Magic Closer Dust

Arthur Rhodes signed with the Oakland A’s as a free agent this off-season and it was immediately announced that he would be their closer. I remember nearly everything I read about the signing included the exact same statistic: During his career, Rhodes was 17-for-44 in “save opportunities.”

It was almost as if the entire baseball-writing world got a memo when Rhodes signed, ordering them to quote that one stat in every article on Rhodes. Here is a small sampling…

From Phil Rogers of ESPN.com:

Rhodes, a career setup man, has never had more than four saves in a season and is 17-for-44 in save situations over his 13-year career.

From Laurence Miedama of the San Jose Mercury News:

Rhodes has converted 17 of 44 save opportunities in his career (Foulke was 43 for 48 last season) but is considered to have the mentality and presence to finish games.

From Rich Hurd of the Contra Costa Times:

Nor does Rhodes have a great track record with saves. He’s converted only 17 of the 44 save chances he’s received in his career, a stat that has followed him like an albatross since he signed a three-year, $9.2 million contract Dec. 22.

From Kent Schacht of MLB.com:

Rhodes, 34, has been Seattle’s primary left-handed setup man for the last four seasons, but never been a full-time closer. He has 17 saves in [44] opportunities in his 13-year big league career and is known for his nasty fastball.

In most instances, the stat was used to express doubt that Rhodes could do the job Oakland was asking him to do. That stat is a great example of how numbers, when used incorrectly, can be extraordinarily misleading.

Going 17-for-44 in save chances is so horrendously bad that it is almost literally impossible if the pitcher in question was actually used as a closer. No decent major-league pitcher, used to close out leads of 1-3 runs in the ninth inning, would convert the save just 38.6% of the time.

So how did Rhodes, who is certainly better than a decent pitcher, manage to do it? Quite simple, actually. He wasn’t used as a closer. When you’re a setup man, which is the role Rhodes has filled for many years, you typically only get a chance to blow saves.

In other words, if a setup man enters a 7-6 game in the eighth inning and gives up a run, he gets a “blown save” on his record. But, even if he had pitched a flawless eighth inning, preserving the one-run lead, he would not have gotten a “save” anyway. Instead, he would have been given a “hold,” and then the team’s closer would have come in to pitch the ninth inning.

So it’s a no-win (or no-save, to be exact) situation. If you are not successful, you get charged with a blown save. If you are successful, you don’t get a save. It’s no wonder Rhodes’ history in “save situations” is so poor.

While with the Mariners in 2002 and 2003, Rhodes went 5-for-13 in save opportunities. That is a success rate of just 38.5%, which is nearly identical to his career rate of 38.6%. Let’s take a closer look at those 13 save opportunities to see if they’re as misleading as I think they are.

A Hardball Times Update
Goodbye for now.

First, let’s look at the eight games in which he got a “blown save”…

1) April 2, 2002: Rhodes entered the game with two outs in the seventh inning and the bases loaded — Seattle leading 3-2.

In order to get a save in this situation, Rhodes would have had to get out of the seventh without allowing a run and then pitch the eighth and ninth innings (a total of 2 1/3 innings pitched). In other words, no matter how he’d done, Rhodes wasn’t getting a save. Rhodes ended up getting a blown save because he gave up one hit, a single, in a game that was at least seven outs away from being finished.

2) April 15, 2002: Rhodes entered the game with two outs in the seventh inning and runners on first and third — Seattle leading 10-8.

Again, this was in no way a legitimate save chance, as Rhodes would have had to pitch 2 1/3 innings in order to record a save.

3) May 9, 2002: Rhodes entered the game with no outs in the eighth inning and a runner on first — Seattle leading 4-3.

Rhodes would have had to record six outs to get a save, which makes it extremely unlikely (although not as unlikely as the first two instances) that it was a real save opportunity.

4) July 20, 2002: Rhodes entered the game with no outs in the seventh inning and runners on first and second — Seattle leading 5-2.

In order to get a save in this situation, Rhodes would have had to pitch three entire innings. In other words, another fake save “chance.”

5) August 23, 2002: Rhodes entered the game with one out in the seventh inning and a runner on third base — Seattle leading 2-1.

Again, not a real save chance. Even if he’d have been perfect, Rhodes would have had to pitch 2 2/3 innings to get a save. Incidentally, this is perhaps the weakest of Rhodes’ “blown saves.” He struck out the first batter he faced, but then allowed an infield single, which scored the tying (inherited) run from third base.

6) April 20, 2003: Rhodes entered the game with no outs in the eighth inning and runners on first and second — Seattle leading 5-3.

