A New Kind of Starting Pitcher?

A reader named Tony Trout, a 40-something data analyst in Madison, Wisc., wrote us a while back with a proposal for a fundamental change in how baseball is played. His lifelong interest in baseball actually predates his birth: “I am named for Tony Oliva, who was voted AL rookie of the year a few days before my birth.”

If you have a hot reliever who is not your closer and can pitch two innings consistently (as in Mariano Rivera in 1995), why not start the game with that pitcher? If you could get two solid innings out of a reliever at the start of the game, your actual starter would only need to pitch seven innings to finish the game. Or, that actual starter could pitch just six innings and hand it over to the closer for the ninth.

I would suggest calling the new relief position “The Opener.”

DESCRIPTIONPhoto: Stephen J. Dubner The opener? Joba Chamberlain firing the last pitch in a recent Yankees victory over the Texas Rangers. Should he be pitching the first two innings instead?

The idea is hatched from a few observations:

First, there are almost no complete games by starters anymore. Number of complete games that are not also near no-hitters or shutouts? Very few. With the high number of pitches necessary to retire modern hitters, starters inevitably flag. Because they want to delay the use of the bullpen as much as possible, managers tend to leave the starters in until they’re tired and ineffective.

They pitch until they fail. I can’t help thinking this is psychologically suboptimal for the large-ego starting pitchers.

So, why ask starters to pitch until they fail? Why not ask them to pitch for just six innings, the third through the eighth? Most starters would be delighted by this “light” load and pace themselves accordingly. I’m not a pitcher or even a baseball player, but I just think there’s a lot of weakness in not knowing how long you’re going to be out there.

The second observation is that hitters are thrown off by a change of pace. Aside from being tired in the seventh inning, the starting pitcher is also facing the same line-up for the third time. The hitters may be getting accustomed to that pitcher’s rhythm. If you could actually start the game with a pitcher who has a phenomenally different pitching style, you could help the “starter” to be successful when they appear in the third inning.

Ideal candidates for this “opener” would be fireballers, side-arm specialists, odd lefties, and knuckleballers.

The third observation is strictly anecdotal: hitters love the first inning. I think hitters really get geared up for the top of the game. How many times does a game start with the starter getting shelled? I think starters are also pacing themselves early on, hoping to get through the first with an “easy inning.” If an opener knew they only had to pitch two innings, he could pace himself and be successful.

Look at closers: one of the benefits of being a closer is that you only need to go out there and collect three outs. Let’s just give another pitcher that same assignment, only double it and put it at the top of the game.

The fourth observation is the effectiveness of pitchers in the All-Star game. Admittedly, they really want to perform well. But so do the hitters. What is known is that they’re going to pitch two innings and two innings only. There is a benefit there.

Finally, if a third-inning starting pitcher had good stuff, he could pitch a mere seven innings and close out the game. On a regular basis you could get through the game using only two pitchers.

The ultimate pitcher for this assignment would be a knuckleballer. Hitters hate them in their own right, but also hate them because they throw off their timing. The third-inning pitcher would benefit from this. Also, knuckleballers are really only playing catch: a knuckleballer could practically pitch two innings every day. Two knuckleballers could cover you for the whole season. The pitching staff would be two knucks, five starters, four long men/set-up men and a closer. Cinch.

There will be institutional resistance to this idea:

1. Starters apparently have this elaborate warm-up routine that would be thrown off by not knowing when they are actually going to start throwing. I think this is crap. The home starting pitcher knows he’s going to throw the first pitch at 7:05 p.m., but the visiting pitcher can start anywhere from 7:08 to 7:45.

2. Starters could get fewer decisions. If the winning team scored all their runs in the first two innings, it would be really hard for the scorer to deprive the opener of the win. He was “the pitcher of record.” On the other hand, according the scoring rules, the pitcher who starts the game can’t get the win unless they pitch five innings. So, I guess the opener would lose out on that one. Problem solved!

3. Starters would be deprived of the opportunity of complete games, shut-outs and no-hitters. To that, I say, boo hoo. Those are three antiquated statistics. We’ve moved on. The No. 1 starters could still start their games. You’d probably want them to. This concept would work great with a fifth starter (and probably a fourth and third starter too).

