# The Baseball Graphs Blog

##### Saturday, February 28, 2004

## A Graphical History of Relief Pitching

One of the most intriguing stories in the history of baseball is that of relief pitching. Intriguing, because relief pitching has changed so much over the course of baseball as we know it. In fact, the use of relief pitchers has never really stopped changing. So writers have a lot of fun studying it, analyzing it and telling it.

Among those who have told the story well are John Thorn, Bob Cairns, Bill James and, most recently, Mike Carminati. In fact, Mike told it so well that he was nominated (twice!) for a Primey.

Mike likes to use a lot of tables in his articles, and I thought it might be fun to see if I could graph a few of them. With his permission, of course. So here we go.

**Click for more...**

##### Thursday, February 19, 2004

## Current Player’s HOF Chances

Here’s an article that looks at the Hall of Fame chances of current players, using Win Shares as the criterion.

It includes a nice list of current player’s total Win Shares—I didn’t realize Mark Grace has so many.

Also, it reinforces the idea that pitchers should be judged on a different scale, as we’ve suggested in the replacement level article. Compare and contrast Steve Finley and Randy Johnson, for instance.

Don’t worry. This article is a lot shorter than Mike’s Hall of Fame series.

##### Tuesday, February 17, 2004

## A Fielding Analysis that’s not all wet

Avkash has done a super job of analyzing fielding stats at his Mets blog, the raindrops. He’s included fielding Win Shares, as well as the other cutting-edge defensive metrics such as UZR.

I need to review this more closely, but it is exactly how I was thinking of tackling Win Shares fielding statistics. Great job.

##### Saturday, February 14, 2004

## A Matter of Scale

In a recent column comparing the careers of Dennis Eckersley and John Smoltz, Rob Neyer compared the Win Shares and Runs Scored Above Average (RSAA) of the two. Isolating their years as starters, Rob found that Smoltz had reached 181 Win Shares and 191 RSAA, while Eck’s numbers were 169 and 119.

In other words, Smoltz had about 60% more RSAA, but only 7% more Win Shares. As Rob said, “I’m not sure why Smoltz has a huge advantage in RSAA and a small one in Win Shares, but either way he comes out ahead…”

I was intrigued by the Win Share angle, and I quickly looked at a couple of the usual culprits, such as differences in their team’s “won/loss differentials” and the fact that Smoltz might be disadvantaged by having to bat. But those weren’t the issues at all.

The reason Smoltz has a huge advantage in RSAA is because it uses a different scale than Win Shares does. This is something Bill James discussed in the Win Shares chapter called “The Baseball Player as an Iceberg” and it’s an important thing to remember.

**Click for more...**

##### Friday, February 13, 2004

## Baseball Crank’s AL East Win Shares

The Baseball Crank continues his review of Established Win Shares, this time in the AL East.

This has been a fun series by the Crank, and he does a good job of using Win Shares, while also acknowledging their limitations in an exercise of this kind.

##### Thursday, February 12, 2004

## Replacement Level

In the Extreme Win Shares article, I used a replacement level equal to 50% of each player’s Win Shares Baseline (which is the average number of Win Shares in the league, given that player’s playing time). Why 50%, instead of the more common 85% (or thereabouts)?

Because Win Shares already has a pseudo-replacement level buried inside it, based on the Marginal Runs concept. So I took a stab at the additional replacement level required to bring the overall system close to something that made sense. And 50% was as good a number as any.

So how did I do? To find out, I subtracted each team’s total Win Shares above replacement from total Win Shares to calculate the implied replacement wins. Win Shares minus WS above replacement equals replacement Win Shares.

The average replacement Win Shares were 122, or 41 wins (rounded off). This is pretty good, I think. The Tigers, who were about as replaceable as you can get, won 43 games last year.

Team, by team, the results varied from 39 (Tigers) to 42 (Astros). This wasn’t quite as consistent as I had hoped for—I think I was hoping for a spread of one win—but it’s not bad.

Looking over the data, I also think you can discern a decent rule of thumb for individual replacement Win Share levels: eight Win Shares for a full-time position player, six Win Shares for a starter, and five Win Shares for a closer.

I know this can and should be improved, but I’ll take it for now.

