Catcher Throwing and Pitcher Hold Ratings

Written by Tom Tippett
November, 1997

We've just finished assigning the 1997 ratings for stopping the running game -- pitcher hold ratings and catcher throwing ratings -- and we thought you might like to know a little more about how we come up with them.

Our approach to rating players has always focused primarily on performance, as opposed to scouting reports and reputation. Scouting reports are professional opinions about current and future skills, and they are useful. Many players have reputations that accurately reflect how well they have performed. But sometimes there's a big difference between a player's skills and reputation, on the one hand, and his performance, on the other. Our job is to rate a player based on how well he performed in a particular season.

So we look for ways to accurately measure each player's performance. When looking at how well a pitcher or catcher defends against the stolen base, it's not all that easy to find relevant information that is publicly available. It has only been a few years since companies like STATS and Elias started publishing stolen base and caught stealing totals for pitchers and catchers. That's a good start, and it helps, but it's not enough.

Let's consider catchers first. As we all know, some catchers throw out a higher percentage of potential base-stealers than others. And it stands to reason that teams don't test the best-throwing catchers as often. Most of the time, both are true -- the catcher with the strong arm is tested less often and also throws out a higher percentage of the runners. But it doesn't always work out this way.

Sometimes a catcher throws out a good percentage of opposing runners but still yields twice as many stolen bases per game as the average catcher. Should we rate that player as average, based on his percentage, or well below average, because teams run wild on him? Maybe his percentage looks better than it should because the opposition feels it can try to steal with runners who aren't in the top tier of base stealers, while only the best runners test the better catchers.

And what about the effect of the pitcher? In the Sporting News 1997 Baseball Yearbook, Kenny Lofton was quoted as saying:

"You steal on the pitcher, not the catcher. You get a good jump on the pitcher, and there's really not many catchers who are going to throw you out."

How can we be sure whether a catcher who looks good based on opposition SB/CS data is simply blessed with a pitching staff that holds runners well?

The Importance of Studying Pitcher-Catcher Tandems

We try very hard to resolve all of these issues when assigning our ratings. We start by purchasing pitch-by-pitch and play-by-play data that describe every 1997 on-field event in detail. To make sense of these data, we've written a large number of computer programs that measure various baseball skills in great detail. One such program compiles SB/CS data for pitchers and catchers individually and as pitcher-catcher tandems.

We know, for example, that Eddie Perez of the Braves yielded 48 SB and compiled 16 CS in 470 innings behind the plate this year. That means he threw out 25% of the runners who tried to steal, which is noticeably worse than the league average of 32%. And he allowed 0.92 SB per nine innings behind the plate, compared to a league average of 0.81. Based on this information alone, he looks like a below-average throwing catcher.

Note: When assessing attempt rates, we don't really go by attempts per inning. We get more precise measures by counting the number of times a runner was on base with the next base open. And we count opportunities to steal 2nd, 3rd, and home separately. But sometimes it's easier to relate to a simpler number such as attempts per nine innings, and that's why we've used that here.

But we also know that Perez had 18 SB and 6 CS when catching Greg Maddux, 6 SB and 2 CS catching John Smoltz, 3 SB and 0 CS catching Tom Glavine, and so on. This is very important information, because Maddux has never been good at holding runners close, while Smoltz and Glavine are very hard to run on. And Perez was rarely used with Smoltz and Glavine but was paired with Maddux in the majority of Maddux's starts. In fact, a full 40% of Perez's opposition stealing opportunities occurred with Maddux on the mound. It's hard for him to look good when he's usually paired with the only starter who is easy to run on.

So, for both pitchers and catchers, we look at both the percentage of runners thrown out and the frequency with which runners try to steal. And we look very carefully at the pitcher-catcher matchup information to determine how to allocate the credit or blame between pitcher and catcher. We also look at the number of pickoffs. When ranking and comparing players, we keep in mind that NL teams run more than AL teams and that both attempt rates and success rates are a little higher on artificial turf than on natural grass fields.

To illustrate how we use this information, let's go over some of the more interesting situations we encountered when assigning the ratings for the 1997 Season Disk.

Ausmus Joins Rodriguez and Johnson at the Top

It should come as no surprise that it didn't take us long to figure out that Ivan Rodriguez and Charles Johnson both deserve the Excellent throwing rating again this year. Rodriguez threw out an astounding 56% of opposing runners, while Johnson came in at 47%. It was a little more surprising to find that Brad Ausmus also qualified for the top rating by throwing out 49%.

Because Ausmus has not previously been in the top tier of catchers, we wanted to learn more. This was his first season with Houston. Was the pitching staff making him look better than he is? Would all catchers look equally good with this crop of pitchers on the mound? To find out, we compared his results for each pitcher he caught to those when Tony Eusebio, the backup catcher, was behind the plate.

The difference was startling. When Darryl Kile was pitching, Ausmus threw out 2 of 3 runners, while Eusebio nailed only 2 of 21. It's true that Eusebio caught almost four times as many of Kile's innings as did Ausmus, but that doesn't account for such a large spread. When Chris Holt was pitching, Ausmus threw out 11 of 17; Eusebio caught only 1 of 7. Overall, Ausmus caught 46 of 93 runners while Eusebio caught 9 of 45.

The Houston pitchers are pretty good at holding runners. Holt and Donne Wall are relatively new to the league but have looked quite good so far. Shane Reynolds and Mike Hampton have been around for a while and have looked good in the past. Only Kile's career shows a slightly-below-average ability to control the running game. But every one of these pitchers posted better numbers throwing to Ausmus than with the other catchers they've had in 1997 and in previous seasons. So we were able to conclude that Ausmus did, in fact, deserve the lion's share of the credit for shutting down the running game.

The matchup numbers are actually more helpful in evaluating Kile and Holt than Ausmus and Eusebio. It's obvious that Ausmus throws much better than Eusebio, and we didn't need the matchup information to figure that out. But the matchup data tells us that Kile's weak individual numbers (20 SB allowed in 24 attempts) were compiled almost entirely with Eusebio behind the plate, and that he looked quite good when throwing to Ausmus.

Brian Johnson Gets a Little Help in San Francisco

Brian Johnson is another interesting case. While splitting his time between Detroit and San Francisco, he threw out 26 of 87 runners, or about 30%, which is very close to the major league average. If this was the only information we used, we'd be tempted to give him an Average rating. But there's more to the story. While in Detroit, he threw out 8 of 39 runners; in San Francisco, he gunned down 18 of 48. This might be enough to make us suspect that the SF pitching staff had something to do with his success, but the matchup information makes it clear.

The 1997 SF staff has an unusual number of pitchers who are very difficult to run on. Terry Mulholland has given up 4 SB in the past four seasons, during which he pitched 659 innings. In that span, he's retired 75% of the runners who dared to challenge him. In the same four-year period, Kirk Rueter has allowed only 9 SB in 433 innings, while catching 67% of those who tried, and Mark Gardner has erased almost half (48%) of the potential thieves. And each of these guys played for other teams in this period, so we cannot attribute their success to the SF catchers. It wasn't hard for us to conclude that Johnson's ability in SF to throw out 18 of 48 (37%) wasn't as impressive as it looked. But it wasn't awful, either. Weighing this performance against his Detroit results and career history led us to give him a Fair rating for 1997.

Widger, Juden and the Montreal Pitchers

We'd be remiss if we didn't talk about Chris Widger, Jeff Juden and the Montreal pitching staff. Widger's individual numbers were awful -- 116 SB and 23 CS -- for both frequency and percentage. And he had almost no playing time in previous seasons to give us a good read on him. Was he really this bad? Or was he the victim of a pitching staff that can't hold a runner to save its collective life? I've twice read quotes from manager Felipe Alou that said, effectively, "when our young pitchers are struggling a little, we tell them to forget about the runners and concentrate on the hitters." Maybe that's the problem.

The matchup information helps. The first thing we noticed is that Widger caught Jeff Juden every time he pitched, and that this tandem gave up an astonishing 40 SB in 45 attempts. This makes Widger's numbers a little more reasonable-looking, since it's clear from other data that Juden is atrocious at holding runners. In only 31 innings in Cleveland, he gave up 9 SB in 10 attempts with Borders, Alomar and Diaz catching. Juden's numbers were equally bad in Montreal and Cleveland, and we know that both Borders and Alomor throw pretty well. And Juden's performance was almost as bad in the three previous seasons. Seems like Juden is the problem and that Widger is an innocent victim.

Felipe Alou is often touted as one of the best managers in baseball. He knew coming into the season that Juden had a history of giving up SB in large numbers. And he arranged it so that Widger caught every single inning that Juden pitched in Montreal. Could it be that Alou viewed Widger as his best-throwing catcher and felt it necessary to pair him with Juden just to give his squad a fighting chance to keep a runner on first?

Jim Bullinger is another Montreal starter with a history of failing to hold runners well. In 1995-6, the two years in which he was primarily used as a starting pitcher in Chicago, only 8 of 46 runners (17%) were caught on his watch. He was a little worse in Montreal (6 of 40, 15%). But not enough worse to make it appear that Widger and Fletcher (who caught Bullinger about half the time) are substantially worse than the Chicago catchers were the previous two years.

Ugueth Urbina is one of the promising young Montreal pitchers who lets opponents run wild. He gave up 8 steals in 8 attempts, running his career totals to 5 CS in 35 attempts (14%). For the number of innings he's pitched, that's about double the number of successful steals per game. And the low percentage speaks for itself.

On the other hand, Mike Johnson moved from Baltimore to Montreal during the year. In Baltimore, he nailed 2 of 5 runners in 39 innings while throwing to Chris Hoiles and Lenny Webster, neither of whom is noted for his arm. In Montreal, mostly with Widger behind the plate, he caught only 1 of 12 in 50 innings. That makes Widger look like the culprit.

So where do we end up on Widger? Although his overall numbers ranked him last in the major leagues by any measure, we decided to give him the Fair rating. We felt that Juden, Bullinger, Urbina and other Montreal pitchers were mostly responsible for his numbers looking so bad. But there was no basis for rating him any higher than that. He made Mike Johnson look awful, and while he didn't play all that much with Seattle in 1995-6, he didn't throw runners out there, either. His 1997 results, after adjusting for the effect of catching Juden, weren't much different from those of Darren Fletcher. We've always rated Fletcher below average, and Widger seems to be in the same category. Hence, the Fair rating for Widger.

Texas and Florida Pitchers

It's essential that we consider who a pitcher is throwing to when evaluating his stats. John Burkett caught 9 of 17 runners this year. That would qualify many pitchers for an Excellent hold rating. But he was throwing to Ivan Rodriguez 92% of the time, and Rodriguez throws out more than 50% with plenty of other guys on the mound, too. Burkett's career numbers are nothing special, either, despite throwing to Rodriguez and Charles Johnson the last couple of years. So he gets an Average rating for 1997.

Kevin Brown is another good example. In recent years, he's pitched for Texas (through 1994), Baltimore (1995), and Florida (1996-7). That means he's spent most of his time throwing to Rodriguez and Johnson. It's hard to know how much worse his numbers would look if he was throwing to average catchers all this time, but you can bet that they'd be down. And in 1997, he allowed 9 of 12 runners to advance with Johnson catching. Most pitchers are better than this with Johnson behind the plate. But it could have been worse -- Rob Nen was 0 for 5 with Johnson catching, and the average pitcher gives up more than 9 SB in the number of innings that Brown threw to Johnson. Looking at his career numbers and his 1997 performance, we concluded that he should get the Fair rating this year.

Personal Catchers

Several teams paired one pitcher with the same catcher almost every time out. In Baltimore, Scott Erickson threw to Lenny Webster almost 90% of the time. In Boston, Aaron Sele and Scott Hatteberg were paired over 70% of the time. For the Cardinal, Mike DeFelice caught Alan Benes over 80% of the time. This can make it difficult to isolate the contributions of the two players. In these cases, we look to the catcher's performance with other pitchers and to the career performances of both players to help us determine how to rate them.

Rookies and Others with Limited Playing Time

My last comment is about players with very limited playing time in 1997. When rating these players, we put much less weight on the numbers. For rookies getting their first taste of the big leagues, we try to find as much information as we can about their performance in the minor leagues. But there aren't many writers or scouts who talk about a minor-league pitcher's move to first base -- they're usually busy talking about his control and velocity and whether his fastball moves in the strike zone. So it can be tough to get a handle on these guys sometimes.

For established major leaguers who didn't play much in 1997, we look to their career record. Sid Fernandez, for example, gave up 2 SB in his only 1997 start. We generally don't like to give the Poor rating based on one start, since anything can happen in one game. But Fernandez has a long history of giving up a lot of SB, and we rated him Poor in 1995 and 1996, so we rated him Poor again this year. Mike Bertotti also gave up 2 SB in two tries this year (though he faced 50% more stealing opportunities than did Fernandez), but his prior record was quite good (no steals allowed in 1995, 1 for 2 in 1996), so we rated him Average in 1997.

Closing Thoughts

My goal with this article is to give you some insight into the way we look at catchers and pitchers. We believe it's impossible to rate hundreds of players accurately without seeking the best information available. In this case, we had to write a program to compile the information because nobody publishes SB/CS data broken out by pitcher-catcher tandems. Even with the best information, it would be a mistake to generate the ratings using some mathematical formula and leave it at that. So we carefully review the results for every player -- about 100 catchers and 500 pitchers per season -- to determine what the information is really telling us.

In the end, the ratings represent our opinions about the player skills in question. We can't prove beyond a shadow of a doubt that Kevin Brown should be rated Fair this year. That's impossible. And with hundreds of players to rate each season, it's inevitable that we'll get a few of them wrong. But we sincerely believe that by applying our best baseball judgment to the best information available, we come very, very close to capturing what really happened during the season. And we're convinced that we get much closer to the truth with these methods than we could by relying on popular opinion, which is what we'd be doing if we based our ratings on what we hear on TV and read in the baseball press.

Thanks for reading.

Copyright © 1997. Diamond Mind, Inc. All rights reserved.