4. Starters take great pride in starting. They’re probably contractually entitled to X number of starts a season. If you manage to get around the ego, you probably can’t get around the contract.

5. Everybody who wears a uniform hates knuckleballers. Hitters hate them. Managers, many of whom were catchers, hate them. I think fans would tire of them too. The knuckleball is underused in baseball. Considering the ratio of their effectiveness to their cost, there should be at least one knuckleballer on every staff. There’s not because of an institutional dislike for the pitch. It’s not real baseball, apparently.

6. If you have a pitcher good enough to go out and pitch two scoreless innings reliably, they’re probably one of your best starters or your closer already. In the zero-sum game of available talent, are you going to want to use up a good pitcher early? This is probably the biggest hurdle.

7. Baseball people are pretty conservative. Notice how long it takes stylish facial hair to appear? And then how long it takes for it to disappear? They don’t like change. If it didn’t work the first few times, it would be disdained and discarded.

Still, I would expect some teams would toy around with it. There should be pitchers who would excel in two innings of work. Thanks for your consideration.

You may think Tony’s idea is absurd — and I’m sure you will let him know what you think in the comments: not all flamethrowers are baseball people, but it seems all baseball people are flamethrowers. If nothing else, I bet you will think of “the opener” the next time your team falls behind 6-0 by the third inning, yanks the starting pitcher (and renders him useless for the next four days), then brings in a reliever who proceeds to shut down the other team.


I think this would be a great idea for some teams, if they don't have a "star" pitcher, or maybe have a below average bullpen, this would let them organize the talents they do have in the most productive way.


Here's the scenario of why this will never happen:

Opening pitcher: 2 IP, 0 ER
"Starting" pitcher: 2.2 IP, 7 ER
Long Relief: 3 IP, 0 ER
Worst reliever: 1.1 IP, 0ER

So now you "used up" one of your best pitchers in a game he normally wouldn't have entered due to poor "starting" pitching. By having the starting pitcher start, you can selectively use your strong relievers in games that are close and where their skills are needed.

Kevin Camp

I would not be surprised if this starts at the High-School level by a risk-taking coach, similar to the football coach in Arkansas that hasn't punted in two years.

Mike B

It might work better, but losing the drama of a no-hitter and a complete game would really kill the TV ratings. Many people are apt to tune in when there's a no hitter going, especially in the era of on demand television where almost any game is available to any fan anywhere. No hitters might happen infrequently, but no-hitters through 6, 7 or 8 innings happen frequently enough that it gets fan's hearts pounding and attracts eyeballs to the relatively mundane affair of a baseball game.

The job of professional baseball teams isn't to win games, its to make money and the system of two starting pitchers duking it out put buts in seats and eyes on screens.


I disagree with two points here. First, I think the comment that knuckleballers are merely "playing catch" is absurd. They are still throwing 70-80 mph which puts strain on their arm, albeit less strain than a pitcher throwing 100+ mph. I highly doubt any knuckleball pitcher could pitch 2 innings every game without breaking down before the middle of the season. Secondly, the no-hitter and perfect game are two of the most difficult feats in all of sports. These events are always headline news, showing they are not "antiquated statistics" and that nobody has lost interest or "moved on." It would be an incredible loss to the game of baseball if there was never a perfect game or no-hitter again.


I've been thinking about this for a while. The only two possible issues, as I see it, are as follows:

1) Whereas your closer only came in for close games, your "starter" would come in for every game, increasing the load from 75 / year to 162 / year. Thus, it would probably only work with knuckleballers and other soft throwers, or you'd need a few of them.

2) In theory, one reason a closer is good at his job is because he loves to be "the guy" during the crucial moments. This reversal takes that away from him. It could be a good use of his physical talents but a mismatch for his mental ones.

Still, intriguing for all the reasons noted, and it would be fun for some team to try.

Johnny E

It all depends on the reliability of your pitching staff. If you have a bunch of good pitchers that can last 8 innings but one reliever who always blows it that would affect how you use the rotation. Maybe have everybody pitch only a few innings but play more games. It might save their arms in the long run. The other team won't be able to get used to one pitcher.