##### Wednesday, February 11, 2004

## Extreme Win Shares

It’s been a fun offseason, but I’d like to stop and catch my breath a bit. In fact, this is probably a good time to add up the Win Share calculations we’ve explored so far this offseason:

We applied park factors differently (not much of an impact).

We opened the door for negative Win Shares, which are important to maintain validity.

Played around with Game Shares and created the Win Shares Baseline (or, Expected Win Shares, to use Charlie Saeger’s terminology). This sets the stage for replacement level analyses.

Looked at relievers, and concluded that one of the pitching claim point calculations favors relivers by giving extra credit for saves. We took out that calculation altogether.

Posted to a Baseball Primer discussion about the relative value of starters and relievers, which included the “unpublished study” that relievers probably achieve ERA’s at least 0.60 lower, on average, than they would achieve in a starting role. I’m tempted to say that any unpublished study by Tangotiger is good enough for me, but I don’t want him to get cocky or anything.

Discussed the relative allocation to offense and defense, with no definitive conclusions. However, there was a lot of interest in the “Fibonacci number” (61% instead of 52%).

- Played with the split between pitching and fielding, including my own approach based on FIP and DER.

We’ve played with all the aspects of Win Shares that I wanted to get to, except for leveraged innings and individual fielding positions, and the latter is going to be a LOT of work. So, I figured, why not rest a bit, pull all these analyses together and see what we get?

I call these Extreme Win Shares, and they’re based on all of the above factors, including:

Establishing replacement level as 50% of a player’s Win Shares baseline. I would tell you that a lot of intricate analysis went into that figure, but that would be lying. Actually, a little fairy suggested it.

Using my proposed solution for splitting Win Shares between pitching and fielding.

- Using the Fibonacci number (61% and 161% instead of 52% and 152%) to allocate Win Shares between offense and defense. I’m not sure this step is really appropriate, but what the heck.

So here are the 2003 Extreme Win Shares, expressed as Win Shares Above Replacement:

- American League Extreme Win Shares

- National League Extreme Win Shares

I’ll be talking about Extreme Win Shares a little bit more in the next one or two articles, but you can see that pitchers are upgraded, relievers are downgraded, and players in important fielding positions (shortstop, catcher) are upgraded.

What do folks think? Is this an improvement?

##### Sunday, February 08, 2004

## Pitching and Fielding, The End

Bill James, Charlie Saeger and I have developed, somewhat independently of each other, our own systems for allocating defensive Win Shares to pitchers and fielders. A summary output of our systems for 2003 can be found in this table.

We began this Win Shares pitching/fielding split adventure in this December article, asking if pitchers are undervalued relative to fielders. The time has come to ask the question again, and to review the results of the three systems. So let’s start with a graph.

**Click for more...**

##### Friday, February 06, 2004

## Pitching and Fielding, Part Four

I’ve been preparing an article that compares the relative outcomes of the three systems for splitting defensive Win Shares between pitching and fielding. As a result of comparing my proposed system with the original James version, and the Charlie Saeger version, I realized that some of my formulas were “overcorrecting.”

I won’t get into the math right now, but suffice to say that I was essentially double-counting my adjustments for errors, double plays, outfield assists and passed balls. So I’ve run the numbers again, and I have attached a new table comparing the three systems.

Once again, I’ve highlighted the teams that seem to vary the most, and I’ll follow up over the weekend with a summary of the differences between our systems.

**Click for more...**

## If nominated, I will not…

The good folks at Baseball Primer have nominated this site for a Primey, which is a real honor in my book. Baseball Primer has been an inspiration and support for both Pete and me.

Unfortunately, our research has taken the form of a blog, and you voters might find it hard to navigate. So here is a “cheat sheet” of some research articles:

Graphical History of Defense

FIP and DER background

Great Fielding Teams

Baseball Economics

Win Shares above average

The Need for Negative Win Shares

To vote for the best “baseball stuff on the web,” run on over to Baseball Primer.

##### Wednesday, February 04, 2004

## Don’t bank on Pudge

## The Giant’s (and Neifi’s) Win Shares

The guys at Fogball take a look at next year’s Giants, including a special look at Neifi’s Win Shares.