April 3, 2009
Stealing Stolen Bases
The stolen base is one of the most hotly contested statistics in fantasy baseball. Some owners swear by it, and will draft early and aggressively to stockpile enough of them, while other owners are willing to treat it as if it doesn't exist in their quest to bulk up elsewhere on offense. While these opposing viewpoints as far as fantasy value have not changed much over the years, the game itself has, with the steal being less of an emphasis league-wide than it used to be. Today I want to take a closer look at stolen bases, to see what trends have arisen in the past 11 years, and what we can glean from that information for fantasy purposes.
First, let's look at the total number of stolen bases and attempts across the league for each of the past 11 seasons (since the league expanded to 30 teams):
Year SB CS SB% Attempts 1998 3284 1504 68.6 4788 1999 3421 1519 69.3 4940 2000 2924 1324 68.8 4248 2001 3103 1408 68.8 4511 2002 2750 1282 68.2 4032 2003 2573 1132 69.4 3705 2004 2589 1100 70.2 3689 2005 2565 1069 70.6 3634 2006 2768 1110 71.4 3878 2007 2918 1003 74.4 3921 2008 2799 1035 73.0 3834
The first thing you may notice is that stolen bases are way down since 1998, with 485 fewer successful steals last season than there were 11 years ago. There were also 954 fewer attempts, and based on the significantly improved success rate, a lot of those shots at base thievery were not missed on a run-scoring level. The number of steals was lower still just a few years ago, with 2003-2005 the nadir of this period, but they've been back on the rise very slightly since then, with last year witnessing a return back at 2002 levels (though with a much higher success rate).
This table makes it appear as if steals are far scarcer for fantasy purposes than they were as we approached the turn of the century. That's true in a way, but there is much more going on here that isn't shown in this one data set. First of all, you don't draft the entire league in most fantasy formats, instead sticking with 10 to 12 teams' worth of players, with roster sizes and arrangements that may not always reflect those of the major leagues. Yes, there are leagues that are much larger than 12 teams, but generally speaking, fantasy leagues use far fewer players than Major League Baseball does during a given season.
This becomes significant once we see what's going on in this next table below. Here, I rank the top 30 and the top 60 in steals for each of the 11 seasons, and then found what percentage of the league's steals they had compiled:
Top30 Top60 Top30 Top60 Year SB% SB% SB SB 1998 33.1 50.4 1086 1654 1999 33.1 51.9 1132 1775 2000 32.7 50.8 955 1486 2001 32.3 50.1 1001 1556 2002 33.0 50.4 908 1386 2003 32.8 50.3 845 1295 2004 34.1 52.0 883 1347 2005 37.0 54.1 950 1388 2006 38.3 54.6 1059 1507 2007 37.1 55.9 1082 1631 2008 36.3 54.5 1016 1524
So sure, total steals have dropped over the past 11 years, but the concentration of steals among the game's top basestealers has risen during this time. Yes, the top 60 are stealing a little less often than they used to, but not in significantly smaller quantities: the difference between the top 60 in 1998 and 2008 is 130 steals, or only 2.2 per player. If you look at the actual list of players, you'll see that fewer had 50 steals, ending up in the mid or high 40s, and while 43 players registered at least 20 steals in 1998, 37 accomplished that feat in 2008.
Given the similarities in the lists both at the top and at the bottom and everywhere in between, it's not really accurate to say that there are fewer players available to draft for stolen bases than there used to be. The bulk of the missing steals can be found elsewhere, with the other 900-plus players that picked up a bat in 2008. This is also where many of the missing stolen-base attempts disappeared to (Att% is Attempts Percentage):
Year Att%(30) Att%(60) 1998 29.6 45.9 1999 29.3 46.7 2000 28.9 46.0 2001 28.3 44.7 2002 29.6 45.6 2003 29.6 45.7 2004 30.4 47.3 2005 33.0 48.8 2006 33.9 49.1 2007 33.4 51.3 2008 33.0 49.8
The top 30 and top 60 in steals have gradually picked up more and more of the attempts relative to the rest of the league, while those who steal fewer bases have seen their chances dwindle through the years. This might explain the improved success rates as well; you have to wonder how much of the trend is the result of teams and managers better understanding certain statistical principles, such as the success rate necessary for the stolen base to become a productive part of an offense. The fact that it's been the players other than those ranking at the top in steals that have seen the bulk of the drop in attempts makes me think that this has a lot to do with it.
Stolen bases are lower among the players that were most likely not drafted for their base-stealing prowess. While in 1998 106 players reached double digits, just 82 managed at least 10 successful steals in 2008. There were 161 hitters who accumulated five steals or more in '08, while 198 did so in '98. (Random fact: Vladimir Guerrero was 161st, while his older brother Wilton Guerrero was 198th 11 years previous.) This would indicate that picking up a few players who are threats on the basepaths is important; you can't necessarily rely on all of your players to swipe 5-10 bags, but what cost does this have for the rest of your statistics?
I decided to use MLVr (and its positional-adjusted variant) to determine whether or not the players in the top 30 and 60 were better or worse in overall production now or in previous years. While MLVr measures from the average rather than replacement level, it is useful for the purposes of this exercise. First of all, the replacement level in fantasy baseball is higher than in regular baseball, since you're drafting the cream of the crop rather than those relegated to bench jobs and the like. You want players who are above average in fantasy, otherwise you're going to lose. Since we're trying to track changes over time as well, we need that adjustment for context purposes. Secondly, MLVr does not include steals in its formula, so I'm not double-counting offensive value for these players. Remember, we're looking to see if the quality of those who steal the bulk of the bases is better or worse than it used to be back in 1998:
Year MLVr(30) PMLVr(30) MLVr(60) PMLVr(60) 1998 .066 .075 .081 .068 1999 .074 .070 .070 .047 2000 .023 .009 .045 .032 2001 .037 .017 .053 .027 2002 .035 .026 .026 .012 2003 .038 .025 .064 .049 2004 .048 .028 .029 .016 2005 .041 .031 .052 .032 2006 .031 .035 .047 .044 2007 .053 .040 .048 .033 2008 .050 .032 .050 .033
I think it's best that we forget about the things we did in order to win our leagues back in 2000 and 2001, given the average run values from the top 30 for those two seasons. For context, over a 162-game season, the average top-30 player from 1998 was worth 12.2 PMLV, while that same player in 2000 was worth 1.5 PMLV, a difference of over a win in real-world baseball. Things balance out over the course of the top 60, but you can still imagine the kind of cringe-worthy overdrafts that were occurring just to secure steals.
What is most striking about this table is that the average run value of the top 30 and top 60 is lower now than it was back in 1998. It's not a massive difference-12.2 PMLV to 5.2 PMLV for the top 30, and 11.0 PMLV against 5.3 PMLV for the top 60-but it's there nonetheless. At least things have climbed from years like 2002 and 2004, when the average PMLV was depressingly low, but this does lend some credence to the argument for drafting players specifically for steals. Proper drafting can land you plenty of the more balanced players (remember, this is an average; the top player in the top 60 from 2008 had a PMLV of .349, while the lowest was at -.315, a difference of nearly 11 wins over a full season), so this doesn't mean that you should run out and forsake other statistics for the sake of steals alone. It does mean, however, that you can probably get away with picking up one of those poor hitters who steals plenty of bases, as long as you tend to your other needs elsewhere.