sell high.

So that's why I mentioned Newton's First Law of Motion to describe the big move in stocks this year.  An object in motion tends to stay in motion - like the stock market since the march lows - until an outside force acts on it.  I've been thinking that the outside force will be "reality" - the reality that things are not really getting better - that we (As a country) have pretty much spent all the money we have and all the money we borrowed already.  The reality that unemployment overrides existing home sales in terms of importance in the economy, and that you can't tax and spend your way back to prosperity.  Perhaps the outside force will be the eventual inevitable Federal Reserve interest rate hike needed to remove the massive fiscal stimulus that's been enacted.

I currently have very little idea of where the stock market is going.  My brain tells me that reality is clearly lower, but if the SPX breaks out of its recently established trading range, I'll have the words of Jeff Degraaf and Isaac Newton screaming in my ears: this object is in motion, and objects in motion remain in motion...



Anonymous said...

The guys name is DeGraaf, not DeGraff.

Kid Dynamite said...

oops - thanks for the correction

Anonymous said...

Happy Holidays Kid! Thanks for the great posts this year. They're always thought provoking and enlightening. As far as I can tell, nothing models this market very well. It is what it is until it isn't. As an alternative to Classical Mechanics I might suggest "dssipative systems". Still not much help with investing but it does include the constant reminder that the ice we're standing on is of unknown thickness.

From Wikipedia - "A dissipative system is characterized by the spontaneous appearance of symmetry breaking (anisotropy) and the formation of complex, sometimes chaotic, structures where interacting particles exhibit long range correlations. The term dissipative structure was coined by Russian-Belgian physical chemist Ilya Prigogine, who was awarded the Nobel Prize in Chemistry in 1977 for his pioneering work on these structures."

GS751 said...

Happy Holidays. I would love to do some analysis on different strategies in the SPX e.g. take all stocks that have rallied 10% short em and close the trade in a week or a specified period of time. Take all stocks that have fallen 10% and buy em and close the trade in the same specified period of time. You could adjust all the inputs but would be interesting.

BrendoD said...

Sweet Sassy Molassy!

I just finished a undergrad biophysics class...and the whole time I was thinking, how can I ever compete with these great physicists?

Now if only I could apply newtons 3rd law to this rally....Will the force applied by the Gov be equal to resulting force applied down? $

David Merkel said...

Good post, KD, from one of your fans. My only difficulty comes when the this becomes so well known that it ceases to work.

EconomicDisconnect said...

Merry Christmas everyone!

Anonymous said...

I need to re-read this. I need to think about the lesson of switching sides in my outlook.

Anonymous said...

i can;t believe someone linked to this nonsense

good luck