This is one of those “possible but not likely” save chances. Even if Rhodes had been perfect in his performance, he would have had to pitch a total of two innings to get the save. Not only wasn’t Rhodes Seattle’s closer at the time, how many guys who are closers pitch two innings for their saves, let alone two innings after coming into a game with two runners on base?

7) August 10, 2002: Rhodes entered the game with one out in the seventh inning and a man on second base — Seattle leading 6-4.

Again, he simply had no chance at a save, even if he would have been perfect, as he would have had to pitch 2 1/3 innings.

8) September 2, 2003: Rhodes entered the game with two outs in the eighth inning and a runner on first base — Seattle leading 7-5.

Unlike all the other situations listed above, it is somewhat possible to think Rhodes may have been given a chance to actually record a save in this game. I don’t think it’s likely, but I do think it is reasonably possible. He would have had to pitch 1 1/3 innings with a small lead, despite the fact that, by this time in the season, Shigetoshi Hasegawa was established as Seattle’s closer.

So, those are Arthur Rhodes’ eight blown saves from 2002 and 2003. Hopefully you can see just how silly this entire thing with blown saves and save percentages is when it comes to guys who are not acting as closers. Rhodes blew eight saves, and there’s a chance that one of them actually came in a situation where he would have had a chance to save the game had he pitched perfectly.

Now, let’s look at the five games in which he recorded a save…

1) June 27, 2002: Rhodes entered the game with no outs in the ninth inning and no one on base — Seattle leading 7-4.

This is what could best be described as a typical save situation in this era of baseball. 1-3 run lead, no outs, no one on base, ninth inning. Rhodes pitched a 1-2-3 inning for the save.

2) September 24, 2002: Rhodes entered the game with no outs in the ninth inning and no one on base — Seattle leading 8-7.

Another textbook save situation. Rhodes had another 1-2-3 inning, picking up the save.

3) June 12, 2003: Rhodes entered the game with no outs in the ninth inning and no one on base — Seattle leading 1-0.

Another typical save situation. Rhodes got three outs without allowing a run to pick up the save.

4) June 14, 2003: Rhodes entered the game with two outs in the eighth inning and a runner on first base — Seattle leading 4-3.

A potential four-out save situation, similar to the one Rhodes blew on September 2, 2003. Rhodes got the final out of the eighth and then pitched a 1-2-3 ninth for the save.

5) June 18, 2003: Rhodes entered the game with no outs in the ninth inning and no one on base — Seattle leading 2-0.

Typical save situation. He pitched another 1-2-3 inning, getting the save.

The numbers will forever say that Arthur Rhodes went 5-for-13 in save opportunities in 2002/2003, just as they said he was 17-for-44 in save opportunities during his career coming into this season.

In reality, however, Rhodes went 5-for-6 (83.3%) in situations that could be called legitimate save chances.

Even the fact that he went 0-for-7 in “fake” save chances in 2002/2003 is misleading, as it makes it seem like Rhodes was horrendous in those situations. Actually though, he was quite good.

In those seven games, had Rhodes been successful, he would have gotten a “hold.” From 2002-2003, Rhodes had a total of 45 holds. Put those together with his seven botched hold attempts that resulted in “blown saves” and you get 45-for-52, which works out to a very impressive 86.5%.

Holds are, of course, an even more worthless stat than saves, but that’s an entirely different issue.

If a pitcher is asked to fill a role that involves almost never getting a chance to get a save, but almost always involves potentially getting a “blown save,” doesn’t it just make sense that their ratio of saves-to-blown saves isn’t going to be particularly impressive?

Saying Rhodes was 5-for-13 in save chances from 2002-2003 sounds quite a bit different than saying he was 5-for-6 in real save chances and 45-for-52 in hold chances, and I suspect, if you looked back through his entire career, you’d find much of the same in regard to his 17-for-44 overall save record.

Baseball statistics are wonderful, powerful things that can help us understand and learn so much about the game. Just like anything else in the world that has some power, baseball stats can be dangerous in the wrong hands.

One baseball writer shares the wisdom that “Arthur Rhodes is just 17-for-44 in save chances during his career,” just about every other baseball writer in the universe picks up on that stat and uses it in their own writing, and suddenly Arthur Rhodes is a guy who has choked when given a chance to close games. And, if the last two years are any indication, that couldn’t be further from the truth.

Oh, in case you’re wondering, Arthur Rhodes is 2-for-2 in save opportunities for the A’s this season.


Comments are closed.