Help me out-Billy Beane my fantasy baseball league!

October 28, 2009 7:06 AM

How do I appropriately valuate players in my competitive fantasy baseball league?

I just took over a horrendous team in a very competitive fantasy baseball league. I'm very familiar with sabermetric stats employed by real-life GMs, but it occurs to me that those same stats are not necessarily going to be helpful to me in fantasy baseball. (Especially because so many of the stats we score on are non-sabermetric in nature.)

To wit: real-life GMs use sabermetric stats to find players who avoid making outs (hitters) or get outs most efficiently (pitchers). In my league, I'm looking to maximize or minimize the cumulative results in specific statistical categories across my lineup.

To that end, I'm trying to put together my own valuation system for players, but it's been awhile since my econometrics and statistics courses in college. Can you help me put together an appropriate formula - or point me in the direction of how I should be analyzing/comparing the statistics - to properly valuate the players in my league?

The details: I'm in a weekly head-to-head rotisserie league based upon the following statistical categories:

HITTERS
Runs
Singles
Doubles
Triples
Home Runs
RBI
Walks
Strikeouts (lowest total wins this category)
Stolen Bases
Batting Average

PITCHERS
Walks (lowest total wins this category)
Strikeouts
Complete Games
Wins
Losses
Saves
Holds
ERA
WHIP (lowest total wins this category)
K/9

I'm not sure if any other parameters are needed, but any help you might be able to provide would be most appreciated.

posted by po822000 to sports, hobbies, & recreation (10 comments total)

Depending upon how you value your time, purchasing a list from one of the big FB sites would probably be easiest.

There are hundreds, shop around.
posted by Setec Astronomy at 8:02 AM on October 28

Well, I'm actually interested in piecing together the math behind it, and I know a lot of the fantasy sites aren't built to our stats (I mean, how ludicrous is it that Complete Games is a stat?)
posted by po822000 at 8:32 AM on October 28

-What are your keeper rules?
-What are your roster construction rules-how many start, how much of a bench do you have?
-How many teams are in your league? This will determine how many adequate players will be on the wavier wire.

-weekly head-to-head rotisserie league

Some clarification on this point would be helpful as well. Rotisserie and Head to Head are the two basic fantasy sports styles. "Head to Head" is style in which you compete against one other team in the league for a period of time (usually a week), and at the end of that period, categories are scored against that other team-sometimes you get a win or loss for the week, sometimes you get a win or loss per category. In head-to-head there are usually playoff weeks at the end of the year to determine the champion. "Rotisserie" used to be equivalent to "fantasy" but it has a technical meaning that people will assume that you are using: you compete against all teams all the time in all categories, with the highest ranked teams in each category at the end of the year getting the most points, and the team with the most points wins the league-there are no playoffs.

Head-to-Head is less popular in baseball that in other sports, because the variance in number of starts by your pitchers can easily swing results week to week. In HTH, you can be successful "tanking" certain categories-beefing up some categories at the expense of others, and winning by a little bit every week. In Rotisserie, you can't do that, usually-you need to have a balanced team to win.

I wouldn't pay for any fantasy content, ever-a ton depends on the factors above and what the rest of your roster looks like, and there's no magic formula. Sabermetric stats are good predictors of performance and lots of the important decisions are tactical rather than strategic.

But in general,

-Pitching performance is harder to predict than hitting performance, and pitchers are less healthy than hitters. Draft for hitting early.
-Look for categories where one result counts for multiple categories-players who do those things will be more valuable. For example, home runs are incredibly valuable in your league, because each HR counts for: R, HR, BA, RBI-guaranteed.
-Your batting categories are slanted heavily toward counting stats, so draft for health with your hitters-especially because hitter health is relatively more predictable than pitcher health. If Vlad Guerro is great for his usual two thirds season, that's less valuable than some healthy OF on the wavier wire in your league. Health is a skill.
-With pitching, follow the strikeouts-you get K/9 and K directly, and they are good predictors of success in other categories.
-Complete Games is an insane category and I recommend ignoring it altogether and complaining about it at every opportunity. So are wins and losses, really-no one knows how to predict them and they're not a good indicator of value-but they're ingrained in baseball culture for better or worse.
-People will draft crappy closers too high because you have to do that in a lot of leagues to get saves. In your league, you count so many categories that you don't have to do that. So don't-saves always come into the league, and it's probably viable to have a staff of high strikeout, low walk middle relievers instead of closers, and tank saves in your league and be competitive. But keep in mind that reliever performance varies wildly year to year, and you probably aren't going to win this year, so don't overpay for them.
-If your team is as bad as you say, leverage old players and injury risks on your roster to contending teams for keepers going forward-be willing to lose on talent in the short run, but focus on building for 2011 and beyond.
-In my experience, there are always adequate OFs and starting pitchers around. Don't hesitate to trade three good OFs for a good SS or C in a year like this when you aren't competitive.

