Friday, March 14, 2014

Model versus man, who is better?

To quote Paul Meehl, an eminent scholar in the field, “There is no controversy in social science that shows such a large body of qualitatively diverse studies coming out so uniformly in the same direction as this one [models outperform experts].”

  1. Superiority for mechanical-prediction techniques was consistent, regardless of the judgment task, type of judges, judges’ amounts of experience, or the types of data being combined.
    --Grove et al. 
I was having one of those endless but always enlightening discussions on what is a better approach to investing, a pure model or discretion. Coming from both environments,  I have to say there is value in both based on the type of decision that is to be made and the environment faced. The important skill is knowing when to employ the right decision process at the right time. If you have a lot of data, the model should always do better. If you have a limited data set or limited events, the advantage may be to follow some other decision process. Yet, the a discretionary process may average opinions or use multiple models weighted in some combination. 

In all cases, there has to be a "model" or process in place for how the decision should be made. Process is everything. 

Is all information useful?


“Decision makers often pursue noninstrumental information--information that appears relevant but, if simply available, would have no impact on choice. Once they pursue such information, people then use it to make their decision. Consequently, the pursuit of information that would have had no impact on choice leads people to make choices they would not otherwise have made.” 


C. Bastardi A, and E Shafir, 1998, “On the Pursuit and Misuse of Useless Information,” Journal of Perspectives on Social Psychology 75, p. 19-32. 


We are bombarded with information in the investment world all day long. Many follow the screens to watch new stories unfold or watch prices to glean more information. Is it worth it? 

Research suggests that is not a good use of time. That is not the same as saying all information is useless. The good investors is able to separate the good from the bad new and information. Critics will say that what is important information will change over time, so you have to look at all. I don't think so. Important stories will arise, but it is likely that these stories will be infrequent and you may be too late to capitalize on it. 

Simple rule, stick to the important information and try to block out the rest. What is important are the few risk factors that have shown through time to have a statistical impact. Being a systematic investors can beat information overload. Simple rules like trend-following will work. 

Monday, March 3, 2014

The value of commodities in a crisis

The Ukraine as a geopolitical flash point increased significantly over the week-end. The result has been a large equity sell-off today but a corresponding increase in commodity prices. This difference in performance is the value of holding a diversified set of asset classes.  The demand for commodities has increased as a precautionary move in a crisis while the expected discounted cash flows from equities has declined as a pullback in revenues is likely. Oil is up over 2 percent, wheat is up 5%, gold has increased by 2%. The Eurostoxx index has fallen 3 percent. 

There is a strong negative correlation between financial and real assets in a crisis. 

Dynamic Trading

How do you adjust a portfolio through time? This is the key question facing most money managers. The allocation decision of static portfolio may have gotten easier with ETF's which can allow easy access to a benchmark, but there is still an important issue of how asset allocation should adjust through time especially if there are costs associated with any adjustment.

Garleanu and Pedersen have looked at this issue in "Dynamic Trading with Predictable Returns and Transaction Costs". They conclude that the optimal trading or dynamic adjustment policy for a portfolio is to: 1. aim in front of the target and 2. trade partially toward the current aim. Simply put, the allocation the optimal trading strategy will take a weighted average of the current portfolio and the portfolio based on the future expected return. Moving to the target or aim portfolio will be affected by the mean reversion of the expected return. If there is slow mean reversion, there will be reason to focus way ahead on what should be the return allocation. Thus,  if there is a slower mean reversion of the the expected return there will be more weight on the aim portfolio and less on the current portfolio. Hence, there is a trade-off between the current portfolio, the Markowitz portfolio based on the expected return, and the aim portfolio which is the mechanism to get the the best expected return portfolio.  The Markowitz portfolio is a moving target but a portfolio manager has to worry about where the target will be in the future not what is predicted today.

A factor based model may predict that growth should lead to a higher expected return or that there is a macro surprise effect, but the issue for a portfolio manager is that there are costs with moving to any new target portfolio. If the fact generating expected return does not have persistence, moving to the new optimal Markowitz allocation may not be appropriate. The market may already be moving in a new direction if there is a mean-reversion or factor decay. 

This strategy makes perfect sense. The more any new strategy has an expectation of not decaying, there should be more emphasis on this new allocation. If there is a fast decay or mean reversion of the expected return there will be more weight place on the current portfolio. Do not make a quick adjustment. You will be disappointed, as you reach the new allocation the returns will no longer exist.

The authors discuss alphas and alpha decays as different return strengths and mean reversion speeds. If there is a lot of alpha or return strength, aim for the new allocation. If there is a lot of alpha decay focus on the current portfolio. The Markowitz portfolio is always a moving target because expected return are constantly changing; however, with transaction cost, an investor cannot just move to any target immediately. There will always have to be a trade-off between where the portfolio currently is and where it should be based on the cost of adjustment and the type of factor expectation.