Playing the A’s? Grow a Beard.

Not exactly the beard you’re thinking of. (via Bob P.B.)

Bring up beards and playoff baseball these days, and a typical fan is likely to think of last year’s World Series-winning Boston Red Sox and their impressive array of facial hair. However, there’s another, more strategically interesting usage of the term that’s been under the radar a long time, one with a golden opportunity to reappear this fall.

A “beard,” a term introduced to the baseball world by Bill James in his Guide to Baseball Managers, refers to a seldom-used managerial maneuver that involves pulling the starting pitcher extremely early in the game for a starter that throws from the other side. If the opposing manager has set his lineup with the expectation of facing a lefty and the replacement pitcher is a righty, this forces the opposing manager to cede the platoon edge for several trips through the order or to burn his bench early in the game. Given that it relies on lefty/righty splits, it’s particularly effective against teams that do a lot of platooning.

While the stratagem is quite uncommon, its history is rather long. Beards go back to at least 1924, when Washington Senators’ manager Bucky Harris pulled it in Game 7 of the World Series, starting the relatively no-name right-hander Curly Ogden and pulling him after just two batters. The maneuver eventually led Giants’ manager John McGraw to pinch hit for Bill Terry in the 6th inning rather than keep letting him face left-handed pitching. (In retrospect, using the beard here doesn’t seem to make much sense, as the Giants only started one other lefty besides Terry, so Harris ceded the platoon advantage to three quarters of the lineup. Harris was probably reacting to Terry’s 1.517 OPS in the prior games of the series and his -5 OPS+ against lefties during the regular season.)

Besides the Curly Ogden game, though, clear-cut examples of beards are hard to come by. For this to occur, there needs to be a confluence of factors: a team in a high leverage game with an adventurous manager and mediocre starting pitching that happens to be playing an opponent that platoons a lot. For the curious, here are a few other examples of games where managers deployed the beard:

• Joe Maddon started reliever Jamey Wright in a game against the A’s on September 1st of last year. The results were mixed, as Wright gave up 5 base runners in 1.2 innings, though he and replacement Alex Torres combined for 5 innings of 2 run ball in a 5-1 Rays loss. The A’s did not pinch hit for their lefties when Torres entered.
• As James mentions in his piece on beards, Jim Leyland started reliever Ted Power in Game 6 of the 1990 NLCS, then switched to lefty Zane Smith in the middle of the third. This eventually prompted the Reds to pinch hit for lefty Paul O’Neill, who had been killing the Pirates that series (1.333 OPS in the first five games); unfortunately, pinch hitter Luis Quinones tagged Smith for the eventual winning run. Still, the tactic was largely a success, as Power and Smith combined for two runs in 6.1 IP.
• Dick Howser pulled Bret Saberhagen after three innings in Game 7 of the 1985 ALCS in favor of Charlie Leibrandt, and this is commonly cited by writers (including Bill James) as a beard tactic. It prompted Toronto to pinch hit for Al Oliver, a lefty who had driven in the winning run twice against Royals’ relief ace Dan Quisenberry thus far that series. While it did have the effect of a beard, it’s unlikely that it was the original intention. Not only does it seem unlikely that a manager would pull this trick with that year’s Cy Young award winner, contemporary accounts note that Saberhagen left due to injury, not for strategic reasons.
• A September 1973 matchup between the Mets and Cardinals, two rivals fighting for the NL East title, saw lefty Mike Thompson, the Cardinals’ starter, replaced for ineffectiveness after facing just two batters. The reliever, righty Rich Folkers, noted after the game that “it was no trick to outmaneuver the Mets. Nobody told me before the game that I was a starting pitcher.” While Folkers pitched acceptably, he was pulled after just seven outs and the Cards lost 5-2.
• According to Bob Mack’s book Bird Hunting in Brooklyn, the Dodgers’ Leo Durocher tried to pull a beard against the Cardinals in September 1946, starting the right-handed Ralph Branca against a Cardinals lineup featuring five lefties. The plan was to pull Branca for lefty Vic Lombardi at an opportune time; however, Branca pitched a three hit shutout, so an opportune time never arose.
• Until a couple of years ago, teams in the Central League in Japan did not announce probable pitchers until lineup cards were exchanged, meaning that it was common to disguise the starting pitcher, a tactic similar to the beard in effect if not identical in execution. (Even now, the Central League doesn’t require announcing starting pitchers until game time during the playoffs.)

All told, according to Retrosheet data, there have been roughly 500 games since 1960 where the starter was pulled for an opposite handed reliever before the 5th inning and with no runs in; however, after investigating the few dozen such games that occurred in September and October featuring a beard pulled by a team in a pennant race, I was unable to find any other examples to supplement those above. (In each case, I could either find a documented injury or there were circumstantial clues suggesting that the beard wasn’t planned in advance.)

Though the beards list above were not an unqualified successes, it still seems like a tactic that should work. It’s somewhat similar to the idea floated in the past to use mostly relievers in high leverage games, but carrying the additional advantage of not requiring a manager to go to his bullpen early if the second pitcher is a starter who can pitch for longer than a reliever.

There are some legitimate concerns about keeping this secret from the other team in the age of Twitter, but, as James points out, it’s not clear how the other team can thwart the beard if it’s implemented appropriately. If the lefty and righty are pretty similar quality pitchers (borrowing James’s example, you’re not skipping Greg Maddux’s turn in the rotation in favor of someone “with an ERA about Cecil Fielder’s belt size”), then knowing what’s coming won’t help the other team. This is because the ways around the beard are to use the left-handed lineup against a left-hander (or righties against a righty), which concedes the issue before the game has even started, or to pinch hit to keep the platoon advantage, which is an imperfect solution and hurts the team in the late innings.