I hope some of this is helpful. Rebuilding years in keeper leagues can be really fun.
posted by Kwine at 9:40 AM on October 28

Kwine,

Thanks for your note - all great things to think about. To clarify, it's a weekly head-to-head: I play another team (18 teams in the league) and we get a win or loss per category listed above.

As far as keeper rules: We can keep eight players per year. One has to have been a rookie in the previous season, one has to be a pitcher, one has to be a hitter. Of the eight you are keeping, only four could have been a keeper from the previous year.

And line-up structure:
C, 1B, 2B, 3B, SS, Middle Infielder, LF, CF, RF, Utility (can be any position)

3 Relief Pitchers, 4 Starters and 4 generic Pitchers (can either be reliever or starter).

The roster I inherited is atrocious; I have a corner infielder and a corner outfielder actually worth keeping, and that's about it. It should be fun, but I'm definitely looking for novel ways to exploit the scoring system that other owners might not have thought of...
posted by po822000 at 10:27 AM on October 28

18 teams is a huge league and there's unlikely to be anything good in the free agent pool ever. That makes health even more of a concern, since you will be unable to replace injuries.

More targeted advice,

1) Draft the youngest, highest upside hitter possible in every round until they run out.
2) Figure out the best deal possible for a up-the-middle guy that you can get for your two good players. You don't want to pull the trigger on it unless you get an up the middle guy who will be in the top five at his position when the rest of your team is good, i.e. a young, really good guy. You won't be able to make a deal like this with a smart owner...so try to find a dumb one. Be patient with this-these are your only assets and you have to make them count.
3) Figure out everybody's "5th best" keeper. Since you can only keep four guys next year that you kept this year, lots of teams will have "5th best keepers" that will be less valuable to them but still might be better than anyone you could keep. Target those players in deals-but think of them as short term not long term assets-remember that if your team were good, you wouldn't be keeping that guy either. This is the kind of guy you might throw into that trade for that top 5 middle infielder. Smart owners won't trade a dollar for three quarters, but you want to get some of them thinking, "I can get Shin Soo Choo, Raul Ibanez, AND Jacoby Ellsbury for Hanley Ramirez?"
4) Suck it up and prepare to be a bad team for a while. I just wouldn't even worry about pitching-you can't count on it to be good and healthy when you need it to be. Focus on getting four good, young hitters that you can build your team around. Get one or two in the draft this year, one for your two good players, and one or two in the draft next year. Pick a year when you think you're finally going to be good and go for broke in the draft-pick a bunch of veteran pitching and roll the die.

As for gaming the system,
Is there an innings minimum and an innings limit in your league? If there's no innings minimum, load up on strikeout RPs with your generic slots-they'll have better rate stats, and you can dominate the RP counting stat categories. If there's no innings limit, load up on SPs in the generic pitcher slots and try to crush everyone in the counting stats.

Is there a transaction limit? If not, turn over your pitching staff regularly looking for extra starts, even if they're by crappy pitchers-concede the rate stats and try to dominate IP and therefore counting stats week to week.
posted by Kwine at 11:19 AM on October 28

Kwine,

Thanks for the excellent advice. Wanna co-manage this crapheap with me? :)

To anyone else that wanders by (specifically math nerds), if you have any insights to my original question, I'd still love to hear it.
posted by po822000 at 11:55 AM on October 28

The big difference between your goal and the goal of a real-life GM is that his goal is to win games, and the best (or most undervalued) way to do that is to have hitters who don't make outs and pitchers who do. Your goal is to win the greatest number of these individual categories. I wouldn't spend a ton of time trying to project individual players - there are people out there who make a career out of it, and a lot of good projections are available for free or cheap. Focus on finding the inefficiencies either in the system of scoring or in the market for the players.

