San Leandro, CA

Sunday, November 15, 2009

Here is another chapter of an upcoming Ebook on investing. It should put to bed a lot of the numbers that the industry has used to persuade buy and holds, backtesting, data mining and more to the naive and unsuspecting consumer. 

How statistics can lie



Most everyone is familiar with the term, 'liars can figure and figures can lie'. Here is the perfect example of absolutely true numbers that can be misused to provide whatever support one desires.

Assume we are looking at the number of cell phone sold from 1959 to 2009. Let's first take 1959 to 1989. I will take a little latitude and say there were none. So if we look at the average number of cell phones sold during those 30 years we get ZERO. A true statistic but with effectively no value.

Now let's take 1989 to 1999- we will use a number of 5,000,000 phones in total or an average of 500,000 per year. Maybe an O.K. statistic but let's also take the full period of 40 years from 1959. That's 125,000 per year. An absolutely true number, but what is the point? The use of the first 30 years of no sales is a useless statistic- but it can be used to show a much lower yearly average in case some "adviser" wants to distort reality.

Let's look at 1999 to 2009. I will use another 15,000,000 phones (far too few but remember this is just an example). That's 1,500,000 per year. Looks great- but it actually is a great distortion of reality. Certainly the greatest number of phones were sold between 2004 and 2009- even between 2007 and 2009. An average does not reflect the huge increases that reflect the later years. But unless you thought about this or were aware of the sales, the average is one that a lot of people might accept. And use to support any future projections.

However, if I really wanted to distort reality, I could simply take 20,000,000 phones in total and divide by 50 years for an average of 400,000 per year. Absolutely a mathematically correct number that has little to do with reality.

I submit that when one is given statistics on what has happened for past results in the market, it is necessary to tear apart what is truly happening with the computations before you commit your funds to an unsupportable position.

That said, this is done all the time since very few people question the numbers from a computer program. I do not necessarily blame then since it takes some time, effort and experience to know what should have been projected. But to have the major B/D firms do it consistently is a violation of trust and fiduciary duty. (See portion of recent report herein)

As one stats professor noted when giving an exam- "you can use any program you want as long as your wrote it yourself. Otherwise you do not know dirt about the end results." 

 

Errold F. Moody, Jr.

PhD MSFP MBA LLB BSCE

Life and Disability Insurance Analyst

San Leandro, CA

510-352-4127