Do the MLB Standings Represent "True" Performance for Teams

[Tanko]

I know many of you hate stats but so what. I love them.

One that I always find interesting is the "actual" W/L record for teams Vs their "expected" W/L record. The "expected" W/L record can be determined using Bill James' equations for determining wins and losses from runs-scored and runs-allowed. Here is what the MLB standings are today vs what Bill James' "expected" W/L record projects. This does not account for injuries or schedule strenghth.


The "so what" of this data....

1. The Braves are as good as they represent (no surprise).
2. The Rangers are even better than their current record indicates.
3. The Brewers and the Marlins seem to be "pretenders" living on luck.
4. The Cubs and Padres are much better than their current records indicate. I could see Padres making a run for the playoffs but for the Cubs, even though they are much better than their record shows, they are still not likely to make the playoffs.
5. Baltimore is also not as good as they indicate. That doesn't mean they are bad, it just means they aren't likely to stay at the 2nd best record in MLB.

 

[Tanko]

Here is the delta between current and expected record for teams.

The teams at the top are underperforming expectations relative to Bill James' projections and are actually a BETTER team than their current record shows.

The teams at the bottom are overperformance expectations and might be considered "pretenders" and their record for the 2H of season may fade somewhat.

 

[Tanko]

For Padre fans - holdout hope. They should get better.
Brewers and Marlins fans need to hope the their luck continues to hold.

Again, injuries can impact the data. This is just general observations about their records.

 

[stormtrooper8]

This is a more compelling stat

 

[stormtrooper8]

Sometimes you get teams that perform well and have high expected W/L but doesn’t translate


I look at if these teams really struggle situationally, late in gaMes

 

[Tanko]

Sometimes you get teams that perform well and have high expected W/L but doesn’t translate


I look at if these teams really struggle situationally, late in gaMes

Not sure but I think the Marlins are in that group.

I've not confirmed it but they have more 1 run games than any other MLB team. If true, that would explain their W/L being over-inflated relative to their "expected" W/L record (which is based on runs-score vs runs-allowed).

 

[Pimike]

Good stuff Tanko

 

[KVB]

Oh wow Tanko. Treemendous thread.

This is actually a video topic of a presentation I already made, including different ways to use those sabermetrics. I filmed it last summer, didn't post it, then tossed it. I went into he pythagorean theorem.

I was just telling my wife last night that I thought I'd make the video again, then add some newer findings to better the topic.

Always remember, regardless of the exponent we're able to derive from the numbers to use the pythagorean theorem, no matter how good of a curve fit with even the most recent data one thing seems to never change.

The expected win loss is closer to reality for those teams in the middle of the pack. Teams that are far below and far above become a little fuzzier to predict. There are mathematical reasons for it but the fact that it's true was a segment in the video I made.

Can't just take the lowest performer and start betting him or the best performer and fade him. You have to weigh it, and include in a formula with other factors, and of course seek value in the market. That's a whole segment too.

That said, you can take this win loss expectation and not just apply it to any stretch of games or a season, but you can also apply it to an individual game. If you are creating a prediction for a game using a runs scored method then you can convert that to an expected win percentage. That is then converted to a moneyling which you then compare to the market.

I noticed BMR had a score prediction. Without such a conversion, how would one know if that prediction has value in the moneyline markets?

Great thread Tanko.

 

[Tanko]

Oh wow Tanko. Treemendous thread.

This is actually a video topic of a presentation I already made, including different ways to use those sabermetrics. I filmed it last summer, didn't post it, then tossed it. I went into he pythagorean theorem.

I was just telling my wife last night that I thought I'd make the video again, then add some newer findings to better the topic.

Always remember, regardless of the exponent we're able to derive from the numbers to use the pythagorean theorem, no matter how good of a curve fit with even the most recent data one thing seems to never change.

The expected win loss is closer to reality for those teams in the middle of the pack. Teams that are far below and far above become a little fuzzier to predict. There are mathematical reasons for it but the fact that it's true was a segment in the video I made.

Can't just take the lowest performer and start betting him or the best performer and fade him. You have to weigh it, and include in a formula with other factors, and of course seek value in the market. That's a whole segment too.

That said, you can take this win loss expectation and not just apply it to any stretch of games or a season, but you can also apply it to an individual game. If you are creating a prediction for a game using a runs scored method then you can convert that to an expected win percentage. That is then converted to a moneyling which you then compare to the market.

I noticed BMR had a score prediction. Without such a conversion, how would one know if that prediction has value in the moneyline markets?

Great thread Tanko.

Thanks KVB. I wanted to keep the information simple to avoid losing the audience. I could go into a few pages of details on this stuff (not at your level) but I'm not sure too many would be interested.

Any additional videos you produce would be readily appreciated.
Very glad you're back.