The following is in response to the commentary, The 5-Year Itch, by Robert Isbitts of Sungarden Research, which was posted on March 3:
Dear Editor,
This commentary looks like data mining:
- I can buy the first chart, showing the 165% return since the low of 2009 and the days from the low to the high (1,813). They are using event timing, i.e. dates of low to the high. But what is the significance of the date 9/1/95 in the second chart? It was not the end of a recession, nor did it mark a low in the market. It surely did not mark the start of the internet. So there is no event-based apples-to-apples comparison, is there?
- The dates are also dubious. The high was in March 2000 not August 2000.
- The returns really cannot be labeled exactly the same.
Consider these facts:
On 9/30/95 the S&P was 584.41.
On 3/20/00 the S&P was 1,553.11 (the high came in March not August)
Not only did Isbitts fail to make an apples-to-apples event-based comparison, but the days are off as well. Also, the returns are close but they are not exact. (1995-2000: 1553/584=2.66) and (2009-2014:1875/666=2.81).
We know that "history doesn't repeat but often rhymes," but this looks more like "historical revisionism" to me.
Best regards,
John DeSantis
Robert Isbitts responds:
In regard to Mr. DeSantis’ feedback, let me respond to his first three points:
- Granted, the dates on the charts were slightly shifted to the right and could have been in a better format. Next time we will tilt them so we can fit more in the chart.
- August marked the point at which the upward trend stopped and bubble began to burst.
- No one is making the case the returns are exactly the same, only that the numbers are similar over a similar time period.
In addition, Mr. DeSantis stated that we used the S&P price of584.41on 9/30/95. But we clearly stated our data began on Sept. 1st, a point where the S&P was at 561.14, according to the St. Louis Federal Reserve. We used August 2000 data, because it marked the point at which the upward trend stopped and bubble began to burst. Using the word “peak” may not have been the best word.
Adjusting for the fact that we used data of the first as of the month shows that the returns matched up:
09/01/1995 – 08/18/2000: 1488/561.14 = 2.65 03/06/2009 – 2/21/2014: 1836.39/695.19 = 2.64