Don't forget the concept of momentum. Maybe you statisticians can figure out if it's harder to score runs if you're behind early in a game. That would affect who you want to be your starter.


There's no "institutional dislike" of knuckleballers, there's a derth of effective knuckleball pitchers. It's hard to find a pitcher who can not only throw the ball without spin *and* still have enough control to throw strikes. You open two innings with a knuckler and you'll have plenty of those games where your "opener" leaves you behind 6-0 in the first inning or two, and an otherwise effective starter is wasted.

It's also a question of risk management and leverage - you want the guy who will pitch the most innings and therefore have the largest effect on the game to go first. If he blows up, you can throw your worst relievers. If he does well, you throw your best relievers, or possibly get to rest the bullpen entirely.


Under this scenario, the "starter" is expected to pitch innings 3 - 9. However, in the NL, the pitcher's spot in the batting order might come up in a key late-inning situation. (Let's say: 7th inning, 2 outs, man in scoring position, game tied or very close.) It would be typical for the pitcher to be lifted for a pinch-hitter in this situation. However, your "starter" has only pitched 4-5 innings, so now you have to get a total of 4-5 innings from your bullpen which is not ideal.


It wouldn't make sense to do this with a real Ace, your Halladay and Greinke and Lincecum, but for teams with some variation of the "Spahn and Sain and pray for rain" it might be helpful to get through those dry patches until the aces are ready again.

I certainly think this could catch on as something that might happen occasionally, maybe 10 times a year for a team, depending on days off and opponents and rotations.


This type of theory surfaces every once in a while. Tony LaRussa tried it -- I believe he dubbed it a "pitching platoon" and would have each guy go three innings. That didn't last long.


It's almost certainly true that having a guy who can get 2 scoreless innings regularly would be wasted as an "opener" instead of being used as a reliever. Any reasonably well-informed fan knows that the more decisive innings (those with higher "leverage" to use proper terminology) occur towards the end of a game. Simply put, it's more valuable to keep a 2-2 game 2-2 in the 7th than it is to keep a 0-0 game 0-0 in the first; you have many more opportunities to come back if you go behind early than if you go behind late.

Second, the proposal practically guarantees usage of 3 pitchers every game. "Closer" usage is basically formulaic in that any time a team is leading by <4 runs in the 9th, if the closer is available, he will pitch to "save" the game. This proposal ensures that in any game where your team is ahead by <4 runs, you get a maximum of 6 innings from the starter, and if your team is losing, you will likely have brought in a reliever anyways. Instead, if you start with a conventional starter and he is pitching well, he may go 8 or more innings in a close, low-scoring game and turn the game directly over to the closer.

Finally, Tony ignores some of the basic research that has been done lately, which suggests that starting pitchers tend to decrease their performance most the third time through the lineup. This would suggest that if you were going to adopt some strategic difference in pitching, you would opt to employ more 4-ish inning pitchers, rather than a 2/6 strategy like this suggestion, as proposed by Ken Funck here: http://www.baseballprospectus.com/article.php?articleid=9228.



To take it a step further, what about a pitching staff made up entirely of "relievers?" This would probably make a lot more sense for a small- or mid-market team that can't afford to compete money-wise for big name starters. Basically treat each game like an All-Star game, in which each pitcher isn't expected to go more than three innings, max. Collect a bunch of low-price rubber-armed guys who can give you six to nine outs three or four times a week and just keep running them out there, either situationally or in a semi-planned rotation. Yeah, you'll miss the glamour of a big name starter, but you'll also miss having high-profile free agent signees bust out or get hurt in spring training.


Hey, I have an idea...let's turn the conventional wisdom on its head and do the opposite, just to be really different!


Why not take it a step further and stack your entire pitching staff with middle relievers who can go 2-3 solid innings 3 or 4 times a week. Then each night trot out 3 or 4 pitchers for your pitching line-up.


Interesting theory. I think its marginally useful. Teams like the Pirates have nothing to lose on this strategy, they have lost for 20 years straight. But you have to consider a few other things.

1. I don't think one guy can realistically pitch 2 innings per game. There aren't enough good knuckle ballers out there, and normal pitchers can't do this, no way. So I think at a minimum, 2 starters would be needed. This may not be a bad thing, having 2 starters, but its realistic.

