I've read this thing around here advocating a starting pitcher come out for short relief 3 days after his last start. Supposedly it's his day to "practice throw" anyway, so why not use a really good starter to pitch one/two hitters to get out of a jam?

This, more or less, was SOP in the major leagues from roughly 1930 to roughly 1965. I think its applicability today would certainly depend upon the pitcher, but conceptually it makes a great deal of sense.

Then read the threads and you can determine what everyone's position is rather than assigning a position.

Way to refuse to engage on the substance of my argument, Backlasher. Didn't the fact I was quoting the thread in question give you a small hint that i might have read it?

Your position stated several times in various forms in various threads is:

"You should build a utilization plan that maximizes your on hand talent. If you must have a default, closer is better, because its better to not risk injury than it is to risk injury and lose all the innings."

The first sentence is the sort of truism no-one can argue with. The second makes a heroic assumption that the closer model minimises injury over all available alternatives. Yes you have data showing career length has risen over the course of major league history (for all pitchers, natch. you certainly don't have any ace reliever/closer specific data). But until someone can disentangle that from advances in sports medicine, conditioning etc then all we can say is that is exactly what we would expect to have happened absent any data and without even thinking about pitcher usage.

You keep on quoting LI approvingly. But LI shows that using the current closer model often leads to pitchers coming in in relatively unleveraged situations (three-run lead in the ninth has a LI of one). My contention is that you can improve the average LI of your bullpens best reliever without using your bullpen in unpredictable ways. Find a top reliever capable of going two innings when well rested and bring them in the eighth when leading by one or tied. Have another pitcher you are prepared to trust with some of your three run saves UNLESS the game is absolutely critical (ie late season game a gainst a rival for a post-season slot, in the post-season etc). Yes there may be times when this missesD key factors a manager would want to take into account - but it looks like a better rule of thumb thsn utilisation based on the save rule.

For those who don't want to plough through 300+ posts in the other threa here are some illuminating posts by Tango:

Tango on the three run save:

1 - Stop using your reliever to start the 9th inning with a 3-run lead. The LI in that situation is around 1.0.

2 - Given that you take 10 games or so off the table, make your reliever pitch in either 10 other games, or give him 10 more innings in his other 50 games. Which 10 games? Well, that's the fun part to discuss.


Tango on Foulke's 2004:

Even if you want to discard the Sox/Yanks as a legitimate comparison, and focus on the fact that the Sox had a powerful hitting lineup where a high-leverage situation may not materialize by the time the 9th inning rolls around because they scored so many runs leading into that, then you bring in Foulke to pitch 2 innings instead. Foulke came into those 19 games just to "get his work in for 1 inning". That's a tremendous waste of games and innings. His 72 games, 84 innings (or whatever it was) is really more like 50 games, and 62 innings where it really mattered. If it was obvious that the Sox team was such a powerhouse that he would only be needed in 50 games, then, maybe him pitch an extra 22 innings in those 50 games.

Tango on behavioural distortions caused by the save rule:

It might also suggest that there's no maverick managers in the midst, and as long as it's the manager's a-- on the line, the GM is limited in his influence to try a different approach.

Managers don't want to lose the game in the 9th. They can live with having a suboptimal player in the 8th and losing the game, but not a suboptimal in the 9th and losing the game.

Managers manage on fear. If they gave out a save based on LI, I'd bet the managers would change their approach, too. And relievers would clamor to go in with the bases loaded and 0 outs, and get lots of LI points for dousing the fire.

Managers manage on fear. If they gave out a save based on LI, I'd bet the managers would change their approach, too. And relievers would clamor to go in with the bases loaded and 0 outs, and get lots of LI points for dousing the fire.

This is one of my favorite elements of this whole discussion. Given that:

- There is no question that the save rule, its publicity and its consequent economic power, has significantly impacted standard bullpen usage over the past 20-25 years

- These changes have been enacted to serve the self-interest of star relief pitchers (and their agents) and managers, and not necessarily to serve the interests of their teams

and

- This dynamic is one of normal, predictable, human nature, not something we can ever expect to change

then the concept of coming up with a more comprehensive, nuanced metric that better captures the actual value of effective relief pitching than the save metric does, is extremely intriguing. Since we know that, over time, if the metric drives the economics, then the metric will drive the behavior, then why not just get a metric to the forefront that would serve team interests optimally?

Obviously it's far easier said than done, but to the extent that it could be done, it would have 10,000 times more impact on modifying real-world behavior than the railings of sabermatricians, even sabermatricians employed by MLB front offices.