Bert, Uncle Charlie, and the Home Run

Following up on my previous article on Bert Blyleven, I thought I’d share some numbers I crunched regarding Blyleven being homer prone, as he supposedly was. I always suspected that Blyleven was prone to giving up home runs because as a curveball pitcher, he was liable to hang one from time to time and present the hitter with a juicy target. The homer-proneness of some curveball-reliant pitchers I’ve seen a lot of recently, like Scott Downs and Dave Bush, led me to surmise that curveball-first pitchers were just more prone to the home run than other pitchers, and that they would just have to make up ground with their fastball-throwing colleagues in other ways.

As it turns out, my supposition was wrong. I did a study involving a group of pitchers who primarily relied on their curveball, using the fine Neyer/James Guide To Pitchers as my bible and selecting only players who:

1. primarily played after 1946, to give a relatively consistent environment for home runs;

2. had a curveball listed by Neyer/James as their #1 pitch, at least throughout most of their career.

I tried not to inject any of my own biases by making separate determinations of whether a guy was really a curveballer or not. I picked 30 such pitchers (everyone in alphabetical order up to Wayne Garland) and looked at their innings pitched, ERAs, walks and home runs allowed, all in absolute terms and relative to their league.

Going the players and assigning numbers, I already knew that my supposition wasn’t going to prove correct. Yes, there were quite a few curveballers (even some good ones) who were really gopher-prone and managed to be successful in other ways. Call these guys the “John Cerutti types.” But there was an equal number of guys who used Uncle Charlie in an entirely different way … much wilder than the first group and very hard to hit for power. Call these guys the “Mark Clear types.”

Anyway, the upshot of it all was that the Mark Clear types and the John Cerutti types were a wash. The curveballers as a whole didn’t look to be giving up that many home runs, probably no more than average. (Taking into account that these were a good group of pitchers, due to the selection bias of getting into the Neyer/James book, they probably gave up 5% more than average.)

I intended to compare these 30 guys to a group of pitchers who had similar numbers and effectiveness but who were fastball pitchers. Then a funny thing happened which gets to the real point of what I’m writing about.

Thinking that I would compare the group of curveballers to Blyleven himself, I threw in his numbers at the end of the spreadsheet. And what did I see, immediately, that made me instantly realize I could have saved myself a hell of a lot of time?

Bert Blyleven gave up 14 fewer homers, over his career, than an average pitcher in his leagues.

Well smack my ass and call me Susie. Ol’ gopher ball Bert was above average at preventing the home run? Why did no one tell me this?

What bothers me is, ever since this research project was stimulated by an offhand remark of Mike D of Batter’s Box, that Bert had a “gopher ball problem,” and I had never bothered to question the underlying assumption. Thinking back over how I could have missed this relatively obvious fact, three things stood out as possible explanations (other than my own cluelessness, which obviously played a considerable role):

1. Blyleven does own the major league record for home runs allowed in a season, with 50 in 1986.

2. My own “mental picture” of Bert Blyleven was formed around this time, when he had moved from the indifferent Indians to the improving Twins. In fact, when Blyleven really captured my attention it was 1987, the Twins were pushing towards an eventual World Championship, and Blyleven had just come off setting that home runs allowed record. The gopher ball was constantly discussed as his Achilles heel. The impression couldn’t help but color my impression of the man.

3. I had become seduced, as I had been for some time, by the “hanging curveball” thesis, again letting it color my impression of Blyleven.

Finally, I think there’s a general impression that Blyleven was homer-prone. There must have been; that’s why Denyszyn would have brought it up, and why no one questioned him when the remark was made, and why I’ve heard similar sentiments being expressed elsewhere.

Recently, I wrote a short post over at my blog that talked a little bit, in a preliminary way, about how we tend to build our knowledge about baseball and ballplayers. One of the most deep-seated biases that we face when we are thinking about players is how statistics can create a powerfully misleading impression of who a player is, and Gopher Ball Bert is a perfect example of this. Bert spent two seasons giving up lots of home runs, and because that fact sticks out in our memory, we use it as a label for the man, instead of looking at his long career in which he resembled a Tommy John type more than a John Cerutti type.

If a guy has played three full years and hit .288, .328, and .304, some people will inevitably describe him as a “.330 hitter.” If a guy has won 20 games twice in 15 seasons, every third mention of his name will employ a description of him as a consistent 20-game winner. We tend to use otherwise-memorable numbers as the standard not by which we measure players, but by which we label them and categorize them. When I was in my early teens, people constantly talked about Nolan Ryan as a guy who would go out and win 20 games a year; at that time he hadn’t won 20 in over a decade and had consistently won 11 or 12 games a year.

