More About Bad Teams and Saves
There were some questions about the best way to see if bad teams earn saves as well as good teams — so here are hopefully some answers.
Should we have used save opportunities as our metric? Better teams should make more save opportunities for our closers, and using opportunities removes the quality of the closer from the equation, right? One problem, before we run the numbers, is that save opportunities exist in the seventh and eighth innings, but the closer is rarely brought into those innings any more. So there will be some save opps that won’t ever turn into saves for our bad team closer… but let’s see what happens:
Actually, this makes it look like even more of a crapshoot.
Winning percentage only explains about 8.6% of the variance in save opportunities. Less than 10%! Bad teams get about the same amount of chances as good teams! Look at how flat that line is.
Okay, so is it that the good teams have good closers that also make the team good? In other words, what if we plot save percentage against winning percentage?
Okay, so here we have something. 20.5% of save percentage is explained by winning percentage. Good teams do have slightly better closers. But that doesn’t help us a ton in the case of a good closer on a bad team.
So, now we’re getting closer to this bad team / good team closer situation. Good teams (marginally) produce more saves, but if you break it down into save chances and save percentage, that effect gets smaller.
The fact remains: there are plenty of good closers on bad teams, and all you have to do to truly reflect their value is knock a few saves off the top of their projection — but not many. Certainly no more than one fifth. We still have lefty closers and bad bullpens to tackle in this mini-series.
With a phone full of pictures of pitchers' fingers, strange beers, and his two toddler sons, Eno Sarris can be found at the ballpark or a brewery most days. Read him here, writing about the A's or Giants at The Athletic, or about beer at October. Follow him on Twitter @enosarris if you can handle the sandwiches and inanity.
Scumbag Me: am a crappy pitcher, leading the AL in saves.
any chance of tackling vulture wins?
this is a terrible approach to save speculation…
fantasy players looking for saves should aim for teams with bad offenses. they may not win as often, so their are less leads to protect, but conversely they have fewer blowouts that are detrimental to save totals as well.
to put it another way, even the worst teams win 60/70 games… but of those 65 games, a larger percentage of those wins are saves
This is a whole series. Will get to your idea too.
Saves (for the most part) come in games where the winning team defeats the losing team by a margin of 1-3 runs.
The margins of victory/defeat for a given team will conform to a particular distribution. Because you are subtracting one random variable (runs scored) from another (runs allowed), the resulting distribution is almost certain to resemble a discrete Gaussian. For a .500 ballclub, it will probably be symmetric and centered right around zero (although zero is an impossible outcome in this discrete distribution).
The team that gives you the most save opportunities is likely to be the one whose margin of victory/defeat distribution has the most mass at +1, +2, +3 on this distribution. It’s not obvious to me which sorts of teams will produce that best, but my intuition is that it will be winning, but low-scoring teams. Consider two teams that, on average, outscores their respective opponents by 1 run. But Team A scores 6 runs a game against 5, and Team B scores 3 runs a game against 2. It seems to me that the distribution of margins of victory for Team B will be more tightly centered than for Team A, resulting in more save opportunities.
I guess I could rephrase my hypothesis more concisely by saying that lower scores (runs for and runs against) lead to tighter ballgames, and tighter ballgames lead to more save opportunities. Here’s the plot I’d like to see: a scatterplot (like one that you’ve created, with Saves on the y-axis and team Winning % on the x-axis) with separate colors for high-scoring teams and for low-scoring teams. Plot the trendline separately for each group, and see if they look any different. Maybe we’ll learn something.
I’m getting a little confused I guess. The purpose of the 1st article was to say that you can get a good amount of saves, aka “a good closer”, from a bad team.
So now we’re talking about finding a good closer from a team with a good closer? Seems circular to me.
The fact of the matter is teams with good setup men tend to get more save opportunities for their closer. I’d say run a graph showing bullpen ERA vs saves, but I have a feeling we all already know what it looks like.
Yeah the last one came back around in a weird way, not so useful. But the first one makes the point best — good teams don’t make many more save opportunities than bad teams.
I agree. The best way to get a good number of saves is to 1st have a player that is a good pitcher and has the job of getting saves.
Good pitcher + not closer = no saves (or very few)
Bad pitcher + closer = some saves (until he loses the job)
Good pitcher + closer = a lot of saves (with job security)
I think you need to narrow the range of years you are using to the period that better reflects the current pitching environment (regarding complete games and bullpen usage). Spot checking the stats on mlb.com, I find that from 2000-2011, no team finished with less 40 save opportunities. In 1974 (the first year you use), the highest number of save opportunities was 45 (with a low of 20).
Can you isolote “save opportunities in the 9th inning” from total save opportunities, and compare those to winning percentage? That would seem to supply the information that is most helpful to fantasy owners.