A Bullpen Usage Revolution
I threw this idea out on in a Fangraphs post where it was discussed how many teams’ bullpens fail them each year and how bullpens can often be the death of contending teams as late inning leads slip away. I think the common logic is that relievers are not good enough to be starters, hence they are relievers. Using this thought process, I begin my hypothesis of what if we just didn’t have “relievers” anymore and had a pitching staff comprised entirely of starters? Instead of thinking of a game as nine innings with someone to start and someone to finish, what if we looked at games as three part three inning outings. If someone is perfect, they will face each batter only once before a new fresh pitcher comes out. Looking at the A’s last year, here is how the club as a whole did each time they faced a hitter a second and third time:
| Starters | SO/BB | AVG | OBP | SLG | OPS |
| 1 PA | 2.25 | .244 | .306 | .359 | .666 |
| 2 PA | 2.05 | .234 | .298 | .356 | .654 |
| 3 PA | 1.78 | .259 | .326 | .424 | .750 |
| Relievers | SO/BB | AVG | OBP | SLG | OPS |
| 1 PA | 2.15 | .245 | .324 | .390 | .713 |
| 2 PA | 1.83 | .301 | .355 | .482 | .837 |
It follows basic thinking that the more times a hitter sees a pitcher eventually they will do better against them, and this holds largely true here (despite the very slight improvement the second time around for A’s starters in 2010). Also it is worth noting how regardless of the situation, starting pitchers pitched much better than relievers. Starters facing a hitter once, pitched much better than relievers facing a pitcher once, and the same holds true with a second go round. Furthermore starting pitchers – who I allege are better pitchers in general than relievers fared better in their second go at a hitter than relievers did in their first.
What if there never was (or there was just a severely curtailed) second plate appearance? The numbers would be quite impressive in theory. Last season the A’s hurlers threw 1431 2/3 innings. In any given year, teams should be somewhere around 1458 innings pitched, this is because some games will be fewer than nine innings (road losses that last 8 ½ innings and weather shortened games) and some will last longer. So let’s say we created a plan aimed at maximizing starting pitcher and minimizing the amount of times a hitter faces any given pitcher. The simplest way would be to have every pitcher pitch only 3 innings. And go in a rotation of pitching three innings. If we divided the 1431 2/3 innings into a reasonable 180 innings pitched we would need to have roughly eight pitchers (it comes out to 8.1 for those of you scoring at home). So what would a rotation like this look like? Let’s take the 2011 A’s. What if you did this using the first week of the season as a guide:
Fri April 1st vs. Seattle: 1-3 Cahill, 4-6 Anderson, 7-9 Gonzalez.
Sat April 2nd vs. Seattle: 1-3 Braden, 4-6 Harden, 7-9 McCarthy.
Sun April 3rd vs. Seattle: 1-3 Outman, 4-6 Ross, 7-9 Cahill.
Tue April 5th @ Toronto: 1-3 Anderson, 4-6 Gonzalez, 7-9 Braden.
Wed April 6th @ Toronto: 1-3 Harden, 4-7 McCarthy, 7-9 Outman.
This week probably as it stands now will be:
Fri April 1st vs. Seattle: Trevor Cahill
Sat April 2nd vs. Seattle: Brett Anderson
Sun April 3rd vs. Seattle: Gio Gonzalez
Tue April 5th @ Toronto: Dallas Braden
Wed April 6th @ Toronto: Rich Harden
Let’s assume everyone does their three innings of work (I am not sure how I’d handle things like extra innings yet, but for simplicity’s sake we’ll assume everything ends in nine innings) let’s compare their usage to what their average start length in their careers (this is rounded to the nearest third) has looked like:
| Player | Traditional Usage Model | Revolutionary Usage Model |
| Trevor Cahill | 6.0 | 6.0 |
| Brett Anderson | 6.0 | 6.0 |
| Gio Gonzalez | 5.2 | 6.0 |
| Dallas Braden | 6.0 | 6.0 |
| Rich Harden | 5.2 | 6.0 |
So despite being used more frequently, these pitchers aren’t being used more.
There was a time prior to Tony LaRussa and Dennis Eckersley that the one inning save was a crazy notion. There was a time when a LOOGY or ROOGY would seem like a wasted roster spot. Is it possible that all relievers are a waste? The 2011 Oakland A’s have a very good bullpen and they are paying for it: the 2011 pen will cost Oakland: $17,086,500 (I used Cot’s Baseball Contracts to come up with this figure, taking the actual salary numbers for all the relievers and using the league minimum $414,500 for the others, I excluded Josh Outman and I assumed Brandon McCarthy would be a reliever, anyone who has yet to make their major league debut was also excluded).
