Good morning and thanks for stopping by this site again today. I really do appreciate it, and I love all of the feedback I have been receiving. I had asked for some guest columnists for this week because I am closing on a house on Thursday and I seriously, seriously should be doing some packing! So much to do, but oh well.
Anyway, I had a couple of people offer to write for me. One of them was Dave Anderson. Dave has e-mailed me in the past and some of his thoughts have been posted on this site. Recently, he mentioned that he has been working on some thoughts on fantasy baseball, specifically stratification for 5x5 leagues. I asked if he would like to have his thoughts published online. He was game, and I certainly appreciate it. So, thank you very much Dave. I think that what he has written below is incredibly interesting, very well thought out, and from the detailed statistical analysis, very informative!
So, if you have any questions or comments on the below, please e-mail me and/or e-mail Dave. I am sure that Dave would love to hear from you. So, here we go:
Making 5 x 5 Fantasy Baseball Work for You
You are reading my first article about fantasy baseball. Hopefully not my last. Although I’m only a recent fantasy addict, I’m a long-time stats nut who has been able to make a correct prediction once in a while. For instance, Twins fans here might appreciate that in May 1982 when the “Twinkies” were in last place (they would lose 102 games that year) and owner Calvin Griffith had just alienated fans by trading away all his “name” players, including Rob Wilfong and Doug Corbett for a minor league prospect named Tom Brunansky, I wrote a letter to the editor to the Duluth News Tribune saying:
“No one will be complaining in 1987, when Tom Brunansky is an All-Star… and the Twins are in the World Series.”
You know how this turned out – Brunansky was an All-Star by 1985, and the ’87 ‘Twins beat the Cardinals in seven. Once the Twins clinched, the News Tribune sports editor interviewed me about how 9th grader could have predicted a last-place team to win the pennant five years in advance. Really it was simple: minor league stats, some knowledge about when players really peak, and some good luck. Okay, a lot of good luck. But the Twins with their core of talented young players had to peak and contend by 1987, and they did.
Fantasy baseball can be predictable too, although of course its complicated by the fact that you’re in a league with a dozen people trying to out-predict you. But a little draft strategy and statistical knowledge will carry you far. I will outline some things that have worked for me.
A note about strategy
First of all, it helps to know what a strategy is vs. what a tactic is. Strategy according to Webster’s is “the art of devising or employing plans or stratagems toward a goal.” Note that it’s an art, and it requires having a plan and knowing what your goal is. On the other hand, a tactic is merely “a device for accomplishing an end.” So “drafting a good second baseman in the early rounds to score points at a weak position” is not a strategy, it’s just a tactic. This tactic could be part of an overall strategy to maximize total points by taking advantage of positional scarcity on offense. Second, whether we are consciously aware of our strategy or not, we do have a defacto strategy – even if that strategy is “wishful thinking!”
Behind the 5 x 5
Of course, your goal in drafting your roster in a 5 x 5 league is to select players to give you maximum points in the 10 categories relative to your opponents. To form the best plan to reach your goal, you must know this: that while each of the 5 x 5 categories has the same value, they do not have the same importance. Some of the categories are more important than others, because they have a higher degree of correlation to success in the other categories (they might not be the ones you think – more on this later). So you need to be strategic about which categories you target with your roster choices.
As you probably know the traditional 5 x 5 categories are
Offense: R, HR, RBI, SB, AVG
Pitching: W, S, K, ERA, WHIP
One thing to notice about these categories is that Offense has only one average-based category (batting average), while pitching has two (ERA & WHIP). This may affect your strategy for protecting a lead or catching up from behind, since averages are more volatile and can change faster than totals. Also, categories based on averages (and therefore Pitching in general) are affected less by injuries.
Another thing to notice about the 5 x 5 categories is that offensive players tend to participate in more categories than pitchers. Batters have the potential to help you in all five categories (of course we all want the 5-category stars). A top starting pitcher might help you out in four categories W, K, ERA and WHIP, but not S. A good reliever will give you S, ERA & WHIP – but fewer K and W than a good starter.