One thing I've been meaning to do a little research on myself is how to best leverage the different categories. Different categories have very different expected totals each week, which means that variance for some categories is much greater. This has much more of an effect in a head-to-head style league, rather than a rotisserie style league. So the question is, which categories should you focus on to maximize your chances of winning each week? Starting from so far behind, as you are, I'd think you should focus on the higher-variance categories (although it being a keeper league will I'm sure have implications I haven't thought about). Random chance is more likely to win you a category like steals or saves than a category like strikeouts or runs. This is especially applicable with your league that has so many categories.

There's also a question of diminishing returns - if you assume that all of the categories are normally distributed (which they aren't, but it's a close enough approximation), then putting more effort into categories where you are near the middle of the pack will have a greater influence. If you're horribly behind in one category, then adding the best player in that category still won't help you much. If you're closer to the middle of the pack, though, then adding that players stats will move you up significantly in the rankings. Similarly, if you get to the point where you have a huge lead in a category, it's less useful to add more.

Statistically, you want to maximize the expected value of the number of categories you will win each week. The expected value of each category is basically the percentage chance that you'll win that category. Consider strikeouts, and assume you can treat it as a normal distribution. Your team will have a mean number of strikeouts and a variance in the number of strikeouts, as will your opponent's team. Each player you add in the draft will affect the mean and variance of your strikeouts, which will in turn affect the chances you win strikeouts each week. You could play with some statistics and see if giving up on a couple categories will increase the chances of winning the other categories enough to increase the average number of categories you will win overall.
posted by hatsforbats at 12:58 PM on October 28

I've performed similar calculations for a rotisserie style FF league I am in. Since these categories are equal in weight, here's how I originally approached the problem. I took some forecasts for each player in each category. I totaled up these values for the league to determine relative importance. For example, complete games might add up to 60 games next year, but hits across the league could be 40,000 (just making these up from thin air.) I then picked a number (say 100) and applied a ratio conversion to each category. In the examples above, I would multiply a players CG projection by 100/60, and hits by 100/4000. This gave me some help in weighting the general importance of these. I sorted the position by total points, then took an estimate of how many players would be drafted at any given position and took that players number as a positional offset. For example, say the total for the 20th player was 87.8, I would have a position offset of -87.8 which would make his value 0. The 1st ranked player might have a score of 381.2, so I would subtract 87.8 from that value making his positionally adjusted score a 293.4. In this league, it was an auction format, so I now had to get dollars adjusted on top of these calculations. I summed up the total points for every spot that would need to be drafted, then divided the total dollars avaiable to the league by that number. That gave me a dollars per point figure which i used to overlay expected value. Then I went to the draft.

Well, I thought I had figured everything out pretty well -- all math based and using fairly good principles I thought, but I made a few mistakes. The first one (which I was able to control fairly well) was mistaking the combined value calculation as the true worth of a player. If I ignored the value components and just drafted off of my final numbers, I could have ended up with a very lopsided team (for example, a team that has high SB numbers, but no power.) As it happens, I brought the full excel file with me to the draft, and I was able to keep a running total of points across these positions. The second issue is that people were willing to pay more than I had projected for the top players in this category. I did not adjust in time to this, so I was left with way more salary cap than I needed, and no star players to spend this on. The third issue was somewhat similar -- the owners in the FF draft did not put any value on defense at all. I had projected much more would be spent there than actually ended up being spent at the end of the draft.

Next year, I revised this slightly. Instead of calculating a hard dollar number, I simply stopped at the calculation of the points assigned to each player. Then as the draft went on, I would write down the dollar amount the player went for next to this figure. This gave me a much better approximation of what any future player would go for, as this helps take into account all of your opponents' biases and tendencies.

