[Lanny_McDonald]

Originally Posted by Doc Hudson

This is a weird statement, how is baseball binary?

He scores/not scores is binary, you can make many data types fit into a "binary" form. But, baseball also has more possible outcomes other that hit or miss.

Statistics in baseball are extremely binary in nature. They are one-to-one events. Pitcher throws the ball - was it a ball or a strike? Pitcher vs hitter - was the hitter out or did the batter reach base? Ball was hit to the third baseman - did he record the out or not? The rules are black and white and very rigid which allows for very accurate data collection. Hockey is not like that at all, because there is potential for multiple bodies to act as an influence on an event that does not exist in baseball. This makes data collection much more difficult and less pure as more uncontrolled variables influence possible outcomes.

Originally Posted by Jay Random

‘Binary’ is the wrong word. Baseball is a game of discrete events. In hockey the action is continuous, and every player on the ice has the potential to affect it at any given moment – which means it cannot be perfectly described by a general mathematical solution. The most you'll be able to do with statistics is find trends and correlations – but you will, New Era to the contrary, be able to do that much.

Very loose trends and no correlations. Go take an advanced stats class and then try and tell me these “advanced stats” meet rigor. Much of what is discussed requires multivariate data collection and analysis, and the data does not meet rigor to do so. The vast majority of these stats are very basic analyses which are not capable of modeling complex environments like those being discussed. I learned a long time ago what simple stats can and cannot model. Hockey has too much chaos in the mix to properly model with such simplistic terms.

It's rather like the n-body problem in gravitation. You can never perfectly analyse the orbital motions of all the bodies in our solar system, because they all affect one another in complex ways. But you can make predictions to a very high degree of accuracy. Hockey, of course, is less predictable because hockey players (as a rule) are capable of making their own choices and planets aren't. But those choices are highly constrained by the situation, and you can measure a player's value by the use he makes of the choices available to him.

This thing is that those bodies behave in the same way, all the time. The rules of physics make predictive modeling possible. The rules of hockey, and the number of variable that can take place influencing any one event, make it extremely difficult to model. For instance, a shot from the boards bounces off multiple bodies on the way to the net and finds its way into the goal. Does that count the same as a clear shot? Should the goaltender be penalized for that? I don’t see an infielder getting charged with an error when the ball goes off an obstruction. Again, the complexities of the game make it very difficult to model using very simple statistics. As you lauded to, the fact that players have the ability to make their own decisions, and operate beyond the rules, makes the game that much more difficult to model. The black and white nature of events becomes much more gray and cloudy because of the chaos on the ice.