you're reading...

Giving 1930s and 1940s players back their missing value mojo

So you have a pile of missing value for a bunch of 1930s and 1940s ballplayers. Now what?
Let’s have a look at some key guys from the 1920s, 1930s, and 1940s whose new PBP data gives us a better glimpse at any hidden value they may have accrued. We looked specifically at players who were either:

  • already elected to the Hall of Miller and Eric, the Hall of Fame, the Hall of Merit, or the Hall of Stats
  • near the in/out line in our own personal position rankings
  • still receiving votes for the Hall of Merit
  • dudes we liked for whatever reason.

We’ve broken them down by position below.
In general, doing this work suggests that BBREF’s regression scores for baserunning may, if our math is reasonable, suppress a good deal of baserunning value—or stinkiness. It appears that BBREF bases its formula for pre-PBP running on steals, steal attempts, and/or success rates. That’s how Ernie Lombardi, a strong candidate for the slowest man to ever don cleats, is listed with positive running value. The reality, as we’ll soon discover, is likely far worse for Lom. Generally, we found a lot of very positive baserunning value. This may stand to reason since we examined the best of the best from this timespan, and good players are often good athletes. Or we need to review our mathematics…. Oh, and here’s this other item. The thing that drives baserunning value isn’t what you probably think it is. Just ask Joe Sewell.

In the tables below, “NOW” refers to a player’s value as calculated by me with my adjustments but without the “missing” value. Which means that “EST” includes the value we’ve calculated for running, GIDP avoidance, and throwing.

           NOW  EST   NOW    EST   NOW   EST
Berra       77   77    62     62     7     7
Hartnett    73   69    56     53     8    10
Dickey      71   69    56     54     9     9
Cochrane    65   67    55     56    10     8
Lombardi    59   50    46     40    16    24
Lollar      41   41    35     35    34    34
Cooper      40   38    34     33    37    37
Ferrell     41   40    32     32    39    40

Two of these catchers are polar opposites. At least among backstops. On one hand, Mickey Cochrane appears to have positive baserunning value, unlike pretty much every other catcher here. He’s also got positive rDP value, which even fellow lefty swinger Bill Dickey doesn’t. Black Mike is the only catcher to gain value in this group.

Then there’s Ernie Lombardi. We’ve run through his story before, but believe it or not, I underestimated how bad a baserunner he was. Here’s how the sad story of Schnozz’s plummeting value goes. Lombardi appears to have surprisingly un-bad stolen base value. Something like -1 against the league in his number of steal attempts. He was only picked off four times in his career, while I figure a league average runner to have been picked off 10 times. That makes Lom about +2.5 runs. Lombardi was a very cautious baserunner, which, despite his incredible slowness meant he didn’t get thrown out very often. He was +6 runs against the league on that account. Despite his lack of foot speed, Lombardi did manage to take 44 bases in non-batted ball situations. That accounts for about 8 runs, where the league would have notched 9. So -1 runs here. On the whole, he’s sitting pretty close to level par with the league. That is, until we account for his taking extra bases on batted balls. Lom took the extra base ahead of the batter on singles and doubles about 32% of the time. The league took the extra base 47% of the time. In our figuring, that means that Lombardi’s legs “earned” -31 runs against the league. So on the whole, His Schnozziness nets out at -25 runs against average.

And then come the twinkillings. No one, not even Jim Ed Rice, banged into so many deuces as this guy on a per-plate appearance basis. He was the lifetime leader in the category for at least a couple decades, but the guy who passed him (someone named Aaron) had about twice the plate appearances. Which means that our estimate for Lombardi is a little more than -60 runs. Add it all up and he dumps about 9 wins of value and falls out of the running for the HoME.

