Baseball is even more quantifiable than football, another game with distinct starting and stopping points. In football, the difference between each possible outcome is relatively small. A team can gain no yards on a play, or one yard, or two yards, or three yards, and so on, all the way up to however many yards are needed for a touchdown. In baseball, by contrast, each at-bat either does or does not result in an out, does or does not change the score, and does or does not alter the position of base runners. Rephrased, baseball's outcomes are more dichotomous while football's are more continuous – with the consequence that baseball stats are more predictive of the game's result. If one team is making outs and the other team isn't, it's pretty easy to predict who's going to win, but if one team gains five yards per play and the other team four yards per play, either team could end up victorious.
All of this is theoretical support of something we've seen in full relief over the past 20 years (and, soon, in a movie starring Brad Pitt), namely that statistical analysis can prove a powerful force in shaping baseball thinking. A decade ago, teams learned to value the walk and cut down on steal and bunt attempts. A few years ago, teams learned to consider a defender's range in addition to the number of errors he makes, with Derek Jeter (few errors, really poor range and a Gold Glover nonetheless) the classic example of an overrated defender.
That managers still steal and bunt nearly as much as ever, that players strike out swinging for the fences nearly as frequently as ever, never mind the walks, and that errors are still the most common way to rate defenders shows that institutional resistance to innovation in baseball has been exceptionally strong, for reasons outside the scope of this article. Still, the mathematical facts remain mathematical facts: data have shown several nuggets of baseball "conventional wisdom" to be inefficient, and were the typical MLB team to retain the same on-field talent, but just employ more efficient strategy, it would win several more games over the course of a season.
All of this was on my mind as I thought about Stanford baseball, sitting on the playoff bubble with the postseason rapidly nearing. If any baseball program at any level should be able to think of a novel, perhaps counterintuitive idea for its benefit, surely it should be Stanford, one of the country's most innovative universities, and one located in the heart of Silicon Valley. A few more runs saved or gained over the course of the season would have resulted in a few more wins, which in turn would mean Stanford would be bouncing in and out of the Top 15, not the Top 25, and we'd all be breathing easier.
So I thought, and I thought, and then I thought some more. Finally, a thought popped into my head, and I double-checked the numbers to make sure my intuition was right. It was. If only I had a platform via which I could let the Stanford community know of my innovation. Oh yeah, I've worked here for a little while. So without further ado, here's my avant-garde, revolutionary idea that could completely transform baseball: No more starting pitchers.
Okay, I admit, those four words are kind of a letdown after all the hype. Big deal, who cares who pitches when? After all, the team with the better pitchers is going to win more often than not.
Fair enough, but look at data from this MLB season, courtesy Baseball-reference.com. We can summarize the main findings in a chart:
|Starting Pitchers||Opp. Batting Average|
|First time through order||.248|
|Second time through order||.265|
|Third time through order||.268|
|Fourth-plus time through order||.272|
Simply put, the longer starting pitchers go, the worse they get. We see the same effect, only at a more dramatic scale, with relievers.
|Relievers||Opp. Batting Average|
|First time through order||.251|
|Second time through order||.282|
|Third-plus time through order||.300|
With two months of games times however many Intuitively, this makes perfect sense. Do anything athletic as hard as you can. Bench-press as much as you can, run a mile at top speed, you name it. You do okay. Now do it another 100 times. Your first rep is going to be better than your 100th. And it's hardly shocking that we're seeing the same thing in baseball. (I can hear it already: throwing a baseball is way easier than sprinting a mile, that's not a fair comparison. True. That's why a world-class athletes can do maybe 10 mile repeats but throw 100 pitches. But the same principle applies.)
Take this logic a step further now. Say you knew you were going to have to bench press as much as you could 10 times. Don't you think you would pace yourself so that your first few reps weren't as strong as they could be if you knew you were just doing two or three reps? Absolutely. In baseball, that's the difference between a starter and a reliever. Starters have to pace themselves, and so they're generally topping out in the low-to-mid 90's at the professional level, while relievers get to go all out, so some relievers can throw 100. As a result, a star reliever has better stats than a star pitcher – your local team's closer almost surely has one of the lowest ERAs on the team. (That the average reliever does worse than the average starter doesn't contradict this, but just shows that teams will hide so-so pitchers in middle relief, whereas their starters are all among the best arms on the team.)
Therefore, my radical system is this: convert every pitcher to a closer who goes no more than three innings at a time. Have your worse pitchers have more one-inning outings and your best pitchers more three-inning outings (and more frequent appearances), and with a little planning you can achieve the same overall distribution of innings that you have now – with star starters and closers accounting for a majority of innings.
I realize this strategy will be mocked as pitching every game like its an All-Star game and goes against 100 years of history, but baseball conventional wisdom has been splendidly wrong before, the stats support this argument, as, again, does common sense: Imagine we're racing in a mile relay, four laps around the track. Your team has the same guy running three laps or until he's absolutely dead tired, then giving it to the anchor/closer, while my team has a different sprinter each lap. Even if your runners are a little better than mine, I'm liking my odds.
There are several additional benefits to such a strategy. We would expect our starters' stats to improve not just to current starters' stats the first time through the order, but even further, to the stats of top relievers, because our former starters are no longer pacing themselves, but can go all out. In the National League, a pitcher would almost never have to bat, resulting in an instant .100 improvement in the No. 9 hitter. (Also, there is no statistical evidence whatsoever supporting the conventional wisdom that pitchers need an inning or two to "settle down", which would throw a major wrench into this plan.)
Admittedly, free agents might be skeptical of signing with your team at first, and the old guard would give you a tongue lashing in baseball circles and in the media. And a pitcher used to seven-inning starts on four days' rest would need a little time to convert to a reliever-like role, though even today we see that plenty of starting pitchers convert just fine to closers. And, yes, pitchers would have fewer off-days, but if a guy can throw 100 pitches in a day, he can throw 33 pitches three days in a row. Again, common sense. Would you rather run four miles today, tomorrow and the day after, or go and run 12 miles right now? Exactly.
So, now that we've presented a new strategy and defended it both intuitively and with statistics, our final question is one of the estimated effect: how much of a difference could switching to "all closers" make? Well, let's look at Stanford baseball. At press time, the Card have outscored their opponents 313-268, which, over the long haul, results in an expected win percentage of .577. (As Carlo Salcedo has pointed out, Stanford has been pretty lucky this year – their great record in one- and two-run games reflects that.) I think this strategy would probably have more of an effect at the college level, as college pitchers aren't in as good of shape as pro pitchers, but conservatively, let's assume that converting starters to closers results in 10 percent fewer hits, or a drop from .267 to .240, about the scope of the falloff we see as starting and relief pitchers continue into their outings. For the sake of simplicity, we'll assume that translated into 10% fewer runs, giving Stanford a 313-241 runs tally. That, in turn, translates into a .627 win percentage, or .050 better than present. Given that Stanford has played 50 games, with average luck, 2.5 current losses would be wins, and Card would be not 30-21 but 33-18 or 32-19. That's not a lot, but Stanford would be that much higher up the Pac-10 standings and I'd be sleeping that much better thinking about their potential playoff path. I think Coach Marquess would too.
So there's my all-closers strategy. 9, I know you didn't use it during the regular season, but, hey, there's always the playoffs.
Are you fully subscribed to The Bootleg? If not, then you are missing out on all the top Cardinal coverage we provide daily on our award-winning website. Sign up today for the biggest and best in Stanford sports coverage with TheBootleg.com (sign-up)!