Last week I told you that we were still working on understanding the field of great GMs. Well, that’s still true, but today we can at least offer a little insight about good GMs versus poor ones. While we’re at it, we’ll introduce some ways to evaluate a GM statistically at a glance.
So far, we’ve logged the careers of five GMs: one great one (Branch Rickey), two strong candidates (John Quinn and Al Campanis), and two poor GMs (Phil Seghi and Randy Smith). As you’ll soon see, many differences are obvious, but each exec has his own strengths and weaknesses.
There are many dimensions we could assess for a GM. With the basic statistics and transaction records we can access on BBREF, these three are the most adaptable to our purposes. We emphasize that these are not the only ways to look at a GM. In isolation, they are not as helpful as a player’s record would be. But when we pair them with a fuller narrative of the GM’s career, they fill in a lot of gaps in the story. Onto those three dimensions!
This is the simplest, especially since we’ve already introduced the evaluation tools during our managers elections. In the tables below, you’ll see the following:
- W/L : Won-Loss record while GM was in office
- PCT: Winning percentage
- Vs EXP: As we explained during managerfest, this is an adaptation of the expected wins formula Bill James introduced in his managers book. Except we use pythenpat records instead of actual records to calculate it.
- OCT: Postseason apperances (starting in 1969).
- OCT v EXP: Measures postseason appearances against the basic probability of any random team making them.
- WS: World Series appearances
- WS v EXP: Similar to OCT v EXP
- Titles: Championships won
- Titles v EXP: Ditto
- MGR PYTH: This is the team’s variance again its Pythagorean record as a measure of how much value the GM’s manager brought to the team.
So let’s look at the careers of the five guys we have info on so far.
TEAM PERFORMANCE VS VS VS VS MGR YEAR W/L PCT EXP OCT EXP WS EXP TITLE EXP PYTH ==================================================================== SEGHI 883-989 .472 -11 0 -2.0 0 -0.9 0 -0.5 - 4 SMITH 566-776 .422 -49 0 -2.0 0 -0.6 0 -0.3 -38 QUINN 2147-2126 .502 +17 0 -0.7 3 -0.3 1 -0.5 - 9 CAMPANIS 1576-1280 .552 +44 6 +2.9 4 +2.5 1 +0.2 + 8 RICKEY 3265-3015 .520 +87 --- --- 8 +2.7 4 +1.5 +46
Obviously, bad GMs have worse teams, and they don’t last nearly as long as good GMs. We also see that good GMs reach the big dance, and the great ones reach it a lot. Good and great GMs beat expectations, often frequently. Part of being a bad GM is routine disappointment. Good GMs may also tend to hire better managers. This isn’t as unambiguous as the other items, but Randy Smith probably has the worst winning percentage of any long-time GM, and he hired Buddy Bell who himself is the worst long-time manager since integration. A match made in…Detroit. On the other hand, Branch Rickey hired Bill McKechnie and Billy Southworth and stuck with Leo Durocher, sharpies all.
OK, but what about the GM, himself. How’d he do? The typical exec is hired because either the sitting GM retires or the sitting GM lost his seat. In the former case, the mandate from ownership may be to keep the good work moving or to take the next step up in competitiveness. In the latter case, it’s either a big shakeup because the front office and the team are in disarray, or ownership want to go all-in on winning and don’t think the guy in the hot seat can take the heat. I didn’t at first think that I could find a common way to measure the effectiveness of execs in all these scenarios, but we might have stumbled upon something that works.
A GM’s job is ultimately to put together a competitive team. The more talent the better. When, like Al Campanis, he inherits a team with excellent players around the diamond, he has less building to do, but he may not have as much budget to work with because those good players he inherited have to get paid. For someone like Joe Quinn, who tore down the Braves in the 1940s and built a World Series team, he had more up front work to do but also more latitude to do it in. The only thing these guys have in common is that in order to compete, their teams have to get to a record of at least .550 (or 89 wins in 162-game notation). That’s the flex point where teams are realistic competitors and where any additional wins they can cadge become massively important. And that’s where we start.
We start by determining how much of a team’s total WAR belongs to players who were on-hand before he got there. In the chart below, we’ll call this the BASE. Then we look at the value contributed by players the GM brought in. Next we’ll determine how many WAR the GM needed to have acquired to reach a .550 record. The calculation for this is straightforward. Using 162-game notation, we know that a replacement-level team would win 49 games. Which means a team needs 40 WAR. From that 40 WAR, we subtract the BASE total to get the GOAL. Finally, we divide the GM’s total by the GOAL to see what %GOAL he arrived at. Note that we aren’t saying the GM could have known what his GOAL would be, only that given whatever circumstances the season brought he either pulled the right levers or didn’t. [Ed’s note: please ignore any rounding errors.]
GM PERFORMANCE YEAR BASE GM GOAL %GOAL =============================== SEGHI 115 239 350 68% SMITH 97 114 247 46% QUINN 222 729 824 92% CAMPANIS 342 407 364 112% RICKEY 388 871 1132 77%
Again, we see how much better the stronger GMs are at obtaining the talent their teams need. Branch Rickey is an especially interesting case. There’s a bit of Connie Mack to him in that some early and late teams really drag him down overall. Also, World War II was especially hard on the Dodgers. Here’s the breakout by team:
Rickey by team TEAM BASE GM GOAL %GOAL =============================== TOTAL 388 871 1132 77% SLB 60 18 54 33% STL 150 722 801 90% BRO 169 126 137 92% PIT 49 12 141 9%
The early Brownies and late Bucco years kill him. The war hurts his Brooklyn days, and he was still learning the role in his early Cardinals days as well. In other words, we need a bit more narrative context to fully comprehend what we see as we look at GMs.
