Monday, June 13, 2011

40k Theory: Moneyball and 40k

Hey all, time for another book report slash theory article.  This book in question is Michael Lewis' Moneyball: The Art of Winning an Unfair Game.  It's a baseball book, and since some of you are foreigners who don't know baseball too well, I'll try to make some of the concepts relatable.

The premise of the book is that in Major League Baseball, there is no salary cap.  Therefore, the rich teams can buy the best players.  This is true for some other sports like soccer, F-1 Racing etc.  However, Lewis notes that in particular the Oakland A's success has defied all odds.  They have a payroll about 1/4th of the Yankees, yet in many years they win as many or more games than do the Yankees.  How is that possible?

Lewis answers the question by describing the methodology of the A's General Manager, Billy Beane.  Billy, using very advanced statistical analysis and economics and market theory, realized that the market for baseball players is highly inefficient.  Baseball stat nerds have known since the early 80's that the major offensive statistics (Batting average, home runs, RBI) have little or no correlation to team wins.  Those are the statistics athletes are paid big bucks for, but those are not the statistics that win games.

So how do teams win games?  By scoring more runs than the other team.  So what statistics are correlated with scoring more runs, if batting average etc. are not?  Without getting too baseball-y, statisticians discovered that On Base Percentage and Slugging Percentage are significantly more correlated to runs produced (and thus wins) than other statistics.  In the mid-90's when Beane took over the A's, this theory was scoffed at by baseball insiders.  They refused to acknowledge that a bunch of book nerds with calculators knew more about baseball than the Old Boys Club that played and ran the game.  But Beane, with a miniscule budget, had to think outside the box.

So using sabermetric principals, he decided to hire players that were cheap because they didn't have the glitzy statistics, but were strong at the important statistics.  Essentially, he was buying players who the market grossly undervalued.  And the results bore out that he was correct.  Long story short, most teams, even the richest ones, utilize the sabermetric principals that he pioneered.

So how does this apply to 40k, you might be wondering by this point?  40k is not a perfectly balanced game, as you might be aware.  The Eldar and Necrons players will tell you so, if you had doubts.  Space Wolves and IG are the Yankees, Necrons (and the like) are the A's.  Billy Beane has a phrase that jumped out at me, "we look for players with warts."  What he is saying, is that they want players who appear flawed by have some undervalued strength that most teams ignored because they were focused on the visible flaw.  Let's say a player had great On Base Percentage.  But he was as slow as molasses.  Beane might be able to get that player for 50% less salary than if the same player had good foot speed.  [Incidently, foot speed is not correlated to wins in baseball, but is highly valued by teams still despite that.]  

Well, the older codices have warts all over the place.  Heck, even the newer codices have units with warts, that the "market" discovers are actually good.  Wolf Scouts come to mind.  When the Space Wolf codex launched, they did not fare well in the unit by unit reviews.  Until someone figured out that despite missing some key 'stats' that people pay highly for (3+ armor), they do some other things really well that actually win games: outflanking double melta units with decent CC ability.  And thus, a unit with warts was discovered to be genuinely valuable.  The key to identifying wart units, especially in older books, is to isolate important stats and ignore unimportant ones.  Unit A can't fight in close combat.  Who cares, it's an extremely shooty unit that point for point compares well with other shooty units from newer, better codices.  Yes, you will have to be extra careful to avoid CC, but if you can manage it, you have taken an ugly unit and gotten one just as effective as a super star.

My assertion is that other units like this are out there.  We just need a metric like On Base Percentage, to discover them.  My other assertion, is that the old codices can be competitive, when they identify their wart units, and build their list to focus on the strengths of the wart units and minimize the weaknesses.  Essentially, Eldar can't try to out Space Wolf the Space Wolfs.  That's a losing proposition.  They need to win differently.  I think some people do this intuitively.  Tau and Eldar and Necrons still win rather big tournaments.  Players who win with them will say, and I agree, that it is experience with the army that allows them do pull such upsets.

But they are doing it subconsciously.  That won't work for new players, or players just trying the army out.  There is a need for an analytical, scientific approach to it based around isolating important statistics and ignoring unimportant ones.  For the past two weeks I've been giving this issue a ton of thought, and I believe I have created a method to analyze units side by side in a vacuum in a meaningful way.  That said, it's still crude and I'm sure with community input it can be much refined and made exceptionally useful.  In my next article I'll lay out this creation for you guys, and you can let me know what you think.


  1. Roids also did not hurt his team performance. I think BA is similar to OBP and HR is also similar slugging but RBI and Runs are to link to how a team does to be good individual player metrics.

    I think good Eldar and Tau players can still do well for two reasons. 1.) Rock, Paper, Scissor nature of the list building. Those races are sort of discounted so people do not tailor there list with items specifically to kill them. 2.) Unfamiliarity of Opponents: Those races are so out of favor that people have forgotten all their tricks and how to fight them. You do not take them into account in your pre event battle plan since 9 out of 10 games you will not play them.

    Not to discount that the players playing them are excellent and experienced in there armies. That is also important as noted in Good Players above.

  2. Cool, can't wait for the follow up article!

  3. @eriochrome Batting average and homeruns might have something to say about ability to hit and hit for power, but they have very little to say about contributing to your team winning the game. Fact is, team ops is the single most important baseball statistic in determining how many wins a team will get in a given year.

    Batting average is to baseball what Ballistic Skill is to 40k. Ballistic skill is an irrelevant statistic. Ballistic skill will tell you how good an individual model in an unit is at doing something. But more important is volume of fire, and quality of the shots. I'd rather have 30 Str 5 AP1 shots at BS2 than 5 str 3 AP5 shots at BS10.

    So while ballistic skill looks nice on paper as a metric to judge units, its almost totally irrelevant. Same for batting average.