Owing to the width of most of the many data tables on this site, it is best viewed from a desktop computer. If you are on a mobile device (phone or tablet), you will obtain a better viewing experience by rotating your device a quarter-turn (to get the so-called “panorama” screen view).
The Owlcroft Baseball-Analysis Site
Baseball team and player performance examined realistically and accurately.
You can get a site directory by clicking on the “hamburger” icon () in the upper right of this page.
Or you can search this site with Google (standard Google-search rules apply).
Search term(s):
Optimum Player Use
“There is no man living who isn’t capable of doing more than he thinks he can do.”
The way in which batters and pitchers are used can have a substantial effect on the degree to which they can or cannot perform up to their full abilities. Here we discuss ways to maximize what a team can get out of its players. (Some of these arguments are rather discursive: please muster your patience.)
Let us begin by setting out some criteria. First, it is essential that every position on the field be backed up at least twice: that means that in a given game, the team can lose any two players and still field a defensively acceptable squad. A team can “lose” men in a game in numerous ways: injury, ejection, an illness or injury that does not rise to the Injury-List level but keeps a man out of service for a day or three; or even just being burned as a pinch hitter who does not remain in the game, or by being pinch-hit for.
Second is play time: a man must get enough to stay sharp, but not so much that he gets fatigued by middle or late season. Do we know what that means in numbers? Yes, at least approximately. Consider the (quite wise) old baseball saying: “Miss a day of batting and when you come back you’ll notice it; miss two days and when you come back, your teammates will notice it; miss three days and when you come back the fans will notice it.” That’s suggestive, though scarcely definitive.
We did a broad-brush investigation a few decades ago and concluded that to play up to his innate abilities, a batter needs to play at least two-thirds to three-quarters of the time. Like the aphorism, that study wasn’t probative, but we regard it as quite indicative, and believe that two-thirds playing time is the floor below which primary players (a term we will define more exactly a bit farther on) should not be allowed to go.
At the other extreme, the “iron man” who plays in almost every game (or, indeed, all 162 games) should be a thing of the past. There will always be the occasional player who can do that over an extended number of seasons and still do well, but: a) “occasional” is very, very occasional; and b) even with those few, we cannot really know if they mightn’t have done better yet with some reasonable rest.
We find it convenient to think of these things on the basis of a seven-game cycle—essentially weekly. There is nothing magic about that number, but—as we will see shortly—it is convenient for discussion.
Thinking in terms of such a seven-day cycle, to reckon optimum use, primary players should get no fewer than five days’ use and no more than six days’ use. (Two-thirds of a week and three-quarters of a week—the minimums—both work out, when rounded to whole days, to 5 days.) Now let’s see how that would work in detail.
We can start with the outfield. There being three outfield positions, in a seven-day cycle there will be 21 man-days of play. We fill that out by having three “primary” players and one “floater”. The primary players are the ones who each play a set position five or (in one case, presumably the best batter of the set) six days a week. The “floater” is a man who can play all three outfield positions (and hit) at least adequately: he also plays five days a week, rotating through the primary threesome’s rest days. (To meet the double backing-up requirement, we also need at least one infielder who can also play an outfield position for a day or two without being a disaster.)
Next consider the infield (for simplicity, we will include the DH as an “infielder”). There are thus five positions, each of which should be manned by “primary” players who play six days out of seven. The seventh day for each should then be filled by a “floater”, who thus plays five days a week (one day each at the five infield positions).
(Sometimes you can find a man who hits decently and can actually play each infield position adequately or better; but if, as is more likely, you cannot—the snag is typically shortstop—it is often not too hard to juggle things—for example, have your primary second baseman play shortstop one day a week with the “floater” playing second twice a week…and so on. It’s not usually difficult to arrange.)
Mind, fulfilling the “every position backed up twice” rule also imposes some requirements. Since we are only considering emergency cases—part of a game or, at worst, one or two days while a callup arrives—we don’t require that the second backup at any position be great at it, just adequate will do. And remember that we need at least one infielder who can emergency-backup an outfield position. Again, it’s usually not too hard to arrange.