          NOW  EST  NOW   EST  NOW  EST
Musial    135  137   98   100    1   1
Gehrig    113  113   88    88    3   3
Foxx      103  101   81    79    4   5
Mize       74   75   61    63    9   9
Greenberg  64   61   57    56   11  12
Terry      62   65   53    56   16  14
Camilli    44   48   43    46   27  25
Hodges     49   49   43    43   28  28
Bottomley  35   34   31    31   51  53

Bill Terry and Dolph Camilli are the stories here. Terry’s surge in value is primarily driven by excellence on the bases. For the seasons we know about, he was picked off only once, made about two-thirds the outs on base that an average player did, had more bases taken than average, and most important, he took the extra base on a hit 56% of the time, versus leagues around 50%.

Camilli, meantime, is an overlooked star. He appears to have been an above average baserunner, not just a meandering slugger, and he was excellent at avoiding the twinkilling (+16 runs career).

            NOW  EST  NOW   EST  NOW  EST
Gehringer    82   86   66    68    5    5
Frisch       83   83   65    64    6    6
J Robinson   65   66   59    60    8    8
Gordon       62   62   55    55   13   13
Herman       60   58   49    48   17   18
Doerr        56   57   47    47   19   19
Lazzeri      50   47   43    40   24   27
Frey         43   48   38    42   27   26
Stanky       40   42   38    39   28   28
Schoendienst 42   44   37    38   30   29
Bishop       42   42   36    37   33   32

There’s a few items of note here. Charlie Gehringer turns out to be an outstanding baserunner, not merely above average, pushing him upward. On the other hand, Tony Lazzeri turns out to be a poor baserunner and below average at DP-avoidance, driving him downward. Billy Herman’s pretty bad on the deuce too. But let’s pause for a moment and look at Lonny Frey.
Has anyone ever said to you, Hey, Lonny Frey was a damn good ballplayer? Well here’s the first time. Frey is little remembered these days, but as a shortstop and second baseman, he combined a fine glove, an above-average bat, strong baserunning skills, and a penchant for avoiding rally-snuffing double plays. Exactly the kind of player who play-by-play data reveals as a source of subtle value. We show him picking up about five WAR, which is 50 runs of value.

          NOW  EST  NOW   EST  NOW  EST
Elliott    52  53    42    43   19   20
Hack       51  56    41    45   23   19
Traynor    47  48    39    39   31   27
Clift      43  46    39    41   30   24
Kell       34  33    30    29   51   51
Lindstrom  27  28    27    27   65   64
Rolfe      25  29    24    28   73   61
P Martin   20  24    19    22   89   79

Because third base is a very clumpy position, small credits and debits can lead to significant movement on the totem pole. Harlond Clift, for example, surges up six slots with only three additional WAR in his pocket. He could run a little and was that rare bird, a righty hitter good at avoiding the double play.

Stan Hack parlayed an even bigger increase into a climb that leaves him this far from the HoME borderline. We reckoned him with 3 rBaser (versus -9 for BBREF) as well as 32 runs for DP avoidance. I suspect, however, that while the former of those could even inch up a little, the latter is not terribly accurate. That’s because Hack was a leadoff man for nearly all his career, and had a minimum of 1350 fewer opportunities than an average hitter would.

But most interesting of all are Red Rolfe and Pepper Martin. These guys were terrors on the bases. Rolfe, who had about half a career, was worth twenty-odd runs on the bases and another passel in DP avoidance. Red was merely above average in stealing, outs on base, and bases taken. But like Bill Terry, he took extra bases like candy: 57% extra-base-taken average versus a 48% league average, worth 15 runs. Then there’s Pepper Martin, who was hung with the famous sobriquet, “The Wild Horse of the Osage.” Like Rolfe, he had about a half a career, and like Rolfe, he ran wild. He was nearly +15 runs stealing bases, +10 on extra bases taken, and another +2.5 on bases taken for good measure. He took the extra base 63% of the time in a league with a 48% extra-base-taken rate.

           NOW EST  NOW   EST  NOW  EST
Vaughan     80  84   67    71    4    2
Appling     82  87   64    66    7    5
Cronin      73  72   60    59   11   11
Boudreaux   67  68   59    60   12   12
Reese       67  69   53    55   16   16
Sewell      58  62   48    52   21   19
Stephens    49  48   42    42   27   27
Maranville  45  44   40    39   30   33
Bartell     47  47   39    39   31   31
T Jackson   41  44   38    40   37   30
Rizzuto     41  41   37    38   38   37

Arky Vaughan slides into the #2 spot at shortstop. He was in a big bunch with Cal Ripken, well behind Honus Wagner. Vaughan isn’t as bad a baserunner as his poor stolen base rates suggest, nor as bad as his BBREF estimate. As a lefty with at least some speed, he turns out to be very good at avoiding double plays. Meanwhile, Joe Sewell, whom I elected with a lot of trepidation, improves his lot and gives me a little piece of mind. Sewell’s an interesting one. As a lefty he gets some double-play avoidance credit, but it’s really his baserunning that pushes him upward. You might be surprised by that since his SB% career-wise isn’t quite 51%, but the league back then ran at around a 55% clip, so it’s not nearly the eyesore it appears. Even so, it’s everything else he does on the bases that helps him. We have just four of Sewell’s seasons, but they account for more than 2,000 plate appearances, enough of a sample to get a good sense of his exploits. Sewell was never picked off in those four years. He’s a run better than the league in both outs on base and bases taken. Given his below average steals value, he’s just above par with the league before we get to extra bases taken. Joe took the extra base about 60% of the time, while the league managed just 50% of the time, good for about +6 runs. So he ends up with about 8 runs of running value for 1930–1933. BBREF gives him -2 runs. When we use the comps method to retrocast him, we end up with a little more than 30 runs total for his career.

I would sound this cautionary note about Rabbit Maranville. I feel very tentative about him. While we have several years of data on him, they come from his age 38–43 seasons. Rabbit missed one of those seasons entirely due to a broken leg, and came back for just 23 games after it. But a deeper look into his stolen base numbers shows a different story. As a young player, Maranville stole with some frequency, gaining double digits in steals every year through 1924 (except for a year lost to World War I). His success rates during those seasons for which we have his caught-stealing information (61%) are probably a little better than average for the time. Then Rabbit started to get old. He lost some time due to injury and ineffectiveness in the mid-1920s, appearing to lose a step in the process. So it’s difficult to say with certainty that the data we have is strongly representative. But for now, it works.

            NOW  EST  NOW   EST  NOW  EST
T Williams  129  129   98    98    2    2
Goslin       70   73   57    59   10    9
A Simmons    71  75    57    61   11    8
B Johnson    62  61    50    49   20   20
Medwick      55  52    47    45   23   25
Kiner        49  48    46    45   25   26
Minoso       51  52    46    46   26   23
Keller       47  46    43    44   27   27
Galan        43  47    37    41   40   38
Manush       39  42    34    37   41   41
Hafey        28  32    27    30   57   53

I didn’t know that Al Simmons was an excellent base runner, but that’s what our PBP data suggests. He was good at every facet of running, whether avoiding outs or taking bases.

This exercise appears to have vindicated certain decisions we made late in our electoral process. We knew that Joe Medwick had issues with double plays, and so we placed him behind Jose Cruz and Roy White in our pecking order. We felt unsure about Ralph Kiner as a fairly extreme peak case. Finally, because we’ve elected solely on Major League play, we didn’t extend any special dispensation to Minnie Minoso. Well the jury is in. Medwick’s double-play addiction cost him about 18 runs versus his leagues. Also, his arm appears less effective than DRA suggests. It’s all enough to push his value low enough that he sinks below Joe Kelley and Minnie Minoso in the rankings and essentially out of sight. Minoso only cashes in outfield arm credit here because his career started after the advent of PBP-based rDP and rBaser. He’s not a good thrower, but he picks up a couple-three runs against DRA, which helps. The fact that he didn’t lose value really helps because if we ever choose to pursue the Negro League angle, he’s so close to the finish line now that even just a couple seasons of above-average play could put him over.