What if instead of spending $17,086,500 for the bullpen we invested that in three other starting pitchers. Who could be had this offseason for that money? Basically you could get: Carl Pavano ($8.25), Aaron Harang ($4M) and Vicente Padilla ($2.0M) with some change left over.
| Pitcher 1 PA | SO/BB | AVG | OBP | SLG | OPS |
| Cumulative
A’s ’10 RP |
2.15 | .245 | .324 | .390 | .713 |
| Carl Pavano | 2.71 | .255 | .301 | .345 | .646 |
| Aaron Harang | 2.29 | .246 | .316 | .398 | .713 |
| Vicente Padilla | 4.33 | .196 | .253 | .297 | .550 |
Now your roster looks completely different. You have only eight spots awarded to pitchers, maybe a ninth to prepare and account for injuries, or for handling extra innings and so on, and you have fourteen spots for hitters which could allow for tons of flexibility in the field.
But I wonder how over the course of a season, a team doing this would fare. Wins would truly be a meaningless statistic for this group as whomever started the game would never pitch long enough to earn a win. Saves may all come in the three inning variety and may be spread among every pitcher on the staff. But for the opposing team, you would be facing lefties and righties, you couldn’t platoon guys, you’d get one look at the guy and then maybe not see him again for two weeks and these are your best pitchers in theory.
I am not advocating any team go out and do this, but I wonder if this could be something that could ever work. When I threw the comment out on Fangraphs (albeit a very brief version), there was just one response (not much time has elapsed but maybe people just don’t care also) which was “Mariano was a terrible starter”. That may be because I wrote something about starters being better than relievers by and large with Mariano Rivera being an exception. But, that to me truly proves what I am getting at. Rivera didn’t work as a starter but when his focus was narrowed, he became the greatest closer in the history of baseball, and to me one of the few that actually has been a true difference maker in the history of the game. What if all our starters became focused like that on only three innings, could they see a similar transformation from passable to lights out?
Also, would this sort of pitching schedule be something that could save Justin Duchscherer and Rich Harden and other pitchers who have experienced arm trouble?
I am curious as to what others have to say.
I’m pretty sure the A’s and another team (Rangers?) tried something like this in the minors at one point about, um, 8-15 years ago? IIRC, players hated it, coaches hated it, and any time someone pitched poorly they immediately blamed this crazy system instead of the usual suspects like poor mechanics or poor concentration or poor pitch selection. I think it lasted about half a year.
Oh wow, any idea on where I could scrounge up any info on that? I’m going to try googling it. I guess it is true, there are no more new ideas?
Ah, yes, there was the minor league one, and LaRussa toyed with it in one of his last acts in Oakland:
Voila, the 9-man rotation:
http://community.seattletimes.nwsource.com/archive/?date=19930720&slug=1711920
Wow, I feel ridiculously stupid, or ridiculously smart now. Not sure which.
I also have to say how fitting that Todd Van Poppel was involved.
Ha, indeed. I’ll try to dig up something one the minor league one. Something in my gut tells me I read about the minor league experiment at Baseball Prospectus, so that’s where I’ll poke around first.
Speaking of Van Poppel as a rookie, I should probably share the fact that my Twitter avatar is a picture of Van Poppel’s first major league pitch. I also wrote about it on this blog entry here:
http://catfishstew.baseballtoaster.com/archives/306598.html
That is fantastic. Love the pic! The view over your shoulder would have been much prettier in those days…
Ah, yes, the minor league thing is called the “tandem starter system”. It’s basically an 8-man rotation, and it’s the brainchild of former (and once again) A’s minor league guru Grady Fuson.
Here is one article about it, and if you google “tandem starter system” you’ll find more.
http://www.baseballamerica.com/today/minors/040130notebook.html
Well, now you have inspired a second post regarding this idea!
So I think my original comment conflated the “tandem starter system” with LaRussa’s experiment. It was LaRussa’s experiment that the players hated, and which was quickly abandoned. I think the minor league experiment lasted a bit longer.
The tandem system is still used by a few teams in the minors, though not above fast-A.
It (I think) is a better idea than the 3×3 method, though they share problems. Among them:
1) Less innings for the best pitchers (this is one reason it’s used in the minors, after all)
2) Harder to make use of “leverage opportunities”
3) It can be awkward conditioning a player from outside your org to your system (or vice versa, for trade purposes).
4) It’s unknown how this usage pattern affects player’s arms in the medium/long term.
5) This could be perceived as a negative when wooing FAs.
