How Pitching is Scored in ottoneu FanGraphs Points Leagues
Interested in playing in my ottoneu league, using this scoring system? See the bottom of this post for info on how you can join!
Last week we looked at hitter FanGraphs Points, which are built on linear weights. A complementary pitching points system was more difficult to devise: you can’t just use linear weights, because then you’d be rewarding the pitcher who gave up the most runs!
Therefore, it was necessary to develop a system that had some objective basis, but would a) reward good pitchers for being better than bad pitchers, and b) have point totals that would correspond to those of hitters. And, just as our hitting system doesn’t track statistics that are substantially affected by other players (like RBI or runs), our pitching system should reward pitchers for what they do as opposed to what their teammates (pitchers and fielders) do. That means no wins, runs allowed…or hits.
After a combination of theory, intuition, and trial & error, the following system was developed:
IP: +5
K: +2
BB: -3
HBP: -3
HR: -13
SV: +5
HLD: +4
The first number is just tracking innings. To some degree, this number was just chosen because it works: pitchers scale closely with hitters, both per season and per week. But it has some theoretical backing as well: since we are using an absolute-runs baseline in our points system, an average pitcher should earn a half of a win for each nine inning game he pitches (i.e. he’ll win half of his games if he is average). A half win corresponds to ~5 runs, which converted to points (x 10) and divided across 9 innings works out to 5 points per inning.
Next are the events in those innings: strikeouts, walks, hit-by-pitches, and home runs. If you’re a number junky, you probably recognize those values: they’re the same coefficients used in FIP. FIP is a very successful statistic for evaluating pitching performances, and it focuses exclusively on statistics over which a pitcher has complete control. If you follow that link, you’ll see Kincaid’s superb demonstration that these values ultimately are derived from linear weights. And, those values work well in this context too.
Finally…there are saves and holds. Wait, saves and holds? For a sabermetrix points system?!? Yes. We need a way to recognize the value of relievers that throw in high leverage situations. In FanGraphs WAR, for example, relievers are given partial credit for the high leverage of their innings in recognition of the extra impact those innings have on the outcome of games. Adding points for saves and holds mimics this, albeit imperfectly: we effectively give closers double-credit for those innings in which they record a save. The reason that holds are set so close to the point values for saves is that it prevents the save from ruling all: Raphael Soriano loses some value by not being the Yankee closer, but it’s not so much to tank his value compared to other relievers. When combined with a 1500 innings cap and 5 RP slots, relievers, not just closers, have a lot value in ottoneu.
Here are the top-10 pitchers based total FanGraphs Points from 2010:
Name | Pts | IP | FIP |
Roy Halladay | 1271 | 250.7 | 3.12 |
Felix Hernandez | 1257 | 249.7 | 3.15 |
Adam Wainwright | 1203 | 230.3 | 2.95 |
Cliff Lee | 1167 | 212.3 | 2.68 |
Justin Verlander | 1147 | 224.3 | 3.09 |
Jered Weaver | 1127 | 224.3 | 3.18 |
Ubaldo Jimenez | 1103 | 221.7 | 3.13 |
CC Sabathia | 1079 | 237.7 | 3.58 |
Tim Lincecum | 1047 | 212.3 | 3.17 |
Zack Greinke | 1042 | 220.0 | 3.45 |
Josh Johnson | 1040 | 183.7 | 2.50 |
Francisco Liriano | 1039 | 191.7 | 2.78 |
Not a bad list, right? The two reigning Cy Young winners really were in a class by themselves.
Now for the fun part: I’m running an ottoneu league based on this scoring system, and would be happy to have it filled with readers. Our draft begins March 8th at 8:30pm EST, and will be held on consecutive nights until completed. So, if you’d like in, please, in the comments below, name one player who you think will have a massive difference in value using this system compared to a traditional roto category system, and explain why. Some of these players may be discussed in next week’s article! I can’t give you out free teams, unfortunately, but I’ll gladly send an invite to my league to those who respond. Remember, teams are just $9.99 through the end of this month!
Update: Let’s just open this up to the public. The draft time (March 8th, 8:30pm EST, and then on consecutive nights at about that time) is pretty inflexible, so I’ll take anyone who is interested and can make that time. Here’s a link for the league: Justin Merry (RotoGraphs)
Update #2: The league is full! Thanks to everyone for your interest.
Justin is a lifelong Reds fan, and first played fantasy baseball on Prodigy with a 2400 baud modem. His favorite Excel function is the vlookup(). You can find him on twitter @jinazreds, even though he no longer lives in AZ.
What were the points totals for the best relief pitchers?
I don’t have SP’s broken down from RP’s handy, but here are all pitchers:
http://bit.ly/i1f9Dh
Quick scan, I see:
Wilson
Marmol
Bell
Wagner
Belisle
Soria
Gregerson
Keep in mind that we use an innings cap in the FanGraphs game. Even though relievers get far fewer points than starters, they tend to have much higher points per IP (this is aided by the saves/holds, but it’s true even without them). That can make them extremely valuable, since you can’t just spam innings to earn pitching points all season.
-j
Thanks for the numbers.
I see what you mean.
Points/IP
Roy Halladay 5.084
Brian Wilson 9.089
Hi Justin,
I have been scouring for a more thoughtful and accurate way of defining stats as I create my own points league over at Yahoo (first timer!). I really appreciate the method described here and think it makes a ton of sense. It also seems as if one thing is missing here, which is defensive statistics. If the aim of these stats is to try to match actual specific value of each player then doesnt the system need in some way to account for their value in preventing runs? In practical terms, it seems that without including defensive stat categories, the value of all defensively geared players goes down (in terms of draft value), and priority will naturally be placed on players who can hit, regardless of their defensive value.
As a San Francisco Giants fan, I am a big fan of plenty of players who may lack relative value in runs created but shine when it comes to runs prevented.
Do i have it all wrong? And if not, do you have advice on how i could create some defensive categories that would mesh well with the existing system you outline here?
Thank you!