           NOW  EST  NOW   EST  NOW  EST
Mays       162  161  114   113    2    2
J DiMaggio  81   84   66    69    6    5
Ashburn     74   74   60    60    8    8
Snider      59   59   51    51   12   12
Berger      47   49   43    45   26   23
Doby        49   49   43    44   29   26
Averill     45   46   40    41   36   33
Combs       43   47   38    41   41   31
D DiMaggio  40   42   37    38   46   41
B Chapman   40   38   34    32   52   56
H Wilson    35   33   33    32   53   58
L Waner     23   25   22    23   73   73

Tommy McCarthy is the worst player elected to the Hall of Fame. I’m far less sure now about the second worst. Is it Lloyd Waner or Highpockets Kelly? We can’t say yet with as much certainty as we’d like because we don’t have PBP info for enough of Kelly’s career to say. But right now, I’m leaning toward Little Poison. I would be a little skeptical that Combs’ is gaining that much ground. He’s definitely gaining because his baserunning is much better than BBREF estimates it, probably by 20 runs. But like Stan Hack, Combs is a lefty lead-off man, and so his estimated rDP of +13 is probably too high. It’s a shame that Averill didn’t reach the majors until his age-27 season and that Dom DiMaggio had the heart of his carved out by the war. Both are coulda-been HoMErs. Which brings us to Larry Doby. Like Minoso above, Doby has inched just a little closer toward the borderline, and he may have enough in reserve during his Negro League seasons to creep over the line, should we choose to go down that path.

           NOW  EST  NOW   EST  NOW  EST
Ruth       181  181  128   128    1    1
Ott        111  117   80    84    3    3
P Waner     80   82   63    64    7    7
Slaughter   60   62   48    50   26   26
Nicholson   49   51   44    46   27   27
Cuyler      50   51   43    44   30   28
Klein       43   45   41    42   35   32
Holmes      42   43   40    41   37   34
D Walker    44   46   37    40   43   37
Furillo     42   42   36    36   49   49

Mel Ott’s a pretty great player. I never really stopped to think about him much. But now I suspect he’s kind of the Frank Robinson of his time. Mantle, Mays, and Aaron dominated the 1950s and 1960s. Robinson was just a notch below. He sometimes outperformed them, but on the whole, the other guys were just better enough that over time a gap in value developed, as well as one of perception. Similarly, Ruth, Gehrig, and Foxx dominated the baseball scene of 1920s and 1930s. Mel Ott, like Robinson would later, did his thing year in and out and wasn’t quite as exciting or sometimes as valuable as his competition. Like Robinson, he also had a diverse set of skills with sneaky speed and great power plus durability and longevity. Certainly Ott was not overlooked, just as Robinson wasn’t, but he never quite equaled those other guys. Through this process, I discovered that Ott was probably a lot better baserunner than you’d think and that his attempts to pull balls down the rightfield line into the Polo Grounds short porch probably kept him out of the double play so much that he excels in that category of our analysis. Meantime, Slaughter is now neck and neck with Vlad Guerrero, and if the actual BBREF data comes through and looks better than these estimates, Country might pass the Impaler. They are both right on the line in right field, and Slaughter’s advantage may be his era. The post-war era is light on honorees. Finally, KiKi Cuyler. After Sam Rice, he’s a big reason why we needed to do this project. We noticed that he was one of his era’s speed merchants, and we knew that he was reputed to have a good arm. All of which turned out to be true, but unless his real numbers are a lot better than what we’ve seen, he’s not going to creep upward.

Overall, the differences we’ve noted are not earth-shattering. Mostly they don’t suggest that we’ve missed players or elected fellows we shouldn’t have. But it does give us a greater sense of the likely value still out there to discover. Of course, once BBREF calculates these figures and creates formal estimates, our numbers will be wiped away—as they should be. Those guys know more than we do, and we trust them. For now we have these estimates to guide further decision making.



6 thoughts on “Giving 1930s and 1940s players back their missing value mojo

  1. Twice you mentioned if you go down the road for Negro League players. Please do.

    Posted by verdun2 | March 22, 2017, 8:55 am
  2. Great stuff! Any chance for some info on why Joe DiMaggio, Goslin, & Appling moved up?

