Rating the Playoff Teams
The Cubs won the third-most games in baseball. In the first round — a round some people don’t even consider the playoffs — they eliminated the team that won the second-most games in baseball. Just Tuesday, in the other first round, they eliminated the team that won the very most games in baseball. Very good accomplishment! Exciting times for the Cubs. It makes it worth wondering: who’s really the best at the moment?
We know that the playoffs don’t always ultimately crown the best team in baseball. There’s just way too much room for randomness, and sometimes superior teams do get toppled. Really, it’s part of the fun. But at the same time, that “best team” label is more complicated than it might appear. Because: when? If you’re trying to figure out the best team, do you mean the best team overall, or the best team at the moment, or what? The Cardinals were just the only team in baseball to win 100 games. They also went into the playoffs without, say, Carlos Martinez, or a healthy Yadier Molina. So what should one make of the playoff Cardinals, relative to the overall regular-season Cardinals?
This is at risk of going too long. I tried to rate the playoff teams. And I mean the teams as they’re built today. I tried to rate the best baseball teams, right now.
What I did, basically, just re-created what’s already going on on our depth-charts page. But one of the big differences: I used playing time as it’s actually been given out. For these purposes, I considered just the eight teams in the Division Series round, because everyone in this round has played four games, so there’s some playing-time consistency. I understand that the Cardinals are dead, as of yesterday, but I still decided to include them.
The long and short of it: for position players and pitchers, I calculated WAR per plate appearance, weighted by postseason playing time. I ran two sets of calculations — one using actual 2015 performance, and one using updated projected performance. When I added the numbers together for position players and pitchers, I arrived at a team rating. The units are weird and we’re looking at some decimals, but the actual rating isn’t important; what matters are the relative ratings, between teams.
I have no sense of how well I’ve explained this. Results might help. First, here are the team ratings, based on how the players actually did this past season:
The best playoff team, by this: the Dodgers, who tomorrow will play an elimination game against what this says is the second-best playoff team. Meanwhile, by the time you’re reading this, the third-best playoff team will be underway in an elimination game against the worst playoff team. Before going into a little more detail, here’s the same plot, but using player projections instead:
You get the same top three, in the same order. You also again get the Rangers in last. The Cubs and Cardinals are almost exactly tied. The Astros fall from fifth to seventh, but the gap between seventh and eighth, here, is bigger than the gap between second and seventh.
An obvious point: to some extent this is going to overrate the Dodgers, because they just started Clayton Kershaw on short rest, messing with the playing-time distribution. Kershaw isn’t actually going to account for 38% of the Dodgers’ total postseason plate appearances. Depending on how you adjust for that, the Dodgers fall back some, but it seems they remain in first place, at least by a hair. And Kershaw apparently can go on short rest and be effective. The Dodgers look like the best, but they’re probably not much better than the Mets or the Blue Jays.
As for everything else, it just comes down to whether you think WAR does an adequate job of capturing the players and teams. If you don’t agree with WAR, you won’t agree with this, and that’s fine. There’s no guarantee WAR has a good built-in league adjustment. It doesn’t adjust for opponent quality. Maybe some of the adjustments it does make are wrong. Clearly, it’s far from perfect on the defensive side. I’m not going to lie to you about WAR’s accuracy, and I used it here because it’s the best we’ve got, and it’s super convenient. These ratings aren’t definitive; they’re just for fun.
Probably, there are Rangers fans who are upset. No one wants to be told their favorite team is the worst in a group. Plenty of Rangers fans have accused FanGraphs of a bias over the course of the season, and there’s nothing for me to do about that but try to assure you this isn’t an attempt to push some sort of agenda. The ratings above are just where the numbers led me. I didn’t manually adjust anything. It’s absolutely within the realm of possibility that as far as the Rangers are concerned, WAR is just clueless, and that’s why the team is still alive today. I don’t have all the answers, and that’s what keeps baseball analysis so addictive. There’s always something new to try to explain, even if you’ve gotten a bunch of things wrong in the past.
Like Sam Dyson. I love Sam Dyson. WAR, less so. It’s a small factor, but it’s something.
I won’t keep you. That’s what I wound up with. That’s my attempt at rating the playoff teams, right now. Now to sit back and find out who actually plays the best over a three-game sample of first-round fifth games. That’s the more fun part.
Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.


How are the 2015 Base Run Champions only #5?
Because Jeff Sullivan is the most biased writer in the history of writing.
Jeff Sullivan is the Ann Coulter of baseball!
Yes, because stripping out sequencing is exactly the smart thing to do. Hey, we don’t understand how much of it is luck vs real so lets pretend it doesn’t exist.
Yes, because taking all sequencing is exactly the smart thing to do. Hey, we don’t understand how much of it is luck vs real so let’s pretend it is the sole cause for all sequencing.
There isn’t a single employee of this site that believes that bs. We mock what we do not understand. Sequencing is at the heart of the next breakthrough in Sabermetrics.
Eh not really upset by this, I figured that the Rangers would be ranked the lowest because honestly we have been pretty lucky this year. I mean I’m very sure I’ve used the “World Series doesn’t decide who the best team is” argument before, so it’d be hypocritical of me to try and claim it now…especially when we’re in a do or die game right now.
As a UConn basketball fan, I know what you mean. Enjoy the ride, even if you didn’t take the sexiest path to get there.
As a Rangers fan, I am not offended. Hell, I’m as shocked as anyone that the Rangers are here. Anything moving forward is just gravy. Thank goodness for fortunate sequencing :-).
Pretty sure Rangers fans have been accusing me of all the bias, not Fangraphs as a whole.
#Sequencing
Yeah, some Rangers fans have taken issue with some of the stuff that Cameron has been saying in the chats over the course of the season. 2 big things, IIRC. One Cameron saying that the Rangers had no chance to compete in the next few years despite having one of the best 3 or 4 farm systems in the league (before the Cole Hamels trade). The other has been his stubborn refusal to acknowledge the 2nd half pitching staff overhaul and +43 2nd half run differential, and his insistence that the season run differential is more reflective of the current roster.
Does that left axis say the Dodgers project 2x better per PA than the Rangers? In WAR/PA, which is essentially proportional to runs per PA?
I did this too. Got casually the same results, but the Mets were actually first.
https://replacementlevel.wordpress.com/2015/10/05/what-if-the-playoffs-made-sense-a-postseason-preview/
So what you’re saying is that Thursday’s game in LA is the real World Series game 7.
So, basically the Cardinals without Martinez and Wainwright as a starter are equal to the Cubs at full strength? That sounds about right … although it might be selling the Cubs short a bit.
The series seemed to reflect that actually. When you’re featuring John Lackey on short rest (and it’s not 2002) you’re fighting an uphill battle.
———————————————-
My mind is strugglin a bit with translating the graph into what it actually means. I’m trying to determine how significant the differences between the teams really are and how that would equate into “runs”.
If I use a number like “37 PA” for every team, does that translate into the Dodgers being 2.4 runs a game ‘better’ than the Rangers?
Lookin at Cubs-Crads because that’s my area of interest … there’s a BIG difference between fWAR and bWAR.
fWAR — StL = 44.9; CHC = 51.9 — CHC + 7.0
bWAR — StL = 50.2; CHC = 41.8 — StL + 8.4
I didn’t scale for playoff playing time, etc … just seemed strange to have such a big disparity between a number that is intended to measure the same things for the same purpose.
Also interesting is that if you take fWAR/2 + bWAR/2 = Total … you get StL 47.55 and CHC 46.85
http://www.fangraphs.com/library/war/differences-fwar-rwar/
It’s pretty common to see big differences between fWAR and rWAR for pitchers, because fWAR is based on FIP and rWAR is based on actual runs allowed. This year the Cardinals’ pitching staff allowed a lot fewer runs than FIP would have predicted, mostly for reasons of sequencing and great performance with RISP that most likely aren’t repeatable based on history. I’m not surprised that the 2015 Cardinals look considerably better by rWAR than fWAR.
Yes, losing Martinez sucked. However, I can’t really say Wainwright factors here when this team actually ADDED him to end the season (even if it was just as a reliever). I mean, the Cubs played without Soler for a large chunk (and he’s been killing it now), and Castro’s looked like a changed man since the position swap. Molina’s injury certainly hurt as well, though the degree to which is hard to measure defensively, while his offense this season wasn’t something to get hyped about.
