Runs Scored + Runs Allowed Model Question

[Professor1215]

Good afternoon. I would really like some help on this.

I am currently working on modeling runs scored and runs prevented in the MLB. As it stands now, I have a somewhat functional model.

My question, is how to properly combine RS/RA to get a total.

For example, lets say the Yankees are playing the Red Sox. Below are my hypothetical model results.

NYY Runs Scored = 4
NYY Runs Allowed = 3

BOS Runs Scored = 5
BOS Runs Allowed = 4

To get a game total, do I simply average these numbers (NYY RS + BOS RA) & (BOS RS + NYY RA)?

Thanks for the help!

[Miz]

without making any comment on how the model produced these numbers, I think your last step is reasonable

[Professor1215]

without making any comment on how the model produced these numbers, I think your last step is reasonable

Thank you for that. Those numbers were simply made up. The model I have gives outputs of 3.1-5.8 RPG for each team.

I am a long way from any real results, but I want to feel good from a theoretical standpoint before I progress any further.

So in this example, the game total would be 8, with a score of 4-4, or +100/-100, 50.0% no-vig ML.

[Stackzilla]

Here's a system I was using with soccer it worked hitting at a high % but I had to stop because the juice varies per game so it ultimately was a failure but for baseball when the lines juice doesnt move you may see good outcomes.

mean of all the home team's runs at home + league avg /2 = A
mean of all home team's runs allows + league avg / 2 = B
mean of all visitor team's runs away + league avg/ 2 = C
mean of all visitor team's runs allows away + league avg /2 = D

A+B+C+D / 2 = possible runs scored

or instead use all season numbers instead of only home/away numbers
i hope this helps , Pm me if you decide to try it out I'd like to see if it works, thanks BOL

[Waterstpub87]

Here's a system I was using with soccer it worked hitting at a high % but I had to stop because the juice varies per game so it ultimately was a failure but for baseball when the lines juice doesnt move you may see good outcomes.

mean of all the home team's runs at home + league avg /2 = A
mean of all home team's runs allows + league avg / 2 = B
mean of all visitor team's runs away + league avg/ 2 = C
mean of all visitor team's runs allows away + league avg /2 = D

A+B+C+D / 2 = possible runs scored

or instead use all season numbers instead of only home/away numbers
i hope this helps , Pm me if you decide to try it out I'd like to see if it works, thanks BOL

I'll save you the trouble, this isn't going to work.

[Stackzilla]

yea i didnt think it would thanks there really isnt a secret system too many variables to consider best thing you can do is your research to come with the decision

[statnerds]

this certainly will work because the starting pitchers are all equal

why do all of this work when the market does it for you?

[evo34]

Thank you for that. Those numbers were simply made up. The model I have gives outputs of 3.1-5.8 RPG for each team.

I am a long way from any real results, but I want to feel good from a theoretical standpoint before I progress any further.

So in this example, the game total would be 8, with a score of 4-4, or +100/-100, 50.0% no-vig ML.

A couple things:

The median will always be less than the mean in baseball totals, esp. for low totals. So if you project 8 runs/game on average, the fair line will not be 8 +100 -- not even close.

As you prob. know, half runs are worth very different amounts depending on the number.

For totals with a heavy favorite, calculate the odds of the bottom of the ninth being played.

This list is far from exhaustive, but some food for thought.