John Quinn is a Jeckyl and Hyde. He was brilliant in Boston/Milwaukee, was pretty good for much of his Phillies tenure, and then faded out long and slow.
JOHN QUINN TEAM BASE GM GOAL %GOAL ================================ TOTAL 222 729 824 92% BSN/MIL 157 421 375 112% PHI 65 308 449 69%
Finally, one more dimension to look at, and that’s the GM’s actual transactions. Here we can see things like how much total value he brought to franchise, even past his tenure, whether he was a trader or a scoutmaster. Was he good at finding treasures in other GMs’ garbage? There are many, many types of transactions, and not all of them are well documented before the 1960s. So I’ll just show you the most commonly known ones and the totals (so the totals won’t reflect everything shown). First we’ll look at the kinds of transactions, then the value wrought from them.
- AM FA: Amateur free agent
- PUR: Purchased from another pro team
- FA: Free agent
- AM DF: Amateur draft (any time of year, only players who signed with the team and played in MLB)
- R5 DF: Rule 5 Draft
- ML DF: Minor League Draft
- TR: Trade
- WV: Waivers
- SLD: Players sold to other teams
- REL: Players released
TRANSACTION SUMMARY INBOUND | OUTBOUND AM AM R5 ML | R5 MI FA PUR FA DF DF DF TR WV TOT | SLD REL DF DF TR WV TOT =================================================|================================ NUMBER OF TRANSACTIONS | SEGHI 10 15 33 38 4 1 106 5 212 | 12 39 2 0 106 1 165 SMITH 11 8 103 34 9 2 65 22 254 | 8 55 5 3 65 15 151 QUINN 137 95 33 25 16 12 113 8 441 | 8 55 25 4 113 3 231 CAMPANIS 40 12 38 78 6 2 69 2 343 | 19 63 12 4 69 7 180 RICKEY 174 78 49 -- 24 11 108 24 748 | 111 68 36 8 108 34 388
Part of what you see here is the difference between MLB before and after the draft. Before it, Rickey and Quinn beat the bushes like crazy and came up with great players. They also used every player-acquisition channel open to them pretty extensively. Al Campanis did the same. Seventy-eight of the players he drafted and signed played in MLB, but he also signed a lot of amateurs for his time. The Dodgers of Campanis’ era eschewed the free agent market in deference to their juggernaut farm system, driving as much dough as they could into player development. Campanis also didn’t make a ton of trades (thanks to the farm system), then again, Swapper Phil Seghi got that nickname because he wasn’t especially judicious about trading. But Campanis used all modes of acquisition. By comparison, Randy Smith was a poor drafter and Seghi perhaps worse. Smith signed a bucketload of free agents to little avail, and he traded a lot with minimal results as well see now:
TRANSACTION SUMMARY INBOUND | OUTBOUND AM AM R5 ML | R5 MI FA PUR FA DF DF DF TR WV TOT | SLD REL DF DF TR WV TOT ====================================================|================================ WAR FROM TRANSACTIONS | SEGHI -4 13 4 32 2 2 275 -1 323 | 2 3 16 -2 253 0 272 SMITH 3 -2 27 36 -2 3 140 10 215 | 2 1 0 0 141 0 144 QUINN 443 45 182 141 25 3 480 1 1320 | 2 1 3 78 496 24 600 CAMPANIS 48 40 2 171 0 1 257 0 519 | 38 13 4 14 298 -1 402 RICKEY 805 118 268 --- 121 1 261 0 1947 | 337 16 64 -3 574 8 1030
Again we see a difference in epochs. For Quinn and Rickey, who worked before free agency, we calculate the value outbound players brought to their subsequent teams forever, or until they came back to the originating team. This is because before free agency, an exec knew he was giving up a player in perpetuity. Today with free agency, GMs are trading for defined terms. Four months of Randy Johnson or however long the contract lasts. That limits the upside and the downside for contemporary GMs. As we look at this table, it’s clear what a poor judge of talent Randy Smith turned out to be. Although he certainly wasn’t always awful, in no kind of transaction was he outstanding. Compare to Campanis who milked 117 more WAR from just four more trades. For that matter, Swapper Phil scored only 18 more WAR than Campanis on 50% more trades. It’s fair to say however that in both these cases, Smith and Seghi beat the pants off Al in terms of what they gave up in value in the trades they made. But as we noted earlier, these are Smith and Seghi’s only value channels, while Campanis spreads it around a bit more.
We should note here that these WAR estimates do not include the WAR value of whatever cash changed hands. And in Rickey’s case a whole lotta deals involved cash.
There are yet more avenues of inquiry available to us, especially one where we look at the legacy a GM left his team. But for now, these three lenses help us see the basics and make our start. We begin to get some objective ways to see exactly how bad GMs fail. At the top level, they fail to meet expectations and may hire poorer managers. In their own performance, they simply can’t find traction, can’t build on the team’s talent base, and ultimately struggle to assess and acquire valuable players. Finally, they don’t find value in all the right places. They may have a strength or a preferred way to buy talent, but in isolation it’s not enough to overcome their many acquisition weaknesses.
Stay tuned. More GMs on the way, they just take a while.
[Edited 3/26/16: Al Campanis’ WAR IN and WAR OUT totals were incorrect. They are now correct.]