smackeddog wrote:H1AND1 wrote:The SCHOENE model has them winning 32 games. Less than last year as well.
I was impressed that the computer last year foresaw Tyson's injury, and the quitting on the coach- even guessed that we'd have a rally at the end of the season! Remind me what else Schooner predicted regarding other teams not called The Knicks? Oh thats right, it was pretty much wrong. (not having a go at you, just the computer!).
Like in life, physics and basketball, you can't create an all knowing, future predicting computer because there are too many variables, and the greater number of variables you include, the greater the degree of chaos created due to the creation of more variables to take into account.
Yes, you are right "computers" cannot predict the future with 100% certainty. But what modern statistical models CAN do these days is give you an idea or a baseline as to what is reasonably likely to expect. There will always be outliers and instances where the computer models err but that is to be expected because as you said there are a lot of variables and nothing is ever _certain_.
All that being said, just because a stat model or computer model cant predict 100% of an outcome it is pretty foolish to simply dismiss any and all statistical models out of hand. Looking at the various projections, and especially when you look at multiple ones and they are each saying the same thing ie: 30 something wins, it stands to reason that the MOST LIKELY outcome this season is a win total around that amount. Could the Knicks win 40, 50 games? Sure. But what these models are telling us is that that is the least likely outcome along the range of win totals.
Again, just cause the projections arent foolproof doesnt mean they have no value. Its a matter of thinking about it correctly. And i will lend more credence to a well thought out computer or statistical model, which by the way is made by humans-the computer simply does the number crunching-than a bunch of fans or so called "expert" sportswriters' projections which are based on their feelings and what they think they see any day of the week (and not that that has zero value, but it IMO it has a helluva lot less than a well thought out statistical model).