    Posted by layson | March 22, 2017, 1:44 pm
    • For each guy the story is a tad different.
      DiMaggio: Great baserunner and worth a lot more than BBREF’s regression-based estimates give him. Once on base, he had great baserunning aptitude and rarely made mistakes. In this way, he’s much like Mickey Mantle who has been revealed to be an outstanding baserunner through PBP stats. Joe D also benefits from an excellent arm. He was truly an elite all-around player.

      Appling: Like many in the 1930s, Appling’s meh steals totals don’t tell the whole story. He was also a smart baserunner who didn’t make a ton of mistakes. Also remember that in this era, someone with a 60% or 65% stolen base rate was adding value compared to the league and its 55% to 60% rates.

      Goslin is a combo of all these things plus one. He turns out to be a close to or slightly better than BBREF’s baserunning estimate. More importantly, as a lefty hitter, he gets a nice DP-avoidance boost. Just being a lefty is probably something like a 70% guarantee that you’ll have average or better DP avoidance value. The faster you are, the more likely that guarantee increases. If you also hit a lot of flyballs, it’s higher yet. I suspect that Goslin was something of flyball hitter. He hit plenty of doubles and triples in Griffith Park which ate cookies like Alistair Cookie, so he probably kept it off the ground. That’s a guess on my part. But Mel Ott is another example of this. He turns out to be an above average baserunner, so he had a little speed. He’s a lefty. And his home run splits are something like 2:1 Polo Grounds/elsewhere. Ott was known for trying to pull the ball down the line where the pole was under 300 feet away. Think of Fenway’s Pesky Pole, only closer. With Ott trying to hit flyballs there, he probably kept himself out of a lot of DPs. Back to the Goose, however, his arm was probably not an asset in the way that Jesse Barfield’s was. He naturally gravitated toward RF for a reason. But his arm appears to be a little better than the average left fielder, and that’s what we’re judging against. Unlike, say, Lou Brock, one of the worst throwers ever to do a left field glove (nearly -50 rOF!).

      In general, however, if a guy is moving up it’s because his baserunning is likely better than the likely overfit BBREF estimates see, that he’s got some serious DP value (which is not calculated before 1948), and his arm is decent. Let me clarify or restate some things about though.
      a) I don’t blame BBREF at all for the rBaser estimates they’ve made. They are doing their best with a very limited dataset and providing a baseline guess. It’s not my intention to rag on them, only to say that I think their calculation is limited and can’t capture all that we now know.

      b) My rDP estimates suffer in some ways from what I’ve just said about BBREF’s running estimates. We don’t have much in the way of DPopp information throughout the 1930s. So I used PAs as my denominator. This creates an issue for leadoff hitters. Those guys are probably getting too much credit from me. Obviously, they will always have something like 20% to 25% fewer DPopps just due to leading off a game. But also they hit behind the two batters in the lineup least likely to get on base. So take that into account for the likes of Stan Hack. It’s also possible, however, that someone like Mel Ott or Goslin isn’t credited with enough rDP because they generally hit 3rd or 4th, behind the best on-basers on their teams.

      c) And, generally, we’ve made an assumption in these two articles and the one coming on Monday that the 1920s, 1930s, and 1940s were enough like the 1950s-2010s that comping is an effective tool for our purpose. I think that’s not too weird to think, but playing conditions were different, and we can’t take anything as gospel.

      Posted by eric | March 23, 2017, 11:31 am
  3. Thanks for all the elaborations. Really appreciate it. As a Cuyler fan (he’s in my personal HOF) I’m glad he moved up through this process, only wish it had been a little more.

    Btw, I’m guessing if/when BBRef updates its baserunning data for the 1930’s & 40’s players it could take a lot of people by surprise due to this heretofore hidden value now being added to WAR.

    Posted by layson | March 23, 2017, 12:43 pm

Tell us what you think!

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Institutional History

%d bloggers like this: