Bullpen management: notes from the CD-ROM

[MARCPELLETIER]

I bought the CD-ROM last month. Lots of fun, especially when TSN is down.

I made a little experiment, nothing scientific. I took a team (basically this team, http://fantasygames.sportingnews.com/stratomatic/team/team_other.html?user_id=212143 , but without Hoffman, whom I acquired recently), and ran several sim seasons.

This team, before I acquired Hoffman, had 5 solid starters, but with many with short stamina (S5 or S6), stellar Nathan in the bullpen, and a butch of 1M relievers. Plays at home in Fenway, and even plays more games at low-hitting ballparks on the road, which makes Nathan even better.

First, I tried 5 seasons with Nathan as a 8th-9th inning set-up man/closer (all SPs at slow hook, Nathan as slow hook, don't come before the 8th inning). During those seasons, Nathan had approximately 50 innings. My team on average finished with 84 wins.

Then I put Nathan as a full-time set-up man/closer (slow hook, don't come before the 6th inning). My closers were 0.51M Gagne and 0.70M kobayashi. Nathan's use jumped to around 150 innings. My team on average (5 seasons) won 88-89 games. This shouldn't come as a surprise for experienced strat players, as the community knows that restricting stellar relievers to closing situations is a costly decision.

But things are more interesting than this: during 4 of the 5 seasons, my team had a losing records in extra-time games and in one-inning game. On ALL 5 seasons, my team underperformed compared to the pythagorian projection.

Then I added Hoffman on my roster (to keep the same payroll, I made some other changes on my offensive roster). I kept Nathan with the same specifications (so he remained my set-up/closer vs both lhp/rhp), but I removed the "slow hook" mark on my 5 SPs. I restrained Hoffman to come after the 8th inning, and I specified HAL to use him in tight games, when Nathan was not available.

On average, my teams won 92-93 games. Nathan's use increased slightly to around 170 innings. Hoffman finished on average with 10 saves and 60-70 innings. Interestingly, after 5 season, my record perfectly matched the pythagorian projection. On 3 of the 5 seasons, I had winning records in extra-time, and I had a losing record in 1-game decisions only once.

Take note that Hoffman didn't pitch really nicely, which is not surprising given that he doesn't have a good card. I didn't calculate his mean era, but his era was often in the 4s-5s, with one season being exceptional (2.18). Nevertheless, he is a C6, and he is substantially better than under-1M-options (unless you go with specialists). The way I see it: my team didn't perform much better with Hoffman, from a pythagorian perspective. But I stopped losing tight game at the same frequency compared to before.

Conclusion:

-bullpen management can easily win you up to 10 games per season.
-using a stellar reliever more frequently than in closing situations pays big.
-teams with short bullpen are likely to lose many tight games, thereby underperforming when compared to their pythagorian projection. If, as a player, there is a trend to always underperform your pythagorian record, you better look at the way you manage your bullpen in tight games.

[MARCPELLETIER]

In case you wonder:
pythagorian projection is the record you SHOULD have considering the runs your offense scored and the runs your pitching has allowed. You can see it on the expanded screen of the standing sheet, at TSN.

Before I acquired Hoffman, my record in this TSN season for X-time and 1-game decision were respectively 1-2 and 4-6, and my team was playing 2 games below the "pythagorian projection", so I felt I was definitively repeating the same trend than the one observed in my simulation seasons.

[MARCPELLETIER]

BTW, I just came across a paper on the web where a guy made simulations based on 1000000 games ( :shock: ).

His results:

[quote:8eba85bf66]
In all cases, the stopper never pitches the tenth or subsequent innings (and always enter in the game at the beginning of an half-inning).

Case 1: Stopper pitches 9th inning if save situation (lead of 1, 2, or 3 runs).

Win Pct = .513 or 2.17 games better than .500 in a 162- game season; stopper pitches 42 innings in a typical 162-game season.

Case 2: Stopper pitches 9th inning if save situation or game is tied (lead of 0, 1, 2, or 3 runs). Win Pct = .521 or +3.39 games; stopper pitches 58 innings per season.

Case 3: Stopper pitches 8th and 9th innings if a save situation (lead of 1, 2, or 3 runs). Win Pct = .525 or +4.09 games; stopper pitches 88 innings per season.

Case 4: Stopper pitches 7th, 8th, and 9th innings (two innings max) if game is within one run (lead of �1, 0, or 1 run). Win Pct = .539 or +6.33 games; stopper pitches 142 innings per season. I will comment on this workload below.

[/quote:8eba85bf66]

This simulation (based on real-life data) pretty much ressembles what I ended up with using Strat. Taking your best reliever from a restricted Closer role to a "go-to-guy-in-close-game" produces roughly 4 additional wins in the simulations (I had roughly 5 additional wins in my own tests).