I had a similar discussion with a friend recently about Owen “Chief” Wilson; looking at a picture that I thought was of Wilson, he remarked that it couldn’t be; this player didn’t look fast, and Wilson holds the single-season triples record. Now it’s not fair to judge someone based on their naive impression of a ballplayer who last played 90 seasons ago and has been dead for over 50 years. But if I could find a more perfect example of a judgment based entirely on statistics that’s both wrong and a misjudgement, that would be it. In fact, Wilson wasn’t an outstanding triples hitter. Good, yes. But Wilson, despite the 36 triples, had a reputation as a slow-footed slugger; triples in the deadball era were a power stat, usually generated by walloping the ball over the heads of outfielders who played close in, and were doubly so in the cavernous ballparks that Wilson played in.

It’s easy to let a few “benchmark” numbers mislead us about a player’s true skills or performance.

Anyway, back to Blyleven. Thinking that maybe his homer-proneness would show up more if I compared him to his peers, the superior pitchers, I decided to look at guys who were Bert’s equals as performers. I compiled a list of pitchers (from 1946 to 2003 only) who pitched over 2,000 innings and had an ERA+ of 110 to 120 (Blyleven’s was 115 according to the Sabermetric Encyclopedia) and looked at their home runs allowed rates.

Blyleven ranked 23rd of the 35 pitchers in home runs allowed. Not great, for sure, but not at all bad either. Some of these guys predicated much of their repertoire on preventing home runs; both Dean Chance and Tommy John, for example, owe almost their entire career performance to above average home run prevention. If Chance had had an ordinary number of home runs allowed, he’d have been about 17 runs above average instead of 129. If Tommy John had had an ordinary number of home runs allowed, he’d have been about eight runs below average instead of 173 above.

The whole collection of 35 pitchers averaged 10% above average at home run prevention, while Bert was 3% better. So compared with his fellow top-quality pitchers, he was in the general range in terms of home runs allowed but not particularly homer-prone. The pitcher on the list who was most homer prone was John Candelaria, Blyleven’s teammate in Pittsburgh for three seasons.

So in the end, I think it would be unfair to say that Blyleven had a “gopher ball problem.” Call it something that, as Studes would say, I didn’t know last week.

Players between 1946 and 2003 with an ERA+ between 110 and 120 and more than 2000 IP, ranked by home run rate vs. the league average.

                                 RATE     HRA    LEAGUE    IP       ERA+    
1    Dean Chance                 165      122      202   2148        119   
2    Al Leiter                   143      169      242   2075        119   
3    Tommy John                  143      302      431   4710.1      112   
4    Steve Rogers                143      151      215   2837.2      115   
5    Nolan Ryan                  138      321      444   5386        116   
6    Sam McDowell                138      164      226   2492        110   
7    Mel Stottlemyre             137      171      234   2662        116   
8    Tom Glavine                 136      268      364   3528        119   
9    Orel Hershiser              129      235      303   3130.1      115   
10   Joe Horlen                  127      145      184   2003        111   
11   Dave Stieb                  127      225      285   2895.1      118   
12   Dwight Gooden               125      210      264   2800.2      113   
13   Jon Matlack                 121      161      195   2363        118   
14   Chuck Finley                118      304      359   3197.1      115   
15   Ken Forsch                  114      155      176   2127.1      111   
16   Tom Candiotti               112      250      279   2725        111   
17   Gaylord Perry               111      399      443   5350.1      117   
18   Jerry Koosman               109      290      315   3839.1      110   
19   Frank Lary                  106      197      209   2161.2      111   
20   Mike Cuellar                106      222      236   2807        113   
21   Vida Blue                   104      263      274   3343.1      113   
22   Early Wynn                  104      313      324   3851        112   
23   Bert Blyleven               103      430      444   4970        115   
24   Steve Carlton               101      414      416   5217.1      113   
25   Bob Feller                  101      164      165   2307        119   
26   Bob Welch                   100      267      266   3092        111   
27   David Wells                  99      330      328   2826.2      111   
28   Virgil Trucks                99      174      172   2306        113   
29   Don Sutton                   95      472      446   5282.1      114   
30   Johnny Sain                  93      172      160   2028        113   
31   Jim Bunning                  93      372      346   3759.1      113   
32   Robin Roberts                93      505      470   4689        115   
33   Dennis Eckersley             87      347      302   3285.2      113   
34   Don Newcombe                 87      252      218   2154.2      111   
35   John Candelaria              85      245      209   2525.2      113