In a given year on offense, there will be a handful of 5-category offensive players (well above average in all 5 categories), several 4-category players and many 3 category players. Compare this to pitching, which in a typical year will have a handful of 4-category players (ace starters) and several 3-category players. There might not be a 5-category pitcher in our fantasy lifetimes.
Now, some fantasy owners would conclude “Hey, since there aren’t many 4-category pitchers, I’m going to use my highest picks to get them – after all there are fewer pitchers on my roster so their performance counts more.” I would disagree – it just doesn’t make sense to spend top picks on (hopefully) 4-category pitchers when there are 5-category hitters available – or any picks on 3-category players while 4-category players are available. Especially considering that pitching stats like W and S have as much to do with luck and team performance as they do with individual performance.
We should remember that we’re not trying to win baseball games, we’re trying to score high in multiple fantasy categories in order to get the most points. We won’t win every category, so its important to prioritize.
Correlations
As I said earlier, some fantasy categories are more important for you to invest in than others, and the reason is that they have higher correlation to success in other categories. If you don’t believe this, you need to.
To give concrete evidence, I took a good-sized data sample: the entire 2003 season stats for all hitters with 400+ at bats, and all starting pitchers with 100+ innings and 5 or more starts, or relievers with 45+ innings and 5 or fewer starts. I ran correlations among the 5 x 5 categories and here’s what I found…
Highest to lowest correlations (r-squared scores) for Offense:
Comparison r-squared value
HR to RBI 0.751
RBI to R 0.357
R to AVG 0.346
HR to R 0.317
AVG to RBI 0.136
R to SB 0.089
HR to AVG 0.059
SB to AVG 0.024
SB to RBI -0.046
SB to HR -0.064
Conclusions? HR, RBI, and R are most highly correlated to the other offensive statistics. Want to see a strong correlation? Look at this scatterplot of 2003 HR vs RBI totals:
The high correlation of HR to RBI is not surprising, considering that each HR automatically generates at least one RBI (consequently, it is nearly impossible to for a player to be among the league leaders in RBI without significant HRs). In addition, each fantasy home run gives you one run scored and one hit toward your batting average, making it the ultimate offensive fantasy statistic.
Offensively, what you want are players who will score well for you in a cluster of categories, at least three or four. The most common clusters will be (HR + RBI + + R) or (HR + RBI + R + AVG). Note that batting order is crucial to getting high R & RBI totals from the same player, e.g. leadoff hitters will have a harder time getting RBI (especially in the National League where pitchers hit).
Based on the correlation scores, what’s going to happen when you fill your roster with high AVG, high SB type players? You’ll have a shot at leading in those two categories, but little chance to even compete in HR and RBI. The reality is you can’t afford to have high-speed, low-power players on your roster, unless they are truly exceptional at what they do – and by exceptional I don’t mean 25 stolen bases in a season. The SB is the dog of the offensive statistics, with a low correlation to Runs scored (someone tell Whitey Herzog) and negative correlations to RBI and HR: the more SB a player has, the fewer HR and RBI they tend to have. No wonder players with speed and power are the rarest fantasy gems!
Highest to lowest correlations (r-squared scores) for Pitching:
Comparison r-squared value
ERA to WHIP (starters) 0.714
W to Ks (starters) 0.448
K to ERA (starters) 0.422
K to WHIP (starters) 0.415
W to WHIP (starters) 0.300
W to ERA (starters) 0.282
K to S (relievers) 0.161
S to WHIP (relievers) 0.113
S to ERA (relievers) 0.082
W to S (relievers) -0.002
WHIP, ERA and Ks all correlate well vs. each other, trailed by Wins. Chasing Ws at the expense of the other categories is silly – wins may be the whole point of real baseball games, but in fantasy baseball they have merely the same value as the other 9 statistics – and less importance than most of them. Wins are highly variable from year to year even for consistent pitchers; correlations with the other categories are low. Kenny Rogers may be good at winning real life ballgames (and writing country songs), but you’d better know when to hold him and when to fold ‘em. I’d say don’t hold him at all: you don’t have room on your roster for a player who’s above average in only one category.
