SEC confirms Dalmady analysis on Stanford

Within hours of my posting about Dalmady’s analysis of possible fraud at Stanford International Bank, I received a comment on my blog telling me that I was spreading lies and that I should recant:

Try to do some investigative work instead of building upon lies ... When you want something successful to fail you present the perception, associate it with something negative (Madoff) and watch the masses panic ... You have now created the reality ... I hope you put as much energy in recanting this story as you do posting them ...

I did not lose sleep over this comment because by the time I read that comment, the SEC had filed its complaint against Stanford confirming most of what Dalmady had surmised.

I found the SEC complaint short on hard facts. Did I really know anything more on reading this complaint than I did after reading Dalmady? I am not sure.

And, there were some things in the complaint that did not sound right to me like the assertion that it is “impossible” for a large portfolio to produce identical returns of exactly 15.71% in two successive years. If exact means that there was no rounding at all in arriving at 15.71%, then it is in fact almost impossible. But then it is quite improbable that a really large portfolio would produce a return which is exact to two decimal places (with no rounding error) in even one year. The return on a $8 billion portfolio at around 15% would be over a billion dollars and would therefore have twelve significant digits when measured in dollars and cents. Suppose that the return in percent is also computed to twelve significant digits. The probability that only the first four significant digits (1, 5, 7, 1) are non zero and the other eight significant digits are zero would then be about 10^(-8) or about 1 in 100 million. Quite improbable!

But if what they mean is that the return rounded to two places was 15.71%, then that is not impossible at all. If the range of returns is say 5% (500 basis points), then the probability of the return being the same as the previous year’s return to two decimal places (one basis point) is 1/500 or 0.2%. Since the SEC examined at least 10 years of data (their example is of 1995 and 1996 returns), the probability that they would find at least one year in which this happened is 1/50 or 2%. Certainly, 2% is not my idea of impossible or even improbable.

Posted at 4:33 pm IST on Wed, 18 Feb 2009         permanent link

Comments