I'm not saying that scoring margin cannot help improve a model. For an individual's model to rank teams, of course use it if you want.coda wrote: ↑Wed May 08, 2024 9:02 am
I am not sure that that is supposed to even mean, just nonsensical rambling. It is pretty simple. You know only have the following information and you have to bet on the game between Team A and Team C..
Team A beat Team B by 9 goals
Team C beat beat Team B by 1 goal.
base on that limited information I can not imagine a person saying "put my money on Team C". Perhaps that would be you, if it is I would love to bet with you.
But we should separate out the debate into two questions:
1) Is it a useful data point for ranking system? (Sure, I don't think anyone is really saying otherwise)
2) Should it be a part of the criteria for the Selection process in some way?
The important question to me is #2. If someone wants to build a model to help them bet on games, that's a question #1 issue. Question #2 is what my concerns about scoring margin are related to. I don't think anyone is really arguing here about Question #1.
But if we are talking about using is in the selection process, that's where it becomes a question of weighing the lift that you'd get from including it and the negative consequences of including it.
In my opinion, the negative consequences outweigh any potential improvement to the ranking process.