Tuesday, March 8, 2016

An Example of Forward Looking Due Diligence in System Development

When most investors do due diligence they look at past returns and expect them to persist into the future.  Unfortunately, they rarely do.  Forward looking due diligence takes past returns with a grain of salt, the goal is to determine where the returns came from and how likely they are to persist.  Forward looking due diligence can be done on money managers, investment strategies, and tactical methodologies.

We recently developed a model for the SPDR S&P 500 ETF (SPY).  The model can either be long SPY, short SPY, or in cash.  From 3/7/2011 to 3/7/2016 it makes $142.6 on one share of SPY, not including any dividends.  For comparison to buy and hold SPY opened at $132.86 on March 7, 2011 and closed at $200.59 on March 7, 2016, so buy and hold would have only made $67.73.  So far so good, our strategy  more than doubles buy and hold over this period.  If we look at the annual returns of our strategy it would have also made money ever year with a low of $3.75 in 2014 and a high of $55.28 in 2015.  Still so far so good.  However, over the past 12 months the strategy made $81.14 and it made $28.24 in February, $14.18 in December, and $14.66 in September.  Those are the best, third best, and fourth best months for the strategy over 5 years, and they all happened within the last 12 months.  On the one hand this tells me that the strategy will do its best in a troubled market.  One the other hand I am worried about whether returns like this can persist.

Backward looking due diligence would place a lot of money into this strategy expecting these recent returns to persist.  In fact our optimization program wants to allocate 50% or more to this model.  Forward looking due diligence is much more cautious.  Yes the strategy has done exceptionally lately, and yes we are likely to stay in a market environment that is conducive to this strategy.  However, the strategy has done so well that the returns actually border on ridiculous, they could persist but they could also revert back to the mean.  Looking at the monthly returns the strategy is up 6 months in a row (March is currently down).  It has only done that once before.  This could be the best model we have ever developed but we have decided to limit exposure to 12.5% until we have a bad month or any type of drawdown. 

No comments:

Post a Comment