2. If the starter gets shelled in 2 innings, and your ace is going to start the third, his talent is basically wasted. His effort is not valuable since the game may already be lost.

3. If your team is winning when the starter is pulled and then the third inning guy comes in and starts giving up runs, what do you do? Your best bullpen guy started the game, the manager is probably short on options. You can't leave the thrid inning guy in, but you don't want to lose the game. This would be agonizing as a manager.

Still, overall I like the idea, I hope some teams at least try it.



I think Tony answered his own question in the first sentence of his plan...

"If you have a hot reliever who is not your closer and can pitch two innings consistently (as in Mariano Rivera in 1995), why not start the game with that pitcher?"

He just name one of the greatest closers of all time, and referenced a time when he was actually the set up man on his team. So he's basically talking about a guy that comes along once every 10 years or so. If you could come up with a few Mariano Riveras on your team this might be a decent plan. Otherwise, you are just shifting some guys who are not good enough to start or close from the middle of the game to the beginning of the game.


The idea listed above will and should not be implemented not because it is unorthodox, but because it is also a poor use of resources. The value of a run in a 0-0 game in the first inning is significantly less than the value of a run in a 0-0 game in the 8th inning. As an extension of this, teams benefit from deploying their ace relievers in high leverage situations, which generally occur from the 7th inning onwards in games that are close. Using a strong reliever for two innings to start the game would mean that the pitcher would be unavailable for a more important juncture in the game. Additionally, if you had (at least) two good relievers that alternated days, their workload - about 81 games and 160 innings - may not be achievable given that pitchers spend their entire careers learning to throw in specific patterns. Additionally If a pitcher is good enough to throw the first two innings of every 2nd game, why would they not be good enough to throw the first 6 innings of every 5th game, a pattern that would result in more total innings being thrown by a supposedly good pitcher, a positive result.

If any unorthodox ideas regarding pitcher usage were to be implemented, by far the most important would be deploying ace relievers at critical junctures. The pitchers that close games tend to only pitch the 9th inning of games their teams lead by 3 runs or less, a very inefficient usage. It does not take a math whiz to figure that a tied game in the 8th inning against a team's 3 best hitters is a more difficult situation to face than a 3 run lead in the 9th against the 7-8-9 hitters in a lineup, and yet most managers would pitch their ace reliever in the later, not the former situation.

A more interested unorthodox approach to pitching would be to use 6 or 7 starting pitchers and perhaps have the two strongest start every five days as they would under current usage patterns, and have the other 4 pitch three or four innings every three days. That way you could match up pitchers of opposite handedness during the same game, and benefit from the fact that pitchers tend to do significantly better throwing over shorter stretches and when facing lineups only once. If anything weaker pitchers would benefit from more frequent shorter appearances, not the better ones.


David L

First of all, the premise is flawed: good managers don't leave a starting pitcher in until he's struggling. That comes down to knowing your personnel. It is pretty typical for a starter to finish out the 7th and be replaced in the 8th by a middle reliever after a solid performance (potential shutouts and no-nos are a different story, and in those cases etiquette and tradition often supplant good managerial decision-making).

The main reason this is a terrible idea, however, is best explained with option theory. The option (defined as having your best strike-throwing reliever available) has the lowest intrinsic value (and highest time value) at the beginning of the game. As the game plays out and the game state changes, the time value of the option decreases, and the intrinsic value (which is predicated on the game state) takes on the greater proportion of the option value. Exercising the option when the intrinsic value is zero is stupid.

Another way to look at it is leverage. On defense in the 9th inning up by 3 or fewer runs is an extremely high-leverage game state, and you want a pithcer who can convert the save. The first inning is a low-leverage situation, because there is nearly 0% certainty about the outcome. In other words, if you burned your closer in the first inning and then the game was a blowout (in either direction), you would have made an error. Furthermore, if you burned your closer early and you ended up in a save situation in the 9th, you would also have made an error, because now the high-leverage game state is handed to an inferior pitcher. So either way, if you burn your closer early, you are almost certain to have made an error.



Unfortunately this idea is really dumb.