Skip to main content

Reply to "Poll - Trout, Cabrera or other for AL MVP"

quote:
bball123- Interesting thoughts. Can't say I disagree. Although I would venture to assume that you agree with the fact that retrospective modeling can lead to future predictions, at least to the best of our capabilities at this time. If Nate Silver can predict the election, then baseball should be able to predict future results fairly accurately. The question then becomes...is it easier to model a prediction based on Silver's political calculations? Does baseball present a tougher task due to more/different variables? Or, is baseball being looked at incorrectly altogether?

I'm not as bothered by Cabrera's victory as I am by Trout's one third place vote, or the inclusion of Raul Ibanez receiving a 10th place vote (What????). Those that vote need to view this award seriously. 28 people is too small a sample size as it is, baseball doesn't need individuals making decisions with no validity behind them.


In my opinion, baseball is the hardest to predict (the outcome of a game) among the three major sports of football, basketball and baseball. For example, if I have the Colts playing the Raiders now, we have high confidence knowing what the outcome is going to be. If you go to Las Vegas, they will have the odds where you can bet your real money. When friends ask me who is going to win in a baseball game, I would say, it's baseball - anything can happen, it depends who will show up in a game. For example, Posey, it batted poorly overall in the playoff, 1 for over 20 at bats (need to verify), until he hit the most critical home-run in the most critical game. His batting average is not 1 for 20, for sure but... The variation and deviation is so great from the average, how could we predict? And this is only for 1 player. You have 9 guys on the field, and 16 guys on the bench. What is the permutation and combination of all the factors. Then you have the same for the opponent, permutation and combination, and you have to match them up 1 on 1 (batter vs pitcher and fielders) for all different different situations. We need an enormous sample size in order to fit a model into this model with huge among of degree of freedom. I believe it is still possible to model but to a lesser degree of accuracy. After all, we could model hurricane, we could model earthquakes to some degree of accuracy. For the political election, you have only two, Obama and Romney. You have millions of people you can take sample from to improve the model's accuracy. Even then, the statisticians on the Romney's camp messed up big time; their computer software, ORCA, did not work at election day. They spent some much money on it and it did not work on election day, go figure.
×
×
×
×