James thus concludes, and I agree, that the reason the beard isn’t used more often can’t have to do with how effective it is; James concludes that it makes the managers’ jobs harder, so they avoid doing it. I’d imagine that it’s also avoided because it could be seen as disrespectful to one of the pitchers. (Those two reasons are not too different from the reasoning many used to avoid employing more defensive shifts.) Of course, one of the perks of being a baseball outsider is that I don’t have to worry about such issues. With that in mind, I decided it would be interesting to estimate how effective this tactic can be.

Thankfully, there’s a very obvious candidate to try the beard against this year: the Oakland A’s, who not only platoon at a large number of positions, but are also likely to end up in the winner-take-all wild card game (98 percent, according to FanGraphs’ playoff odds). I decided to try to use simulation to roughly estimate how much of a benefit their possible opponents could get from bearding. While there’s a lot of uncertainty in the results, the initial findings suggest the gains are substantial.

A few ground rules that I established for the simulation in order to make things simpler: the A’s are be the home team and start Jeff Samardzija. I further assumed that the opposing team will be one of the Mariners, Tigers, and Royals and will only employ a beard with two pitchers of otherwise reasonably similar ability, and eventually decided to test two pairs: Danny Duffy and Jeremy Guthrie for the Royals and David Price and Max Scherzer for the Tigers. (No pairs for the Mariners stood out as good candidates for this.)

The actual simulation process was a bit convoluted, but I will try to explain it as simply as possible. For simplicity’s sake, I decided to use Matt Hunter’s SaberSim tool, which is free and very easy to use. However, it has two main drawbacks for purposes of this analysis: it doesn’t include platoon splits, and it doesn’t update its projections (it uses projections from the beginning of the season).

To correct for those issues, I used the methods described in this THT article by Bojan Koprivica and this FanGraphs piece by Matt Klaassen (both borrowing from The Book) to compute the expected wOBAs against left- and right-handed pitching for each member of a typical A’s left-handed lineup (i.e. the lineup that faces right-handed pitching) based on their historical splits and their current Steamer wOBA projection. (All my data are as of September 7 and come from Fangraphs.) I then found players whose overall preseason projections were similar to those platoon wOBAs and with similar power and plate discipline numbers and substituted those for the A’s lineup in the simulation. This is a quick and dirty way of replicating the effect of the platoon disadvantage in the simulator.

As an example, my calculations suggest that Adam Dunn’s wOBA projection against lefties is approximately .290, while it’s .324 against righties. Conveniently, Steamer’s preseason projection for Dunn was .324, so we use the simulator’s version of him as Dunn facing righties, while we use Tyler Flowers for Dunn against lefties.

I didn’t make the platoon adjustment for pitchers, tacitly assuming that the pitchers have league average platoon splits. (Obviously, depending on the pitchers involved, this might be an important omission.) I did, however, do the same current-to-preseason projection adjustment, so I used a pitcher that had a similar simulator FIP—Madison Bumgarner for David Price, for instance.

After making all of the above adjustments for both teams, I simulated 30,000 games for each of the two opponents. 10,000 were with the lefty starter against the A’s anti-lefty lineup, and 10,000 were with the righty starter against that same lineup. The final 10,000 were using the right-handed starter but with the A’s numbers adjusted as if they were pitching against a left-hander, to control for the fact that the lefties used in this test are a bit better pitchers than the righties. Changing the starter each time is equivalent to assuming that the original starter faces 0 batters, which isn’t quite how things would play out in real life (the starter has to face at least the leadoff man) but is pretty close to the execution of the original Curly Ogden beard.

Before I present the results, I want to stress that these should be taken with a pretty hefty amount of salt. The simulator doesn’t include defense, pinch hitting, relief pitching, or a number of other things that would be affect the results of a real game. Moreover, the ad hoc way I generated projections is going to have its own error, both numerical and human. Thus, these numbers are nothing more than a starting point for how effective beards can be.

If these winning percentages seem wrong, you’re right to think so; not including defense and the bullpen in the simulation really helps the team with an awful bullpen and a bottom 5 defense and really hurts the team with lights out relievers and a great defense. We care less about the exact percentages, though, and more about the differences between the sets of sims, which let us estimate how useful this strategy really is.

As it turns out, it seems extremely useful; the Royals see a 23 percent drop in RA9 (14 percent after controlling for pitcher quality), which is quite important if you’re hoping to get 5+ innings out of the starter. The Tigers, with much stronger pitchers, still get a 16 percent drop in RA9 (13 percent after the pitcher quality adjustment). Even if the simulator exaggerates the effects on starter RA9 by a large factor, it seems obvious that teams should be considering the beard in high-stakes games against the A’s when they have the appropriate pitchers available.

The A’s can respond, of course, but they can’t pitch hit for everyone, and every substitution cuts into their defensive flexibility and leaves them open to disadvantageous bullpen matchups. Remembering that the beard has been employed in the past specifically to get one good player out of the lineup, I imagine Ned Yost and Brad Ausmus would be thrilled to see the A’s making wholesale changes early in the game.

In the end, it seems pretty clear to me that this is something teams that play the A’s (or any other heavily platooning team) should consider for games during the pennant race or playoffs. While I’d be (pleasantly) surprised if someone uses a beard this year, I do think there’s a distinct possibility that this eventually becomes a part of standard October strategy as teams start to embrace previously frowned-upon tactics. Maybe in a few years, the playoffs will be just like they were in 2013, with a team employing beards to capture the attention of baseball fans everywhere.