Fleecing your friends for fun OR How I won my fantasy league with a little knowledge of statistics

If you’re like me, you probably just finished off your fantasy draft for the year this weekend. If you’re like me, you’re also sitting there wondering how on earth you managed to end up with an outfield of Dave Roberts (hey, he steals bases!), Nick Swisher, and Pat Burrell. You’re angry at your friend who drafted Chase Utley two spots in front of you in the first round, and then Garrett Atkins in the fifth. Secretly, you are plotting his demise. (But enough about my weekend.) I’m never an advocate of harming anyone… but there are some people who I hope decide to join an international humitarian relief organization… and move to another continent… where they don’t have internet access.
Fear not, my wayward friend. You do not need to go collecting brochures for your friend. You can exact your revenge in a much more sneaky way. All you need is a little understanding of the basic principles of statistics and another owner in your league who’s a little bit gullible. Part of the fun (and the skill) of fantasy baseball is in making skilled trades, even if it means that you have to double-talk your cherished friend from fifth grade and know that you are raking him over the coals. You feel a little guilty afterwards, but then again, you know about the embarassing thing he did on Spring Break during your senior year of high school and have laughed at him constantly for it over the past ten years. And he’s still your friend.
What follows might be considered Sabermetric tricksfor fantasy leaguers. Oddly, many fantasy owners swear that they wouldn’t dare employ such methods, yet shell out big bucks every year on and plan their strategies at draft time around the wisdom in the big pre-season annuals. (Who do they think writes those books?) Some of them draft a player based on a book tip that the player will have a breakout year, but swear that they would never dip so low as to stop making decisions based on “what they know.” In other words, you’ve got them right where you want them. They’re literate enough to know what you’re talking about and stubborn enough to try to prove you wrong.
What’s correlated with what?:Are you in a Roto-style league (e.g., 5×5 league) that counts such silly things as how many HR, RBI, and SB a batter has? (We’ll save the “RBI is a silly stat” discussion for another day.) How can you (quickly) identify a player who has the ability to help you in multiple categories? To do this, we turn to the wonders of correlation. Correlation is a measurement of how closely two things are related. If two things are highly (and positively) correlated, it means that if a player is high on one of the stats, he will likely be high on the other. For example, in 2005, the correlation between HR and RBI among hitters with more than 400 PA was .88. The closer you get to 1.0 (which is the maximum), the more closely related the two are. So, generally, if a player hits a lot of home runs, he will drive in a lot of runs. On the other hand, the correlation between home runs and batting average was .22, which is relatively small. Does this mean that players with high HR totals will have low averages? No, it means that knowing a player’s average doesn’t tell you all that much about his HR total. He might hit a lot, he might not.
So, what one stat correlates nicely with many of the usual stats used in roto leagues? A hint: it’s one that few people bother to look up. Runs scored. The correlations for runs scored to HR, RBI, AVG, OBP, and SB are all pretty good, if not spectacular (all between .41 and .65) So, a player who scores a lot of runs is likely to be good at a lot of things fantasy-wise. (Exhibit A: Your 2006 MLB leader in runs scored, Grady Sizemore. Let me guess, he went in the 15th round to the guy who grew up in Cleveland?) Your best bet at leveraging this one is to find the owner who drafts a lot of power hitters, because to him “power hitter” means “good hitter.” Offer him a high power/low-average/no stolen bases guy for one of his high run-scoring guys who’s not a proto-typical power hitter. You probably lose a little bit in HR and RBI, but you are more likely to gain in runs scored, OBP, AVG, and SB. If your league measures three of those categories, then you’ve just traded two categories for three. Remember, your job is to fleece your friends, and if they are in love with the long-ball, you can use that to your advantage.
Regression to the mean: In 1996, Brady Anderson famously hit 50 homeruns. The previous four years, his totals had been 21, 13, 12 (in a strike-shortened 1994), and 16. It suggested that he wasaprettygood bet to hit somewhere between 15-20 in 1996. After he hit 50, I’m sure that in folloing Spring of 1997, he was over-priced in plenty of drafts, by people who thought that he would hit at least 35 again. His totals from 1997 to 2000? 18, 18, 24, 19. What happened? (I know someone’s going to say something involving the word “steroid” and will have just as much evidence asI doof whether or not that’s the case: zero.) He wasn’t hurt to my knowledge. He just… didn’t… do it again. In fact, he went back to hitting roughly what we might have expected him to hit before that freak season. In statistical terms, he regressed to the mean. It doesn’t always happen like this. Some players actually develop new skills or find a new batting stance or a new diet(ary supplement?) that makes them better hitters. However, in the large majority of cases, players who have freakishly outstanding years regress back to where they started. In layman’s terms, we call this a “career year.” It also works the other way for players who have horribly bad years when their track record andlack of an injurysuggest otherwise. They usually come back to where they were.
If you find yourself with a player who had a career year last year, don’t be expecting more of the same. It could happen, but it’s not likely. (When you’re trying to trade him to your friend, talk up the fact that it could happen… really… it could…) Look for the guy who your friend is despondent at having been “stuck with” because he’s “so clearly on the decline.” Chances are, your career year guy will fetch more than he ends up being worth, and the bad year guy will be obtainable for less than he ends up being worth.
