Monday, February 29, 2016

Using Intra-Day Models to Improve Tactical Asset Allocation

Just about any form of tactical asset allocation (TAA) will work extremely well in a straight up market and a straight down market.  However, many methodologies will struggle in a choppy market that either has no real trend or that moves so quickly that a TAA strategy just doesn't have time to adjust.  In the past I have written about a number of different ways to improve the way TAA strategies navigate choppy markets, this post will talk about adding intra day models. 

The main reason that TAA works so well is that markets are not random.  For the random walk theory to be valid the day to day movements in the stock market would have to be like a coin flip, where what the market does today has no bearing on what it will do tomorrow.  Markets don't work that way.  They move back and forth between trend following environments, where an increase today is likely to be followed by an increase tomorrow, or mean reverting environments, where an increase today is likely to be followed by an increase tomorrow.  However, in a choppy market the day to day movements do start to become more random in nature and each day becomes less motivated by what happened previously.  At the same time choppy markets tend to be more volatile and the spread between the open price of the market and the close tends to widen out. 

Just like markets tend to trend over longer term time frames, they also tend to trend during the day.  Ghao, Han, Li, and Zhou wrote an interesting paper in 2015 showing that the first half an hour return in the market tend to predict that last half an hour return.

This means that a momentum model can be constructed for intra day trading just like a momentum model can be constructed for longer time periods.  This offers a number of important advantages:

1. In a choppy market there is no real momentum over longer time periods but this should not impact momentum intra day.

2. In a choppy market the risk of holding positions overnight increases, intra day models close out all positions at the end of the day and start fresh the next day.

3. In a choppy market most TAA models would be out of stocks entirely and would not be able to participate in gains on up days.  Intra day models can participate in up days.

Intra day models can substantially improve TAA strategies by allowing them to do better in choppy markets and participate in intra day gains when  they normally would not be positioned in stocks.

Thursday, February 4, 2016

Why We Use Counter Trend Models

Counter trend models seek to buy into market weakness and sell into market strength.  They expect that markets overreact and will be mean reverting in nature.  When the market rallies they expect that the market will go to far and that some of that rally will eventually be retraced.  When markets decline they expect that market will overreact and some of the decline will eventually be retraced.   Because these models go counter to the trend I sometimes get questions from clients who wonder why we are buying when the market is going down and why we are selling when the market is going up. 

Counter trend models shouldn't be looked at in a vacuum, they are part of a diversified tactical portfolio.  Imagine a world where traditional asset allocation actually worked like it was supposed to and you had a diversified portfolio of all different kinds of stocks and bonds.  Lets assume you also had an allocation to an alternative asset class like managed futures.  In 2008 your stocks would have declined, your bonds probably would have held steady, and your managed futures would have gone up a lot.  You may not have understood why managed futures went up, typically these strategies are a black box, but you wouldn't have questioned it.  Then in 2009 your stocks would have gone up and your managed futures might have gone down.  Now you would start to question why you have an allocation to managed futures, conveniently forgetting that they saved your butt in 2008.  Of course this isn't the right way to think, you can't take one asset class out of a diversified portfolio and look at it under a microscope, you need to look at it as part of an overall portfolio and as a diversifier. 

So why do people sometimes have trouble wrapping their head around a counter trend methodology?  I believe that in our minds we are all trend followers, we believe the most recent past will equal the future.  So if the market is up big one day we think it will continue to go up, and if it is down big one day we think it will continue to go down.  We then have trouble understanding why you would sell into a rally and buy into a decline.  However, the data supports the fact that markets really are mean reverting.  CSS Analytics did an interesting study on the frequency of daily runs by decade:

They basically looked at each decade from 1950 to 2009 to see how long market runs were.  What they found was that in the 50's one day market runs, meaning the market went up or down and then reversed the next day, happened only 40% of the time.  Meaning that if the market went up or down today it had a 60% chance that the next day would be in the same direction.  Fast forward to the 2000's and one day market runs happened 53% of the time, meaning if the market went up or down today you have a 53% chance the next day will reverse. 

I found similar study on the woodshedder blog.  They tested a daily follow through strategy vs. a daily mean reversion strategy.  The daily follow through strategy would buy when the market was up and sell short when the market was down.  The daily mean reversion strategy would do the opposite.  what they found was the daily follow through strategy was only profitable 34% of the time and would have lost money.  The daily mean reversion strategy was profitable 64% of the time and would have made money.  The results are below:

So the disconnect is simple, in our minds we are trend followers and in reality markets are mean reverting.  When counter trend methodologies are doing well there is no problem but when we hit a period where counter trend methodologies struggle then we get the disconnect. 

In a tactical portfolio counter trend methodologies are a powerful diversifier to trend following methodologies.  They tend to do best in market periods when trend following methodologies do their worst and they can profit in bear markets without having to go short.  Where they struggle is in market corrections where weakness continues for a long period without enough strength to sell into.  There are a number of things you can do to minimize the impact of this in the overall portfolio..