I also think that doing this math only takes you so far, and you do need to apply a lot of the principles listed from other responses in this thread.
posted by newper at 1:36 PM on October 28

Here's how to win the league (or at least, be the number 1 seed heading into the playoffs):

This is a head-to-head league. A "win" means taking 11 categories each week. Since you want to play Moneyball, there's a very obvious inefficiency in your league that, since it's there in the first place, I assume most of your competitors will not exploit.

Your pitching categories are:
Walks, Strikeouts, Complete Games, Wins, Losses
Saves, Holds, ERA, WHIP, K/9

I assume there's a pitching innings minimum for your league. Your goal each week should be to just barely meet this minimum and load up on top middle relievers and closers. A team of top middle relievers and closers (assuming you meet the minimum innings requirement) will win you Walks, Losses, Saves, Holds, ERA, WHIP and K/9 (7 categories) every week, with a chance to also win Wins. That's 7 categories under your belt every week, without having to do much managing of your roster.

Depending on the innings minimum, you want to draft 2-3 starting pitchers MAX, who will be used solely to ensure you reach your innings. Focus on starting pitchers with high K/9 and low BB/9.

Very valuable for you will be elite relievers who qualify in the SP slot. These players will depend on your league's requirements & how things shake out over the next 4 months or so.

The great thing about this strategy, assuming no one else uses it, is that you don't waste your time drafting starting pitchers, who are generally hard to predict from year to year anyway. This means you can pick up more hitters to help you win the other 10 categories.

Does your league have a games maximum for hitters? If it doesn't, the rest of the strategy is very simple, and fits in perfectly with what we established by ignoring starting pitchers: You use all of the slots you didn't use on starting pitchers to pick up extra hitters, and then always start a full lineup. This will give you the edge in 8 out of 10 of the hitting categories, regardless of who your actual hitters are, since you have more at bats.

If you do have a games played maximum, you want to exactly hit the maximum each and every week.

Let's look at hitting. The easiest skill to predict for hitters from year to year is their batting eye, ie. how often they strikeout vs. how often they walk. Hitters that strikeout very infrequently should be your #1 target, as they will have the highest batting averages. They also tend to hit at the top of the order and thus score more runs and collect more hits. Some of them will also steal a lot of bases. If you find one who also walks a lot (or Albert Pujols) even better. Some players as an example: Dustin Pedroia, Denard Span, Jacoby Ellsbury, Yunel Escobar, Shane Victorino. You should be able to fill up most of your roster with these types of players while others chase "sexier" home run hitters.

Home Runs and RBI are the stats that everyone will chase, which is why in a head-to-head league you should ignore them. It doesn't matter if your team loses HR and RBI every week since your only goal is to win more categories than your opponent.

Using this strategy, most weeks (90%) you will win : AVG, R, 1B, SB, K

You'll have a decent shot (50%) at: BB (depending on your players), 2B, 3B (crapshoot category, but you put yourself in a good position to win it by playing all top of the order/speedy players)

You probably lose (eg win 10% of the time): HR, RBI

That's 6.2 points per week, plus the 7 you are virtually guaranteed by employing only top relief pitchers. That's enough to win you your matchup virtually every week. There will be some weeks where your relievers bomb and you get creamed, but that's fine. You don't need to win every week, just enough to get into the playoffs and (depending on your league) earn a bye.

If you commit to this strategy and no one else does, you will almost certainly end the season in first place. Everyone else in the league will hate you, and probably change the rules for next year, but whatever. You won.

Keep in mind, as Billy Beane famously noted about his team building strategy, this shit don't work in the playoffs, ie. in such a small sample, when only one week matters, anything can happen. But that will be true no matter what your team looks like.

This is all counter to the "best answer" you marked above, but you know...
posted by hamsterdam at 3:03 PM on October 28

Hamsterdam,

Those are VERY excellent points - counter to what I was thinking, but totally make sense.

FWIW, there are absolutely no minimum/maximum for either pitchers or hitters. Thanks so much for everyone's help.
posted by po822000 at 6:20 AM on October 29