At catcher, there is a primary catcher and what we may call a secondary catcher (not a “reserve” in the sense we have used that term before, because the secondary still gets to play a good bit). Catcher is a very physically demanding and draining position over a long, hot summer, and so catchers need more off days than other players. We reckon that two days out of three is probably optimum (and if at all possible, never catching a day game after catching a preceding night game). That breaks the by-weeks pattern, but catchers are a special case.
If you add it up, we have four outfielders, six infielders, and two catchers: that makes twelve men, which is one fewer than the thirteen normally alloted to non-pitchers. The thirteenth man would ideally someone who can play almost every position, including catcher, at least well enough to get through a game or two. He needn’t be much of a hitter—but what a bonus if he is!—because he is only an emergency spare part. We say “only”, but the nature of the game, with nagging little injuries, not to speak of outright Injured List trips, will usually mean that that thirteenth man will get nontrivial playing time. That slot is not so simple to fill, but the shrewd team will seek out such men, or develop them in their minor-league affiliates.
Obviously, no team could ever adhere rigorously to such a schedule: “stuff happens”, as they say. But it represents a useful template that management should strive to follow as closely as day-to-day circumstances allow.
These are probably the two simplest—and possibly most important—equations on this site:
pinch hitting = BAD
platooning = BAD
Let’s see why.
Pinch Hitting
Teams that regularly pinch hit based on handedness are showing a lack of grasp of baseball analysis. We don’t need to here re-invent the explanations wheel: this matter has been exhaustively researched over the years by numerous analysts. For instance, in the excellent book titled simply The Book, Tom Tango et al. go through an in-depth analysis and reach this conclusion (on page 172 of the paperback edition):
“Because batters perform significantly worse when pinch hitting than when starting, bringing in a pinch hitter to face an opposite-handed pitcher is of minimal value unless the batter he is replacing is much worse.”
And Tango is scarcely The Lone Ranger on this topic. Here are a few more representative analytic analyses of pinch hitting, with a salient quotation from each:
“In fact, this brings into question the usefulness of pinch hitting as a strategy—if players perform
that much worse coming off the bench, the player being replaced had better be really bad.”
“Gaining the platoon advantage? Fine, though you won’t get a huge pickup unless you have a great pinch hitter. Not getting
the platoon advantage? Don’t do it! Fifteen years on, the math of pinch hitting looks basically the same as it did in 2006.”
“Collectively, pinch-hitters are much worse than the average bat.
They probably always have been and probably always will be.”
And, as a footnote, we remark that no well-constructed team should ever have a batter in the line-up who is so inferior (“the batter he is replacing is much worse”; “really bad”) that a pinch hitter might actually make sense.
Platooning
Once again, we don’t need to re-invent the wheel. In The Book, Tango et al. reach this conclusion (on page 158 of the paperback edition):
“[A] good right-handed hitter with an average platoon split is much better against lefties than an average right-handed hitter with a large platoon split (and of course, is much, much better against righties). Likewise, we see that the variation in platoon splits for left-handed hitters is significantly less than the variation in overall OBP or wOBA skill levels. This doesn’t mean that we should ignore variations in platoon for batters, but it does indicate that it is of secondary importance.”
In other words, a good batter is probably as good or better on his “bad” side than an average (or worse) batter on his “good” side. You can see, if you hunt about on the web, that others generally agree; here’s one example from a different FanGraphsanalysis by the above-cited Ben Clemens (“The Platoon Advantage Will Mislead You”, August 17, 2020):
“In other words, which teams have the platoon advantage most and least often is often just a proxy for which handedness their best hitters are, and whether they have hitters good enough that they should play every day regardless of matchup.”
And all that ignores the biggest negative of platooning. Roughly 73% of all batter-pitcher encounters are with a right-handed pitcher; that means that in a strict platoon, the right-handed member will only get 27% of the plate appearances. That is nowhere near enough for the typical batter to maintain his best level of performance. Platoons weaken overall team performance.
Putting a lesser hitter in the lineup for a better one just because the lesser has the “platoon advantage” and the better one doesn’t is just plain bad baseball.
