Pitchers and Golden Gloves

Now that the Golden Glove Award votes have been announced, there are a few things I’d like to add to my voting story.

Among the pitchers, I voted for Hiroshima Carp lefty Kris Johnson, and Rakuten Eagles righty Takahiro Norimoto. Both of these guys finished third.

The CL vote went to Tomoyuki Sugano of the Yomiuri Giants for the third straight year, while Takayuki Kishi of the Rakuten Eagles won his first. I had Sugano rated very low, because he was below average in the three hard counting percentages I measured (see below).

One thing about Golden Gloves is that reputation goes a long way. Had he foiled a single bunt all season, he would have rated much higher. But here’s the thing, and John Gibson (@JBWpodcast) and I discussed this Monday on our latest podcast. A reputation has value on the field — if opponents limit tactical options to avoid someone’s perceived strength. I know it’s backward, but that happens.

I once calculated that by throwing out runners at an astonishing rate, Hall of Fame catcher Atsuya Furuta was helping opponents create runs, because they abandoned trying to steal against him. This meant his Yakult Swallows were given fewer easy outs on defense that other clubs were getting when opponents tried to steal against them.

If opponents don’t want to risk bunting against Sugano, that’s something in his favor, I suppose. But my pick, Johnson, has been outstanding at starting double plays the past two years and fields bunts really well, and has an outstanding number of plays considering the number of untaken ground balls in the infield and singles that aren’t stopped before getting to center field.

Kishi was, before I did a more thoughtful analysis, my first choice, above average in four of the five measures I looked at. He and Sugano are both reasonable choices, I suppose.

I don’t think too many people care about which pitcher wins an NPB Golden Glove, but the information is so scarce, I thought I’d contribute to what little discussion there is by trying to answer a few questions.

Because tomorrow is the deadline for NPB Golden Glove voting, I’ll throw some things out there. There’s very little info available for pitchers so I spent way too much time on this.

I wanted to measure how pitchers compared to their NPB peers in errors, bunts foiled, double plays started.

Those three are concrete, because we know the number of bunt attempts each pitcher fielded and about how many times each pitcher got a ground ball in a GDP opportunity – and how many of those weren’t taken by one of his teammates.

I have two more categories that are iffier.

1. The number of “bonus” plays a pitcher makes more than the NPB norm from balls at are :
a) infield ground balls in play to the pitcher, catcher or first baseman.
b) fly outs to the pitcher.
c) sacrifice hits.
d) Other singles that stay in the infield or get through to center field.

I don’t want to put much weight on that since this context is mostly noise.

2. The number of first base infield singles, since a fair number of these occur every season when the pitcher is not in position to cover the bag.

Again there are lots of other infield singles to first that don’t involve the pitcher, so I wouldn’t give that very much weight.

The tables below are for the pitchers who are eligible for Golden Glove votes having pitched 143 innings or appeared in 47 games. They are ranked in the order of points I assigned to each category – which I’ll put at the bottom for those of you inclined to look under the hood.

CL pitcher fielding plays above NPB norms
PL pitcher fielding plays above NPB norms

Point weights used for rankings

2.25 points for each error fewer than the NPB average per fielding chance, and for each GDP started more than the NPB average for ground balls in GDP opportunities.

2.5 points for each good additional good outcome (no fielder’s choice, hit, error or sacrifice) on bunts fielded by the pitcher.

.25 points for: every play made by the pitcher from the set of balls in play (listed above), and every additional first base infield single.

Whys, wherefores and Win Shares

So, how many win shares was that Neftali Soto home run worth, smart guy?

今年のMVP、最優秀新人選手、ベストナインの投票について各NPB選手のWin Shares(ウィン・シェア)を計算しました。Win Sharesの日本語の説明はこちです。

I’ve written in the past that I use Bill James’ Win Shares methodology to generate my short list of candidates for my postseason award votes, the MVP, Rookie of the Year, Best Nine and Gold Gloves. It’s not an easy system to use, but has two very attractive features.

  1. The wins attributed to a team’s players cannot exceed the number of wins the team achieves.
  2. Because it works from the concepts of league defensive norms at different positions rather than raw fielding numbers — that are often skewed toward poor teams that have allowed more base runners.

The system also has many inherent flaws particularly the lack of loss shares — something James is ostensibly still working on — to account for defensive responsibilities and playing time.  The proposed new system would give every player a win-loss record, where the current system has to make do with a single figure.

This comes into play when you compare, for example, two DeNA BayStars outfielders, Yoshitomo Tsutsugo and Neftali Soto. Win Shares assigns Tsutsugo, the Japan cleanup hitter, with 22.5 percent of the club’s offense win contribution, and Soto — who also hit 38 home runs but in just five months and who was vastly more productive with runners on base and in scoring position — with 21.3 percent.

But here’s the catch, Soto made 7.7 percent of the BayStars’ batting outs, while Tsutsugo made 9.4 percent, a 20 percent increase. If we were assign the two loss shares to go with their win shares, I’m guessing Tsutsugo would be 22-11, while Soto was 19-7 and probably more valuable.

This sometimes creates disconnects for a variety of reasons. Two players with identical production on teams playing in similar offensive contexts can have quite different win share totals. If one team wins more games than its totals of runs and runs allowed would suggest, that team will have more wins to divide up among its players than a team that got fewer wins than expected.

A surprise from this year’s win shares calculations involved Yakult Swallows closer Taichi Ishiyama. He finished second in pitching win shares in the Central League behind Tomoyuki Sugano. How could that be?

Because the Swallows’ parks this year gave them the highest run adjustment — meaning runs were easier to come by in their games than in any other team’s in Japan this year — Yakult’s 658 runs scored — the second-highest CL total was not nearly as impressive as their CL-worst 665 runs allowed. Win Shares estimates the Swallows scored 297 runs more than the worst offense imaginable would score, but saved 416 runs more than the worst possible pitching and defense would allow.

That gives a huge amount of the credit for their 75-win total to the pitching and defense, and since the Swallows also won more games than expected given the number of runs scored and allowed — it means pitchers who excelled for them this season could potentially be credited with contributing to more wins than expected.

According to Win Shares, the top 10 CL pitchers this season were, in order:

1. Tomoyuki Sugano,
2. Taichi Ishiyama
3. Katsuki Azuma
4. Daichi Osera
5. Yasuaki Yamasaki
6. Onelki Garcia
7. David Buchanan
8. Yasuhiro Ogawa
9. Randy Messenger
10. Kris Johnson

I would never argue that’s right, but there are bits of truth that the system can illuminate.

writing & research on Japanese baseball