Reading too much into small samples: Chris Shelton. Those who understand statistics and followed baseball last year will know exactly what I mean. For those of you who don’t meet one of those criteria, Chris Shelton was the Opening Day 1B for the Tigers last year and smacked two HR on Opening Day. By the end of April, he had smacked 10 and people were wondering if he might hit 60. This was after 92 AB. He hit 6 HR the rest of the way, and spent a month in AAA. In other words, Shelton had a hot April and a not-so-hot… well, rest of the year. In statistics, the more observations you have on something, more certain you are of its true value. It’s called the error of measurement. The more observations you have, the lower that error goes. For example, we can be pretty sure that Neifi Perez is not a very good hitter. After eleven seasons, 5000+ AB, and a .268 career BA with minimal power and speed, I even wonder how he still wears a major league uniform. After 92 AB, you can’t tell much of anything firm about a hitter.It’s now easy to see thatShelton probably gotlucky over those 92 AB, and had the good sense to do it from Opening Day onward, where he would be the story of the young season. I suppose that deep down Neifi could be a .350 hitter who’s just been really un-lucky for all these years, but I highly doubt it.
In your fantasy league, target the sentimental guy who loves great stories. The thing about small streaks like Shelton’s is that they are very much in the here-and-now and make for great copy for beat writers who have to write about the here and now. In fairy-tale land, this scrappy underdog who no one suspected will keep it going all year. In the real world, which happens to be where most baseball games are played (except, apparently this one), a small sample is a bad way to measure anyone, and like above, most players will revert back to their mean. But, your sentimental fellow owner might be fleeced by virtue of his love of a good story.
The gambler’s fallacy:Speaking of streaks, suppose that you flipped a coin, and five times in a row, it came up heads. If you had to guess what the probability of tails coming up on the next flip, what would you say? The correct answer is 50%, no more, no less. The gambler’s fallacy is the belief that after a streak of bad luck (and lost money), that the probability of a run of good luck is “due.” Ifthe gambler isplaying the coin-flipping game, he will bet his money in such a way that shows he believes that the probability of coming up tails is actually greater than 50%. This error happens in many different situations, including your fantasy league. Let’s say that a player who has a track record of consistently hitting .300 goes on a month-long tear where he hits .500. What is the most likely estimate for his batting average in the following month? If you said .100, I’d like to talk trade with you. The correct answer is .300. He could very well hit .100, but there is nothing about his previous performance that compels him to do so. Yes, over the long run, if the season continued into eternity, he would be likely to have a month in which he hit .100 to balance out the .500 he hit last month, but there’s nothing that says that the month of .100 has to immediately follow.
Sportscasters love saying, “Well, he’s a career .290 hitter, but he’s only hitting .260 at the All-Star break. So you know he’s going to go crazy after the break.” One of the owners in your league believes everything that the sportscasters tell him.That owner will do a little bit of math andthink he’s going to be getting a .320 hitter from you for the second half. He’ll pay like he’s getting a .320 hitter too, if you play your cards right. He’s really getting a .290 hitter (not bad!), but a trade is all about maximizing the value that you have.
Trading away a “clutch” hitter in August: Do you own David Ortiz? Do you want Alex Rodriguez? Here’s all you have to say to your fellow owner in late August. “Well, you know how clutch David Ortiz is. I’m worried about dealing him, because we all know that he’s going to turn it on a couple notches in September because of the pennant race. But, if you’re offering me <insert name here>… wow, that’s a hard one to pass up.” You’ve now played hard to get and implanted a false idea in your fellow owner’s head. (Some might call that a lie. I prefer “negotiation tactic.”) In fact, there has been quite a bit of study of the issue, and there is very little to no evidence to suggest that players have a specific ability to “turn it on when it counts.” In fact, players seem to perform about the same in the clutch as they do at any other time in the game. “Clutch hitters” generally do what they’ve already been doing all along, just in situations where more people are watching. However, in feeding this common (and erroneous) belief of “clutchness” to your fellow owner, you have inflated Ortiz’s value. So ask for that top-line starter in return, rather than a second-tier guy. And chat up the guy who owns A-Rod. Implant the idea that he should be worried about A-Rod faltering in the heat of the pennant race. Then offer to take him off his hands.
A warning: One more thing that you must understand about statistics before running off and using these techniques. Statistics is a game of probability. Just because it is likely that a player will regress to his mean, it does not mean that he will. It’s just more likely. Just because clutch ability seems to be an illusion when held up to the statistical light, David Ortiz may indeed hit .400 for the month of September in the middle of a classic Sox-Yankees dogfight. The career .290 hitter who’s hitting .260 at the break might just hit .320. Statistics, as ascience,looks for overall trends and patterns, but it can not predict what the actual outcome of individual cases is. It can onlytell us what the most likely outcome is. Theproblem is thateven if I say that 55% of the time, event X will happen rather than event Y, I am still going to be wrong 45% of the time. Over the long run, I will be right more often than not, but in the short run, the strategies may look like an absolute failure. (See the entry on drawing conclusions from too small a sample.) If you want to use these strategies, you’ll need to be disciplined to ride out when they don’t work. But, a little knowledge of statistics and probability will go a long way in helping you to fleece your fellow owners in the fantasy trade market.
At least, in the long run it will. Happy trading.


