Friday, August 7, 2009

Misleading perception of accuracy

In his book, When Genius Failed, author Roger Lowenstein said that Long Term Capital Management reported the Value at Risk on their portfolio with such precision that their investors felt confident in their manager's control of their assets. LTCM's client letters went further than most managers because they did not merely concede the possibility of a loss, they calculated the supposed odds of it occurring, and to precise mathematical detail. But as David Dreman pointed out, "there's a false precision to analysts' forecasts that enable over confidence."

The Wall Street Journal ran a story earlier this week regarding the precision of fuel consumption accuracy (http://online.wsj.com/article/SB124940692698405243.html#mod=todays_us_opinion) whose title, "When Precision is Only 92.11567% Accurate" speaks to me regarding our industry's sometimes naive attempts at improved accuracy, given the degree of noise that we face in our system. The author quoted Peter de Nayer, an independent U.K. auto tester who said "It is ludicrous to suggest that you can get fuel-consumption accuracy anywhere past the first decimal place, let alone the second." Might the same thing be said about some of the information reported to clients and prospects? We know some managers who report returns only to one decimal place; perhaps this isn't such a bad idea. On the other hand, I've seen reports with returns to THREE decimal places: precise, but are they accurate?

When I was Mayor of the Township of North Brunswick (NJ), my finance officer would give me reports down to the penny; I told him to just stop at the thousands: not that I questioned his accuracy, but because I simply didn't need to know that level of detail. I also wouldn't have cared if he later found an error that was so immaterial that it wouldn't have mattered to me and yet would have required him to provide me with a revised report. And while I'm aware that many individuals DO like to see such precision, I can't be bothered. At least in some things.

In our training classes I sometimes wax poetically about this topic, and mention the number of people there are in the U.S., but immediately correct myself because surely someone was just born or died. The article, too, uses the census metaphor to make their point. And their point is mine: "Still, decimal places lend the aura of authority and the veneer of verisimilitude. So the modern world is awash in squishy numbers wearing the many-figured garb of faux precision." When we show numbers to increasingly more decimal places people actually believe them ... unfortunately, they're often incorrect. So why bother? Is that detail REALLY necessary? Especially when in the end, the noise surrounding them may cause errors to exist that we're simply unaware of?

This topic, precision vs. accuracy, is one that we've taken up at our annual PMAR conference and often engenders debate. The Bible's book, Ecclesiastes, perhaps, should have included the line "a time to be precise, and a time to round," or something to that effect. If we're attempting to put someone on the moon, we better be accurate to several decimal places or they may miss their landing point by quite a distance. But there are other times when we can back off a bit.

The reality is there are differing views on the subject. This article may cause you to think, however, about the similarities in our industry with what we encounter when attempting to report returns and other measures to clients.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.