This is not to say it’s unwise to pursue this (if I were running a really poor club and got the green light to take bold risks, this is something I’d look into). But there are reasons teams aren’t doing it…
Regarding point #1 – I agree, there would have to be some sort of method of ensuring that your better players get a few additional outings so that they pitch more innings. With respect to #4 I can’t say. But with respect to #5 – I imagine it cuts both ways. A pitcher looking to rebuild value may be excited to be a part of a new system. A AAAA pitcher gets more innings than he normally would. A starter who is fighting for a roster spot would have more spots to fight for on this team presumably.
Interesting to think about if anything. Kinda wish some manager would be brave/desperate enough to try something different.
Ever think of running a simulation on this? Do they still make Out Of The Park Baseball sims? They were fun and played out fairly true, I believe. Perhaps someone else knows of a newer/better product. But I’d be curious to see how it would work at the MLB level over a lengthy period of time.
Of course, there’s the question of offense too. You can have a staff of Hall of Famers pitching any number of innings you like, but if the offense can’t score runs conistently it won’t matter. Even with guys like Carpenter and Wainwright and the emergence of Jaime Garcia, and hitters like Pujols and Holliday, the 2010 Cardinals fell flat.
Best of luck, I applaud your outside-the-box thinking in a sport so driven by tradition. Bob Costas and Andre Dawson for Commissioner and vice president of rules and on-field operations, respectively, and cheers!
I’m thinking of trying it as a sim in Baseball Mogul. It would take a decent amount of time however! I don’t know if there’s a more rapid method that could be used to do it and have them hold strictly to the 3 inning outing.
This idea seeks to maximize pitching and squeeze out a better performance from your pitching staff, but as always a team needs to be balanced to win, an 81-81 club last year in Oakland proved that much! Thank you for reading!
Hey,
Interesting article, I was excited to read what you had on it, as I had daydreamed something like this a while back but lack the skill/motivation to actually delve into it.
One thing I was reminded of by this article is a system the Brewers have toyed around with in the last few years. It’s where in the low minors they start games with relievers, and then bring in starters around inning 4 or 5 or so in order to get them used to finishing games (which is rare in the minors because of pitch counts). I’m not sure but I think this has been scrapped, as I’m fairly sure players did not like it. Thanks again for the article.
-Kyle
Kyle, thats interesting regarding Milwaukee’s approach. It almost seems like a backwards version of what I am proposing, taking minor league starters and prepping them for relief work int he pros, if not solely for the sake of mimicking playing deep into games. I’m going to look into that a bit. Check out the trackback below the comments, it goes to a piece I wrote afterwards about the A’s botched experiment with something similar in 1993.
It’s an interesting idea. However, I see a couple problems as well:
1) What happens if someone is pitching really well/not well on that day. Do you leave them in longer/take them out sooner? Or is this a strict “3 innings each, no matter what” deal.
2) Extra innings are a pain. You’d need at least 10 pitchers on the roster at some times (9 so that each guy only has to usually throw once in each day, plus an extra guy for extra innings).
3) It would only work in the American League. PH/double switching in the NL would screw this up.
4) There’s a difference between throwing 100 pitches and getting 4 days off and throwing 50 pitches and getting 2 days off. Even if it’s the same # of pitches over a season, how you spread them out is as important as the total number thrown.
5) If each guy is stricly going 3 innings, then sometimes they’ll end up throwing 70 pitches and sometimes they’ll end up throwing 30. THis could be somewhat cured by having each guy work on a pitch count, but then you’re going to need 10 (possibly 11) guys on the staff in case you have days where too many pitches are required…..
Not that it’s a terrible idea, because it’s not. But it’s got a few flaws, and I don’t know exactly how you’d work around them.
With respect to #1 I think that you do take them out even if strong, thats sort of the key to the system of minimizing the views a hitter gets at a pitcher, but if they are just throwing up batting practice, yeah you don’t want them in a full nine. I assume you’d need to have some “mop-up” type guys to make it work, also to account for extra innings (your #2 point). You are completely right regarding the DH, hadn’t even occurred to me, with me being an AL fan that’s right it wouldn’t work in the National League at all, or would create a multitude of problems though perhaps the added roster spots allotted to hitters could create more flexibility for an NL game, it would frequently mean replacing superior hitters I imagine through numerous double switches. Regarding #4-5 yeah I agree, not sure how one would handle that. The main premise of this whole idea is just to see if there’d be a way to minimize the advantage hitters begin to build when becoming more familiar with a pitcher. I appreciate your comments, lots that hadn’t even occurred to me.