So for pitchers you want a cluster of success in at least three categories. Common cluster to look for are (WHIP + ERA + K) or (K + W + ERA and/or WHIP). Your first priority should be starting pitchers likely to give you above average WHIP, ERA and K and hopefully some wins.
Saves have the lowest correlations of all the pitching statistics, since the greatest predictor of Saves is save opportunities, given to relievers by their managers and circumstance (can you say Danny Graves?). Saves are like the stolen base of the pitching categories – don’t sacrifice too much to get them.
On the other hand, we could say that Strikeouts are the home runs of fantasy pitching – Ks correlate highly with the other pitching stats. We should not be surprised, since each K is an out that lowers ERA and WHIP and increases the chance of a W or S.
With HR and Ks so important to fantasy baseball success, is it any wonder the game is so popular? Those exciting Home Runs and strikeouts are at the crux of our game, just like they are for the casual baseball fan.
My 2004 Draft
Going into the 2004 draft, my strategy was to maximize overall points by drafting the best multi-category players I could with one eye watching scarcity at the thin positions (C, 3B, 2B). I would steer away from players promising good SB totals unless they also had some power potential. I tried to make as many value picks as I could (choosing best available) rather than tactical picks (“gee, it’s about time I picked a good closer”).
The draft took an early run on starting pitching (Curt Schilling in the 1st round, many others picked in rounds 2 and 3). I stuck to my guns and picked the best players available on my top 200 chart.
By the time the draft focus shifted back to position players, I had already stocked several offensive stars, and started to pick overlooked pitchers including Roger Clemens. With most of the owners in my league from the New York area, there seemed to be a bias against former New York players, perhaps because their every failure had been chronicled in the local media. A case in point, I also added Armando Benitez to my team when he slid lower than market value. (It’s hard for any of us to be objective about players from your favorite teams – I blame this lack of objectivity for making me actually draft Kyle Lohse!)
In a real-life example of mis-prioritizing the pitching categories, this year someone used a top pick on a closer (Eric Gagne). Well, Gagne may have been a legitimate National League MVP last year, but even if he duplicated his 2003 performance this year (he won’t), what would you get? A pitcher with stellar ERA and WHIP totals (multiplied by fewer innings than a starting ace), an average strikeout total (vs. starters), and lots of saves. This owner got a 3 category pitcher in the first round and I (fortunately) got Carlos Beltran, who will be a 5-category offensive player if he can stay healthy and keep his average up.
Closers came out early just like starters, but I tried not to chase them until the right time. After getting Benitez and Arthur Rhodes (which seemed like a good idea at the time), in the 24th and final round I picked up Latroy Hawkins. I figured I’d rather get good ERA and WHIP totals from a potential closer than an established closer with bad ERA & WHIP. Today, Hawkins is closing for the Cubs and giving me above average ERA, WHIP and S totals. If he loses the closer job, he should still help my ERA and WHIP.
My middle-round value picks included Frank Thomas and his left-handed alter-ego Adam Dunn, both gigantic sluggers who walk enough to score as many runs as they get RBI – archetypes of the (HR + RBI + R) cluster. Value picks in the later rounds included Marquis Grissom, who people may have overlooked because he doesn’t steal bases anymore (at least with Barry Bonds hitting behind him). Yet even without SBs Grissom can be a good producer in the other 4 categories.
Today I’m in first place in the league. Injuries may still take me out, but I had the satisfaction of drafting players I felt gave me a solid chance to win.
© 2004, Dave Anderson