Fact: the higher in the lineup, the more seasonal plate appearances. The #1 slot gets about one-eighth of the team PAs, while the #9 slot gets about 10% (and the #5 slot, midway down, gets the average one-ninth). In 2019, the average MLB team had 6,217 Plate Appearances; that means that on average, the PA difference between the #1 slot and the #9 slot was a full 155 PAs*. Looked at another way, that’s about 17 annual PAs per slot increment (that is, moving a man from, say, #4 to #5 costs him 17 annual PAs, and vice-versa for moving him up from #5 to #4.) And at season’s end, the team runs scored—and thus games won—will change in some part owing to how high up (or far down) its better hitters are placed in the lineup.
* = In The Book, Tango et al., using actual PA data from 1999 - 2002, got 151 PAs as the average difference over those years. Obviously, the rule of thumb is quite good.
The naive begining would thus be to start with a lineup order based solely on TOPs (career TOPs, that is), with highest going first. But the matter is hugely more complex than that. Once again we are saved the trouble of re-inventing the wheel owing to the work of Tom Tango et al. as disclosed in The Book. Their analyses are thorough, fact-based, and—we feel—conclusive. Chapter 5 of The Book covers the matter, and is well worth examining, but we will here cut to the chase:
“Your three best hitters should bat somewhere in the #1, #2, and #4 slots. Your fourth- and fifth best hitters should occupy the #3 and #5 slots. The #1 and #2 slots will have players with more walks than those in the #4 and #5 slots. From slot #6 through #9, put the players in descending order of quality.”
Mind, immediately after they refer to “how little impact the ordering actually has.” It is also worth quoting this:
“Don’t consider the strikeout, or the ability of a hitter to move runners over on outs, when constructing your starting lineup.”
If you think all that through, you will see that it really doesn’t vary much from the naive “put ’em in TOP order” approach. The only difference is having the #3 hitter (who used, alas, to traditionally be thought of as the best hitter on the team) should be your fourth-best hitter. That makes a lineup that looks like this:
#1: best hitter (but also consider on-base rate)
#2: second-best hitter (but also consider on-base rate)
#3: fourth-best hitter
#4: third-best hitter
#5: fifth-best hitter
#6: sixth-best hitter
#7: seventh-best hitter
#8: eighth-best hitter (but see below)
#9: ninth-best hitter (but see below)
In general, if the call is close on rankings, put the batters with the higher on-base percentages higher, and the ones with the higher total-base percentages (TB/PA) in the middle.
There is one final note on lineups from The Book, but it only applied to the National League, in which they used to play real baseball, with pitchers batting. The analysis has to do with the reality that when the lineup turns over, the fine hitters at the top will be a hair more productive if they have a hair’s worth more likelihood of batting with a man on base.
One does not need to be a sports physician to grasp some basic points about repetitive stress exercises. The key point for us is that fewer repetitions more often produce better per-rep results than more repetitions less often. Let’s turn that mouthful into a simple example.
Imagine an athlete doing weight-lifting exercises—say curls. Imagine that he or she is using a weight such that in a set of 10 repetitions, he or she can just barely complete the tenth curl. Further imagine that he or she is repeating a 10-curl set every 20 minutes. (That is only a thought experiment—it may not be realistic as to the exact numbers, but it is the principle we are looking at.) In an hour, the lifter will have completed 30 curls, and the last two or three in each set will be sloppy form because the lifter is always approaching his or her fatigue point for that weight lifted that many times.
Now suppose that instead of 3 sets of 10 curls every 20 minutes, the lifter switches to a regimen of 6 sets of 5 curls each every 10 minutes, for 6 sets in an hour (so still 30 curls). How will the lifter’s form will be on each fourth or fifth curl? Lots better is the answer.
Our use of pitchers is still, even well into the 21st century, ultimately derived from the 19th-century notion that the pitcher is like any other player on the field: he plays all nine innings. Anything less is a failure. Sure, we don’t consciously think that way, but down in the subconscious we still do. And it shows in pitching use.
