Message Font: Serif | Sans-Serif

No. of Recommendations: 35
Nice demagoguery, snoop. No really, I mean it, well done.

My griping about an unfair quarter was my favorite part. A stroke of genius.

My turn for a made up conversation:

NFL Ref: Chains!

Commentator: Looks like the Chiefs got a favorable spot on the ball, but this is going to be close.

Dennis Miller: Sort of like Jean Paul Marat's decision to take a cold bath.

Commentator: Shut up, Dennis. Here comes the measurement. They've made it just by the nose of the football!!

Ref: First Down, Chiefs!

Datasnooper: Wait just a minute, that measurement wasn't valid.

Ref: Who the hell are you?

Commentator: Who the hell is that?

Datasnooper: If my calculations are correct (and my calculations are always correct) The distance by which the ball made it past the first down marker was only 1.3 inches, which falls in the 33rd percentile of the distribution of outcomes for measurements that are, in fact, set by human hand and thus controvertible.

Ref: Huh? It's a first down. The Ball passed the marker! Get off the field, math boy!

Datasnooper: First you must tell me that the original 10 yards were accurately measured and not eyeballed. Since you did not line up the markers with the ball using GPS and vectoring, then we can assume that there is no way to really be able to tell within a 2.7 inch margin (I'm right about this) whether or not the chains were placed in true relation to the ball. Moreover, you have run through a three play sample, each time the ball was moved and replaced. Never once was the ball placed in the exact same position, so each play added an error rate that you can neither calculate nor adjust for. So for each play run you are using but a sample, not a population that completely accurately shows the exact level of performance by the team in its quest to advance the ball. Therefore after three plays any distance within 1.7 yards of the first down marker must be ruled inconclusive.

Datasnooper: Shut up, Dennis.

Ref: So it's impossible to tell whether or not the team actually got a first down.

Datasnooper: No, not impossible. If a player scores a touchdown, you can assume that he has also made a first down. Of course, you would have to know that the original placement of the ball was fraught with errors and 99% of the player's ability to advance the ball into the endzone was based on the luck of him happening to run where the defense wasn't. Your entire system of measurement does not hold enough precision, and therefore the entire game is invalid.

Ref: OK, well I have an idea. If a team gets within 1.7 yards of a first down, we can have a coin flip. I have this really cool coin we can use.

Datasnooper: yeah, funny thing about coin flips....

Chiefs Coach: Hey, ref! What the hell? We playing ball here or what?

Ref: Seeing as the attempt to measure progress is futile, I guess the entire foundation upon which football is built is invalid. Nah, Coach. Just go home. I'll call the NFL and get them to send you your last check.

----------------------

What you are saying is that without the full population, any amount of information you are using is sampling, and thus only tells you about the sample. Well we don't have data going back to the 1400's, nor do we have data from the future, so we cannot have a population.

Because even if we HAD every bit of the data from the past, every statement for every shmoe who has ever owned a mutual fund or an index fund we wouldn't have future data and thus we could make no determination of historic superiority?

Wow, what a racket.

Let's see where else we can apply this logic:

Datasnooper: Hello, Merrill Lynch? Yeah, I've had an account with you for 35 years and my overall performance has only trailed the S&P 500 by .7% per year. That falls below the threshhold for meaningfulness, so we have to assume that X=0. I need you to send me a check for \$127,324.73.

Datasnooper: Yeah, Olympic committee? you're going to need to make the gold medal platform in speedskating a little wider. You see, the gold medalist only won by .014 seconds, which is statisically meaningless in this sample. You actually have 11 statistical winners.

Datasnooper: Hey, Osama? This is datasnooper. I wanted to talk to you about that 73 virgins in heaven for martyrs thing. Yeah, you're wrong to assume that each martyr receives 73, in fact within this sample it would be valid to say that a martyr would receive negative 73 three virgins. What's a negative virgin you ask? I don't know. I'm a statistician, that's real world stuff.

---------------

None of this comes close to addressing one of the suppositions of indexing over active management, that being tax efficiency. Particularly for the high turnover funds, we may be talking about inches where the true outperformance net of all fees, expenses and taxes to the end user changes the equation quite drastically.