[Mean Dean]

[quote:014adb1230]Taking your best reliever from a restricted Closer role to a "go-to-guy-in-close-game" produces roughly 4 additional wins in the simulations (I had roughly 5 additional wins in my own tests). [/quote:014adb1230]It's amazing that the closer role has developed the way it has. I know that SOM managers have the advantage of not having to warm the "fireman" up first, but, yeesh.

[apolivka]

The strictly assigned "roles" that many managers love to cling to for their bullpens these day, in my opinion, cost teams quite a few wins. What really irks me is putting in the closer when being up by 3 in the 9th with the bottom of the order coming up. Pretty much anyone in the bullpen can probably "save" that game. Then the next game the score is tied in the 8th and you've got 2-3-4 coming up, but the closer never comes in!!! In many games the 7th or 8th could be much more important than the 9th, but the rigidly assigned bullpen roles can't be tampered with...

Come on managers. A manager's "role" is to win games, not to get saves for his closer. Drives me nuts. The save is one of the most overrated stats in baseball. Luckily in simulations, we don't have to worry about contracts with bonuses for saves.

[maligned]

Well put, apolivka.
It's difficult to quantify the human element, though. I'd love to see full season stats in save versus non-save situations for, let's say, the top 10 elite closers in the game. I feel I've often seen closers tank non-save situations because they can't mentally re-focus on a different situation than the one to which they are accustomed. Maybe these tanked situations just stand out in my mind; I wonder if anyone has seen data on this?

[MARCPELLETIER]

[quote:176ff11045] I wonder if anyone has seen data on this?[/quote:176ff11045]

yeah, it's been done. Closers get into washout games sometimes because they still need to pitch here and there to remain in shape. And from I remember, their performance is just the same. Perhaps they lack concentration when leading 11-2. but so do the hitters!!!!

Until officials incorporate stats like leverage, or Win share, it's gonna be hard for coaches to apply what apolivka says. Get into the game to face #2-#3-#4 without finishing the game is golden way to finish your career with the most blown saves, and lowers your number of saves. Be the career leader in blown saves, even though your the best reliever of your generation, is not an easy way to get into the hall of fame, ask Goose Gossage.

[maligned]

I did a few minutes of research.
Let's assume our discussion above pertains to prolific closers only (guys who are used almost exclusively as closers and are highly successful). Obviously, we don't care if a guy with an ERA of 4.50 gets into more games or not.

Top 10 current saves leaders with an ERA of under 3.00:
Rivera, Nathan, Hoffman, Aardsma, Papelbon, Street, Franklin, Rodriguez, Broxton, Bell

Collective Innings Pitched: 445
Average Innings Pitched: 44.5 per pitcher
Collective ERA: 1.98
ERA in Save Opportunities: 1.84 (two thirds of IP came here)
ERA in Non-Save Opportunities: 2.25

Stats of the 10 relief pitchers with the most innings pitched in relief from these same 10 teams:

Collective Innings Pitched: 517
Average Innings Pitched: 51.7 per pitcher
Collective ERA: 3.08

Difference in IP between top closer and innings eater: 7.2 IP
Difference in ERA between top closer and innings eater: 1.10 ER/9 IP

Analysis: 0.88 runs lost for the season by not having the top closer take on an innings eater role (this becomes only 0.66 runs if you consider the difference between non-save opp. ERA and innings eater ERA)

Conclusion: The issue is more that NO relief pitcher is pitching many innings these days--not that the closer isn't pitching enough. These teams haven't even lost an entire run on the season by not having their prolific closer pitch an "innings eater" amount.

[Mean Dean]

Hmm? I don't think we're talking about total IP, we're talking about [i:4c2a0bbccd]which[/i:4c2a0bbccd] innings they are. Ideally, the best reliever would be pitching all of his innings (however many that turns out to be) in the scenarios when the game's outcome is most at stake. This is impossible to do perfectly, because 1) a game's progress is unpredictable, so you never quite know when the most important moment will be, and 2) in real life, you have to warm up the pitcher for a couple of batters before he comes in... so, you have to be able to anticipate that the key situation is coming up before it actually happens... and you also can't warm the pitcher up too many times without bringing him in. So you can't always have the best guy in the game when you need him most, but with the modern closer model, you're not really even trying.

[maligned]

Even that point isn't as bad as we think.

Collectively, 80% of the innings pitched by these 10 were in late inning pressure situations (1 run games or tied or tying/go-ahead run at bat)--some save situations, some not. That's honestly a much better use of their innings than I anticipated.

I did the earlier calculation with innings pitched because I have often thought that top-line closers were babied a bit. I wanted to know how much difference there really is between their innings output and their own teams' inning-eating tendencies. I realize it's only one element of the discussion, but it was something that could be researched quickly.