We have at least finally noticed that a starter typically does materially worse in his third (and beyond) run through the opposition’s lineup. Part of that is the batters seeing his stuff for a third time, but another and definitely non-trivial part of it is fatigue. Today, the average MLB starter goes barely over 5 innings per start (here is a most useful discussion from Steve Potter). As a rule of thumb, the starter stays in till he shows signs that he is reaching, or has reached, his fatigue point; all too often, those signs are poor performance and thus excessive runs given up in an inning. By the time he is finally taken out, the damage is done. (That sort of pitcher management is what we have often referred to as the manager letting the game manage him, a parallel to the old criticism of an infielder “He let the ball play him”.) Surely there is a better way?
Of course there is a better way. Indeed, there are several variations on “a better way”, each with its own advantages and drawbacks.
The simplest in principle is a bit crude, but still likely a lot better than current practices. It is to schedule three pitchers per game, which we can refer to as the “opener”, the “primary”, and the “finisher”. The opener goes 2 innings, the primary goes 5 innings, and the finisher goes 2 innings. (We will deal later with the complications that arise if one of those men needs to be relieved ahead of schedule.) The advantages are these: no pitcher is over-extended; and each pitcher knows to a nicety just when he is coming into the game and how long he will be expected to pitch. The drawbacks of that simplistic approach are that an “inning” is a flexible thing: a man who is struggling, but not so very badly that he needs to be pulled, can throw a lot more pitches than expected. Further, also because the outings are innings-based, a given pitcher may see an opponent more times than expected. But it’s a conceptual starting point.
Other approaches possible could be based on pitch count or on number of batters faced. Those have substantial advantages over the naive 2-5-2 approach, but also problems of their own. The chiefest such is that neither of them will necessarily finish out a game. Take men faced, probably the best approach; we would like the opener and the finisher to face 9 men each (that is, go once through the lineup), and the primary 18 men (twice through the lineup). But that only totals 36 batters faced. Nowadays, the average per-game BFP total is between 37 and 38 (37.56 in 2024), and that’s the average: individual games, even without extra innings, can easily run to a good bit more.
Let’s sidetrack to consider how such a method (9-18-9) would affect pitcher demands. We take the primary pitchers back to four, in the belief that a man can readily pitch five innings every fourth day (which seems plausible, as it will always be a maximum of five innings). The short-run pitchers need depends on how often a pitcher, once acclimatized to it, can pitch to 9 men (roughly, but not exactly, two innings’ worth). Obviously not every day; equally obviously, at least every fourth day. Despite decades of asking that question of presumably knowledgeable sources, we have yet to receive a definite opinion. Our own opinion is that asking a man to do that every second day is asking too much, but that it should be quite feasible to ask him to do it every third day (recall again that he will know exactly when he is coming in and for exactly how long he will be expected to go, which eases the warm-up demands). If we assume that, then we need six men for those roles (each of those six would alternate between opener and finisher at each appearance). That means we require ten “scheduled” pitchers. Till very recently, that was a major drawback, but with the (long overdue) opening up of rosters to 26, allowing 13-man pitching staffs, it becomes feasible, as we would then have three true “relievers” available.
Probably the best assignment of the three “unscheduled” men would be two “firemen” and a swing or utility man. The firemen would be used solely when the current pitcher is struggling and there is a crisis. A “crisis” can be readily and exactly defined by the widely available probable-runs tables that cover all 24 outs/runners-on cases. For example, if the appropriate Table (you can easily Google up such Tables) shows that with runners on first and third with one out, the inning-total run expectancy (not counting any runs scored before getting to that situation) is 1.140, is that a “crisis”? It depends on your personal standards: the highest inning-run expectancy (in that particular Table) is 2.282, for bases loaded and none out; the lowest is 0.095—close to zero—for bases empty and two out. What level is a “crisis” depends on your tolerance for risk. By “tolerance for risk”, we mean what average probability of run-scoring for a particular outs/runners situation is the threshold for deciding to pull the current pitcher? Is it 1 run? 1.5 runs? 0.75 runs? That’s a managerial preference, but unless the manager has a good idea of what his Table of preference shows, he’s just guessing.
