I’m setting up my own league based on linear weights for the first time this season. Last year I was trying to think of a scoring system that would accurately represent things, yet for some reason I didn’t think of LW – I just sat there trying to crank out my own values. Oh well – I had some down time, and it was my first year really doing fantasy baseball, so it’s probably just as well that I got a little experience before trying to do my own (I’d done fantasy baseball before, but those usually consisted of picking players and not paying much attention until the season ended)
One thing I’m having trouble with now is what to do about pitchers. The only LW info I found right off hand is for batters. Currently, I’m just assigning them most of the same categories as hitters, just with inverse points (an IP = 3 PA. An event gives a batter X points, the same event will give a pitcher -X points). Batters get credit for SB and penalized for CS, and I gave pitchers a little bit of points for Ks to make up for the fact that pitchers seemed worth so much less even though batters don’t get penalized for Ks in my league.
Also, I’m unsure what to do about relievers. They don’t get any credit for pitching in high leverage situations, but giving points for Saves/Holds doesn’t feel right even if it’s only a small amount.
I look forward to reading how you decided to score pitchers.
Right, you can’t use lwts for pitchers because the pitcher getting the highest point total would be the one who gives up the most runs! :)
I will write about the pitcher points next week. I probably should leave you in suspense, but you can get a preview if you look at the support page for ottoneu: http://ottoneu.fangraphs.com/support
Short story is that it’s built on FIP, with additional credit for innings that makes them match up to hitters. And then there are saves and holds…which I’ve concluded are a necessary evil, unfortunately. More on it next week.
At least one other lw league spawned out of the book blog last year, which I joined. We had a lot of success ignoring SV/H using a 5 RP format with an innings cap (I think it was 1450 but I don’t remember). My relievers were the most valuable part of my team per unit (IP or PA). Relief also generally had the highest ‘reward’ in net points between the elite and replacement level aside from catcher and shortstop.
@Brad, that does require an IP cap. And that rules out head-2-head points leagues, which is what this system was originally built to use.
But yeah, I can see that working. Relievers definitely have better FP/IP rates in our system, even without the saves. Our game at FanGraphs does use a 1500 inning max, so this could be something that we need to look at. I hit 1800+ innings last year in another league using lots of two-start starters and SP/RP players, so it’s an issue.
Very cool way to see wRC compile as the season progresses. I may try out this system this year but I’ll be sure not to use the aforementioned Izturis no matter how shallow SS gets this year.
Comment by Half Full — February 18, 2011 @ 12:02 pm
As a fangraphs frequenter and fantasy enthusiast, I actually prefer more classic, dummy fantasy scoring formats. I find that there are two problems with the type of scoring system presented; it’s too predictable, and it takes out your ability to ‘game’ the system. Getting lucky or unlucky on a pitchers ERA is fun. You can still use advanced stats to make good guesses. I also like the idea that i can snag all the homer hitters and then use that for huge trade leverage. Its a game.
So question, I may sound ignorant asking. Where is the accountability for speed, other than steals? Lets say Billy Butler and Carl Crawford are on second for their respective teams. Someone hits a frozen rope to the corner of right field. Crawford goes home, while Butler huffs and puffs to get on third. Unless I am missing something they would both get the same amount of points, which does not account for their actual worth. I am probably missing something.
You are correct that there is no explicit accounting for speed in this system, beyond the rewards for SB/CS and triples.
However, the value of baserunning is typically pretty small over the course of the season: based on EqBRR, the league leaders in baserunning are usually around +7 or +8 runs per season once you remove SB’s. It matters, but it’s not huge, and I think our game still does a nice job of reflecting real hitting value. wRC/wOBA/WAR at FanGraphs does not reflect that level of baserunning either.
One last thing–because most of the premium positions have a good number of speedy players, they get a bit of a value boost simply due to scarcity (and while ours gane doesn’t do this, some leagues do use LF/CF/RF to further give CF’s a boost in value).
Basically, I’m using PA since Yahoo (and ESPN) don’t allow batters to get penalized for outs, which is really what LW counts. So I just deduct for all PA instead, then adjust all values of positive events by a corresponding amount.
With pitchers, 1 IP = 3 outs, so I just say IP = 3PA and then I don’t adjust the other events.
I should have clarified that earlier, but didnt’ want to make my post too long.
Nice work Justin. I do believe this is the best way to quantify all of a player’s tangible performance factors into a fantasy output. I would be interested in throwing my hat into your Ottoneu league when it becomes available. By the way vlookup () and bubble charts rock!
Comment by darrenmaybee — February 22, 2011 @ 1:49 am
Depends on what you’re after. If you like the basic idea of a fantasy game, especially a points-based game, but want a scoring system that matches up to reality, then this is a good option. It’s a simple game: pick your players, and watch their performances accumulate. It’s no more time consuming to play than normal fantasy, and you can do so in normal format leagues (or our cool year-round auction leagues at fangraphs).
If you want another level of realism, you probably do have to venture into the world of sims. But with it, you get added complexity, time commitments, etc. And, at a certain point, the fate of your team lies not directly in the performances of your players, but in the algorithm that runs the sim…
Very interesting though if the points are based off each at bat and the number of runs that an outcome would produce for a team (single = .463 runs), then shouldn’t a strikeout for a batter be an additional subtraction of points? It was my understanding that a strikeout results in less runs over the course of a season than a out on a ball put into play. Maybe this difference is just too small?
An out (by strikeout or other out) does result in -1 points, because that results in an AB in almost all cases (except for SF’s, which get a pass here–whether that’s appropriate or not is an open question).
Strikeouts do result in fewer runs than a normal ball in play, but it’s an extremely small difference: in this study (http://www.tangotiger.net/bsrexpl.html), it’s -0.111 runs instead of -0.098. Rounding to the hundredth place, and then multiplying by ten gets us to -1.0 points per strikeout and -1.0 points per out. Part of the reason that the difference isn’t greater is that balls hit into play result in double plays a lot more often than a strikeout, and so strikeouts can sometimes be preferable to a ground ball. Obviously, that depends on the situation, but on average, they really aren’t much worse.