10 Responses to Fleecing your friends for fun OR How I won my fantasy league with a little knowledge of statistics

  1. Sean Smith says:

    Thanks for putting up the first post of our new group effort.
    As I mentioned in a previous email, I’m not sold on clutch hitting being a myth, but for fantasy it sure is irrelevant. Alex Rodriguez hitting a 3 run bomb in the 5th inning of a 16-2 blowout helps you just as much as Ortiz hitting another 9th inning, 2 out walkoff.
    Unless, of course, Ortiz hits that bomb off your opponent’s closer 🙂

  2. David Hannes says:

    Great post, especially about probability. I usually use RBI’s as my primary gauge, as runs are too great a function of who hits behind a hitter…well, I take that into account, as well.
    You should try a head-to-head league, where you can try to incorporate splits into your weekly line-ups…suddenly, Rockies, Orioles and Reds you wouldn’t normally consider can provide you with great numbers when they have both series at home for the week…and Kevin Millwood at home can kill your stats.

  3. David Hannes says:

    I’d like to see standard deviation data as well…someone whose weekly RBI stats for the first 5 weeks are 8, 9, 10, 11, 12 is probably more valuable than another player whose numbers are 4, 16, 3, 17, 10…unless you can identify why they shot up in various weeks.

  4. David Hannes says:

    Also, the numbers won’t help you avert some dumb moves…in ’05, my starting pitchers were Matt Clement, Carl Pavano, Erik Bedard, Scott Kazmir, and Gustav Chacin…one from each of the teams in the AL East…of course, they often faced each other, almost killing my chances for multiple Wins on the week. Even having B.J. Ryan and Bobby Jenks last year came back to bite me, as the Jays and Sox had 3 series against each other last year.

  5. Cory Humes says:

    That’s a great post to dust the cobwebs off this space.
    And I can sympathize:
    “If youre like me, you probably just finished off your fantasy draft for the year this weekend. If youre like me, youre also sitting there wondering how on earth you managed to end up with an outfield of…”
    Grady Sizemore, Reed Johnson, Scott Thorman and Wily Mo? Granted it’s MVN’s 16-team keeper league, but did I fall asleep with my hand on the draft button?

  6. John Beamer says:

    Pizza Cutter,
    Great to see you picking up the mantle at this site with Sean and Matt. I love you blogspot blog and will continue to check in here daily.
    Keep up the good work,

  7. Pizza Cutter says:

    My fourth outfielder is actually Eric Byrnes.
    Sean, it’s not that clutch will actually help you over the course of a season. The point is that you can implant the idea of Ortiz hitting better as a ploy to drive up his value in the mind of your friend.

  8. SABR Matt says:

    Great to see the blog already running some decent traffic…
    I had to get settled back into college and start working on my senior thesis paper, but I’ll be making my first couple of posts shortly.
    Hey PC – You must play in some seriously watered down leagues…none of your tricks work in the highly competitive leagues I participate in each year. 😀

  9. […] On the surface, Kinsler smells like this year’s version of Chris Shelton. Shelton hit 10 HR in April 2006 and was hailed as the next big thing, but ended up in Toledo for part of the year and hit 6 more major league home runs during the rest of the season. I used Shelton as an exemplar in an earlier post on the dangers of reading too much into small sample sizes, and Kinsler’s 46 PA at the time of this writing is a mighty small sample size from which to draw any conclusions. […]

  10. […] In this post on the blog, I talked about how you can use correlation to your advantage in fantasy baseball. Knowing what’s related to what can be helpful to find good players to help your team, but Sabermetricians also use (and mis-use it) to study what skills in a player seem to “go together.” There is a way to test whether the correlation is significant or not, but it’s heavily influenced by sample size. That means that even weak relationships can look very “significant” if you have a 1000 people that you’re testing. […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: