THE BOOK--Playing The Percentages In Baseball

I haven’t finished reading the article yet, but the graph seems to be misleading because 100 is not average for team OPS+.  100 is average for non-pitchers, but team OPS+ includes pitchers hitting.  I think average for a non-DH league is usually around 93-94.

Kincaid: ah!  That would be an excellent point, for sure.

Two questions:

1) Why use OPS+ and ERA+ instead of R/G+ and RA/G+?

2) How does the same graph look for World Series LOSERS? Are they less dependent on pitching and fielding, or is the graph the same, shifted left/down just a bit?

Charlie: does Sean have that option?

A more minor complaint:

Some of the figures used in the article, such as comparing the median/average OPS+ to ERA+ or counting the number of teams with a better OPS+ than ERA+, are additionally affected by the scale of ERA+.  Even if you correct for the non-100 baseline for OPS+, you would get different results if you used lgERA as the denominator for ERA+ than you get using lgERA as the numerator.  Using lgERA as the denominator in ERA+ spreads out the variance of ERA+ figures across the league, which also drives up the average ERA+ of WS teams since they are overwhelmingly above average.

Charles/#3’s suggestion of using R/G+ and RA/G+ would be more appropriate since they are on the same scale.  If using OPS+/ERA+ is necessary due to limited data available to the author, the comparisons of one number to the other are going to be biased and make it difficult to draw any conclusions about the magnitude of each.

The article mentions that the 2010 Giants were the favorites to win the WS, but it doesn’t really back it up.  Paap uses a simple weighted average of ERA+ and OPS+ to say that the Giants were a perfectly average WS winner. 

On a whim, I looked up the ERA+ and OPS+ of the other playoff teams.  None of the others came close to 108.  Most of them hovered around 104 or less.  Regardless of the scale of ERA+ or OPS+, this is a telling metric. Especially since the Giants were practically unanimously thought of as overperforming.

I wonder how well this predicts postseason performance and making the playoffs.  I wonder if it would be improved if you just took a weighted ERA+ and OPS+ of postseason starters.

Thanks all for commenting on where I misstepped. I assumed (never a good idea) that league-average OPS+ was set at 100, like ERA+. I also hadn’t thought of the scale of ERA+, thanks for the tip.

I’ll post an addendum to this on THT Live, hopefully, that will temper my original argument based on this information.

I used OPS+ and ERA+ because I didn’t have RA+ and RS+ available. I was also building on something Bay City Ball had done.

I mentioned in the article that the Giants were NOT the favorites, but maybe should have been given their elite pitching and advantage using a simple weighted average of ERA+ and OPS+. I did this from 2010 to 1997, and it was hit and miss… about 50% of WS winners led in this.

Here’s 2010:
SFG: 108
NYY: 107.5
PHI: 104.5
TB: 104
MIN: 105.5
TX: 106
CIN: 104
ATL: 105.5

In 2010 and the latter half of the interleague era at least, the AL had a huge advantage such that a team with a 105 ERA+ and OPS+ average in the AL would be a huge favorite over a team with a 105 average in the NL.  So, you cannot compare ERA+ and OPS+ across leagues without doing a league adjustment, any more than you can compare them across any other leagues, like AAA and MLB, college and AA, etc.

This is the same old argument we seem to have every year or so.

There is no particular reason why good pitching is any better than good hitting in any game or series, except…

You can leverage good pitching a lot more than good hitting in the post season - obviously. You use your top 2 or 3 starters much more in the post-season and you can use your best relievers until their arms fall off.  And if you have a great #1 and #2 starter, you will tend to have a very good pitching staff overall (high ERA+).  If you have a weak #1 or #2 starter, you will tend to have a weak pitching staff.

For hitting, there is not much more you can do in the post-season than you can do in the regular season.

In addition, there is a slight advantage to preventing runs (as opposed to scoring runs) in the post-season because the run environment is (much) lower (small ball strategies, colder weather, more use of elite relievers, etc.).

There ain’t no magic going on here.  Good pitching does NOT beat good hitting in the regular season, in the playoffs, in the kitchen....

MGL: I really appreciate what you’ve said. This does come up often, but more than that I don’t understand why we take numbers from day one of the season to the end of the World Series and say “so and so is what wins games (or doesn’t)." 
Why not treat them as two separate events? I’m not convinced it evens out over time.  The input of influenes surrounding the aspects of baseball   (pitching, hitting, relief, etc.) that wins in the season vs. Postseason I feel is too different.  Maybe I’m crazy, but it seems almost a two step process.One example wld be over the course of the season at some point a #1 or #2 starter is going to have a bad day, or make mistakes, and the hitters will get to him. A team needs those runs to get to the postseason. If those same starters have a bit of luck combined with what we know wins games in the postseason  mixed in with their excellent pitching in the postseason they are going to beat the hitters everytime.   

I’d love to see these type of exercises broken down as two seperate events.  
 

"If those same starters have a bit of luck combined with what we know wins games in the postseason mixed in with their excellent pitching in the postseason they are going to beat the hitters everytime.”

Wow!

What word? I hope luck wasn’t offensive.

No problem…

Per Anna’s point, though, I’d be curious whether 2nd-half OPS/ERA is a better predictor than full season OPS/ERA.

IIRC 1st-half/2nd-half wins does not matter, but maybe with these narrower metrics, it would.

I know it’s anecdotal, but I recall some GM being quoted as saying that he uses the first 2 months to figure out what his team needs, the next 2 months getting it and the last 2 months winning.

I seem to remember doing this, and the weight would be something like 52% / 48% for 2nd half / first half.

Tom,

A common definition of halves of the baseball season is before and after the All Star break.  This usually comes around 89 games or so, not 81.  If your 52%/48% is based on 73/89 games then the per game importance of the 2nd half grows.

I use the mathematical half.  Half as in half.  Not the Yogi Berra half.