Dear Editor,

I am not questioning any study indicating that, as a group, investors are poor market timers.  This may very well be the case.  Certainly there are some good ones out there and there are undoubtedly some bad ones.  My issue is how we should measure the success of market timing, as we measure the success of any investment strategy.  We regularly analyze investments based on both risk and reward, not just on reward.  So why when we study the effectiveness of market timing are we only concerned with the reward part?  (A general bias against timing, I would surmise.)

If I told you that there is a timing strategy that over the long run has under-performed the market by 1.5% per year but it had a beta of .3, and a drawdown one quarter as much as the market, would you consider that a successful strategy?  I know a lot of people who would.  But you would never consider it a success by only looking at the reward side of this strategy.

Thanks for listening.  I have been receiving your e-mail newsletters for a while and I find them interesting and thought provoking.

Keep up the good work,

Dan
Daniel J. Traub
Braver Wealth Management, Inc.
Newton, MA 

Avi Wohl, a professor of finance at Tel Aviv University and author of one of the studies cited in the article, responds:

Dear Mr. Traub,

If the risk does not change over time then timing increases slightly the risk. Look at a simple (and exaggerated) example: assume the risk free rate is zero and stock market return may be -10% or 30% with equal probabilities. There are two periods and there is no correlation between market returns. Investor A and B have initial investments of $100 .Investor A invests 50%-50% at the risk free rate and the stock market every period. Investor B invests 100% at the risk-free rate in one period (say the first one) and 100% in the stock market in another period. Investor B’s terminal wealth is $90 or $130 with equal probabilities. Investor A's possible returns in each period are -5% and 15% (because half of his investment is riskless). Therefore the possible terminal values are:

($100*0.95)² = $90.25            probability=0.25 (losing in both periods)
$100*0.95*1.15=$103.7875     probability=0.5 (wining in one period)
($100*1.15)²=$132.25            probability =0.25 (winning in both periods)

The expected end of period value for investor B is $110 and the standard deviation is $20. The expected end of period value for investor A is $110.25 and the standard deviation is $14.88.  So, if the expected stock return is the same in every period, then an attempt for timing increases the risk (over the entire period) and decreases the expected return. Practically speaking, these effects are small.

It can be, however, that the periods will differ in their risks and expected returns and someone who invests more in stocks in safer times and less in riskier times may look like a poor market timer. This is the reason that our paper (Aggregate Mutual Fund Flows and Subsequent Market Returns) does not call the negative relation between mutual fund flows and subsequent market returns "bad timing." "Bad timing" is a possibility but there are other possibilities (e.g., periods differ in their risks and expected returns, temporary price pressures following mutual fund flows) as well. We didn't deal with this in our paper. This is indeed a very interesting question and we may investigate it in the future. Thank you for your comment!

Best regards,

Avi

Display article as PDF for printing.

Would you like to send this article to a friend?