We need to say a word here about “appropriate Tables”. The actual probability numbers in a run-expectancy Table depend on the “run environment”—that is, on the average number of runs per game. That average changes over time, and indeed to some extent by park. There are tools available that will generate a Table for any given run environment, and even for the identity of the batter at the plate. Using an inappropriate Table will give incorrect results. The Table we used for the example just above (simply the first that came conveniently to hand) set the run environment at 4.15 runs per game.
Arguably even better as a gauge is Tom Tango’s “Leverage Index”, which has the advantage of a larger, and thus finer, scale for ranking risk, going up to as high as 10. It is derived from the Tables, but is more nuanced. You can see a sample at Tango’s site.
But in fact what one wants to set the “crisis” level at is immaterial to the gravamen of this discussion. The point is that the firemen are reserved for cases of a struggling pitcher in a crisis, however defined; when they come in, they remain in only till either the inning ends or the crisis dissolves (which can happen if the reliever gives up a big hit). After a fireman has been and gone, the usual approach would be to just bring on the next scheduled pitcher and go on from there; what that means, though, is that the finisher will almost surely be done well before the game ends.
Since the finisher will almost always be done before the game ends, what does one do when he is done? That depends. If the game is not quite close, one would probably bring in the swing man to finish it. Or, if one (or both) of the firemen have not had much work lately and the game is close to over, let one of them wrap it up. If the game turns into an extra-inning extravaganza and the swing man has been used up, one would have to dip into the pool of other short men (the openers and finishers) or even that of the primaries. No one can claim that any pitching-use program will eliminate all difficult decisions, or work splendidly no matter what strangenesses come along. But that approach, 9-18-9 plus, is quite likely the best available.
(That probably means that if teams started using it, or something much like it, the Commissioner would ban it, on his well-established policy of “No thinking allowed in baseball!”)
Some further notes: that approach does not use—and indeed cannot really tolerate—pitching changes made more or less purely on handedness: no taking out a pitcher doing well just because the next batter is of the “wrong” handedness. While pitcher platoon differentials tend to be wider than those for batters, unless the current pitcher’s platoon differential is huge (in which case why is he on your staff?), the risk of replacing a man pitching well with a new man over-balances the platoon differential.
Finally, we can note that experiments with various sorts of such unusual programs have been tried at the minor-league level with no reported problems.
About “Closers”
Note that in that scheme the stupid “closer” role has disappeared. Analysts have been consistently arguing for years that using probably your best pitcher just to pitch one clean inning to finish a game is an absurd waste of his talent. Consider that over 70% of all innings pitched are scoreless (72% in the period 2015 – 2020); that implies that a perfectly average pitcher should be able to get a “Save” at least 72% of the time (“at least” because in another 15% of innings only one run is yielded, and a “closer” will sometimes come in with a two- or three- run lead). The folly of using your best pitcher in a situation where a perfectly average pitcher will very often suffice grew out of letting a stat control game usage. If Jerome Holtzman had never dreamt up and argued for the “Save”, we could have been spared these many years of mis-used pitching talent. The time to bring in your best is when the game situation is at its worst. That does not seem like rocket science.
The 2024 MLB average Saves rate was 64%; that is startling. If we look at ESPN’s list of the “best” closers, by team, in 2024 (“best” by number of Saves, not Save percentage), we see Saves rates from 97.1% down to 64.3%—but only two are below that 72%. Looking closer though, we find that those 30 top relievers accounted collectively for only 46% of Save opportunities. That means that 54% of Save opportunities were handled by someone not that team’s best closer—and those someones have a collective Save percentage of…49.1%. Wow. (Mind, many “Save Opportunities” are not starting a clean top of the 9th: often the reliever who will get the Save or Blown Save stat has come in with runners left on base by a previous pitcher, and on average about one-third of inherited runners score.)
So we see that teams have primary closers who almost all do well, some quite well, but then by and large leave the rest of closing to what seem to be pretty awful pitchers. Remember that an average pitcher should get an at least 72% success rate, then look at that 49%. It is weird, to say the least. Clearly, though, the “closer” role is a lurching zombie.
) in the upper right of this page.