Still, the Cubs are moving on, and I’m quite concerned about how they handle it. They won yesterday with a starter whose second-half FIP is 4.54. They sent out 5 relievers, four of whom were failed starters (Grimm, Cahill, Wood, Richard). They also threw out the mess known as Fernando Rodney, and their “dominant” relievers are a Rule 5 pick closer (Rondon) and a setup guy with no consistency who was a throw-in in the Feldman-Arrieta trade (Strop).
Re: Wainwright – I think your parenthetical (“even if it was just as a reliever”) makes all the difference, no?
You can go a lot farther on the Cardinals’ problems: I don’t think anyone trusted Wacha for more than 3-4 innings (bad 2nd half, threw lots of innings this year); Holliday was awful coming off his injury; Grichuck’s arm; Garcia’s illness; Piscotty’s concussion (not that it seemed to slow him down). Still, the Cubs torched their bullpen, which looked like their one mild advantage. The Cubs just have a deep lineup, especially with Fowler and Castro hitting around their expected talent level, which neither did in the first half.
#Sequence this Fangraphs
/points to crotch
On average, when you point to yourself, you’ll be pointing at your bellybutton. Therefore, if we remove luck from your gesticulation, you’re not particularly offensive.
Yes, because stripping out sequencing is exactly the smart thing to do. Hey, we don’t understand how much of it is luck vs real so lets pretend it doesn’t exist.
Actually, I’m pretty sure it’s mostly luck.
If it’s a skill then SOMEONE should demonstrate it consistently for longer than you’d get from random luck.
And then correlation studies on first half vs. second half, or year vs. year, should show such a skill relatively trivially.
In this case absence of evidence is in fact evidence of absence, because “sequencing is mostly skill” is a hypothesis with testable consequences. And those consequences largely fail to show up where they’d be expected.
There appears to be a small skill element to clutch for hitters, almost too small to effectively measure on individual hitters, but real.
There is presumably a skill in bullpen management.
Other than that? Why do you think sequencing is a skill on a team level? Evidence isn’t that hard to check for, pick a set of years and a measure of sequencing production (baseruns minus actual runs for example), plot years in pairs, and do a fit. Things like that pretty consistently come out with a fairly flat best fit with r^2 values of 0.00 or so.
That looks like luck. Heck, I’ve tested commercial pseudo-random number generators and gotten results that looked less like luck (admittedly on much larger samples, but still).
My eyes are up here!
. . . and batting ninth
I have no problem with this.
Me too.
Hahahaha
These numbers mean something, right?
“Like Sam Dyson. I love Sam Dyson. WAR, less so.”
WAR 1 Jeff 0
whoopsadoodle
Jose Bautista loves Sam Dyson even more than WAR.
WAR 0 Jeff 1
(I would still want him (back) on the Blue Jays…)
oops, meant WAR/Angry Bautista 1
Bautista 1, Dyson + the laws of physics 0
only thing i might want to try different in this evaluation is use a 50/50 ra9/fip WAR number for the pitchers.
i did a less sophisticated Active Roster War total using the 50/50 split before the playoffs and i believe the ranking came out:
1. TOR
2. CHC
3. HOU
4. LAD
5. NYM
6. STL
7. KC
8. TEX
this is the right way to evaluate things, at least for Toronto, since they have Dickey and Buerhle in the fold
Preview to the 2016 Fangraphs predictions: Royals finish last in the American League with a 10-152 record.
Preview to the 2017 Fangraphs predictions: Royals finish last in the American League with a 5-157 record.
I am perfectly fine with Fangraphs under-rating/ disrespecting/ ignoring my favorite team. Fangraphs’ odds of the Rangers even making the post-season were microscopic, so keep it goin’, Dave and Jeff. Pick us last again next year. Please!
Your logic excludes injuries to key players on 25-man. For Jays, loss of Cecil and Loup’s impromptu vacation left Jays without a LHP in pen. Cecil has been a hell of a weapon in the 2nd half (.135 avg) which no doubt bumps his 1.4 season WAR by quite a bit.
Curious where the Pirates would have ended up.
I have LA and the Mets as the 2 worst teams. Toronto and Cubs the 2 best. I think thats because I weight 2nd half performance more highly, and take into accounts SOS