When I was a kid, I used to sneak into my dad’s home office and pull Bill James books off the shelf.
My dad grew up in Kansas City and was a big baseball fan. Somewhere along the way, he came across James’s early writing. So by the time I was old enough to care about baseball, books like the 723-page tome The Bill James Historical Baseball Abstract were sitting there waiting for me. I read them constantly. At the time, I certainly did not understand all of the statistics (though I obnoxiously started calculating Runs Created for my little league team). What stuck with me was something more basic: a lot of things that everybody knows turn out not to be true.
That idea has probably influenced Data & Dice more than anything else.
The Bill James approach
Bill James was an unlikely person to change how baseball was understood. In the 1970s, while working nights as a security guard at the Stokely-Van Camp cannery in Lawrence, Kansas, James began writing his own baseball analysis. In 1977, he self-published the first Baseball Abstract. A few years later, he coined the term “sabermetrics,” which he defined as “the search for objective knowledge about baseball.”
The appeal of James’s work was never simply that he liked statistics. Baseball already had plenty of statistics. He was interested in whether the statistics people used actually answered the questions they thought they answered. Baseball had generations of accumulated wisdom about what a good player looked like, which statistics mattered, how teams should be constructed and how the game should be played. Scouts and managers had spent their entire careers around baseball. They knew far more about the game than James did. And yet some (much?) of the conventional wisdom was wrong.
Sometimes baseball valued things that looked impressive more than things that actually helped teams win. Sometimes relatively boring skills were being undervalued. Sometimes a statistic that had been treated as meaningful for decades turned out to tell us less than people thought. James kept asking whether those assumptions could actually be tested.
That basic way of thinking eventually became much more mainstream. Moneyball told the story through Billy Beane and the Oakland Athletics, who used sabermetric ideas to look for players and skills that the baseball market was undervaluing. James was not the protagonist of Moneyball, but his work helped lay the foundation for the approach. That part always interested me more than any particular baseball statistic.
The useful idea is that conventional wisdom creates opportunities when conventional wisdom is wrong.
Wargaming has plenty of conventional wisdom
Spend enough time around competitive wargaming and you hear the same kind of claims: Everybody knows this faction is broken. Everybody knows that unit is terrible. Everybody knows shooting armies have an advantage. Everybody knows Fourth Edition is swingier. Everybody knows lists are spammier now. Everybody knows the first turn matters too much.
Usually, these claims are not coming from nowhere. Good players develop strong instincts from playing a lot of games, attending events and talking with other good players. Experience matters. My reaction, though, is usually the same: Does everybody know that?
Maybe the claim is completely correct. Maybe it is directionally correct but exaggerated. Maybe it was true six months ago and the meta has moved. Maybe it is only true for certain factions or certain kinds of players. Maybe we remember the dramatic examples and forget all of the ordinary ones. Or maybe everybody really does know it because everybody is right. I am perfectly happy with that answer, too. The interesting part is finding out.
Turning opinions into questions
That is where the analytics come in. Take a statement like “Fourth Edition lists are spammier.” We could argue about that for hundreds of Facebook comments without getting very far. Everyone can produce an example of a list that supports their position. Or we can turn it into something measurable. Are players taking more copies of the same unit and unit size than they did in Third Edition? Are players within the same faction converging around the same units and packages? Has the range of competitive builds actually narrowed? Now we have questions.
The same approach works with faction balance, matchup strength, scenario performance, player skill, unit efficiency or tournament scoring. The data will rarely give us a perfect answer. Kings of War is far messier than a spreadsheet. Players have different skill levels. Lists interact with each other. Terrain matters. Scenarios matter. Tournament samples are often small. But even an imperfect measurement can move the discussion beyond “everyone I know thinks this.” And sometimes the result surprises me. Those are usually my favorite articles.
Data should be able to change your mind
There is a trap here too. Once you become the person with the spreadsheet, it is very easy to become just as attached to your numbers as someone else is to their conventional wisdom. That misses the point. The goal should be to test the belief, including your own. If I start an analysis expecting to find that a particular faction is too strong and the evidence does not support it, that is useful. If I think a unit is mediocre, and it consistently shows up in successful lists, that is worth investigating. If the data confirms what experienced players have been saying for months, that is useful too. And yes, 4s are, in fact, swingier.
There are also plenty of questions we cannot answer very well. Tournament results can tell us something about balance. They tell us much less about whether an army is fun to play against, whether a rule interaction feels awkward, whether a faction still has the right character, or whether a game is asking players to spend too much mental energy remembering rules rather than making decisions. Those questions still require judgment. Data is most useful when it helps sharpen that judgment rather than replacing it.
Borrowing tools from other fields
I do not think the useful part of what I do is inventing new statistical ideas. Almost none of these tools are mine. They were developed by people working on completely different problems, often long before I started applying them to toy soldiers.
What I enjoy is recognizing when a tool from one field might help answer a question in another.
The Gini coefficient, for example, comes from economics and is usually used to measure inequality. I have used it to look at how unevenly power is distributed within a Kings of War list. The Herfindahl-Hirschman Index (HHI) is something I first encountered through antitrust and merger review as an energy regulatory lawyer, where it measures market concentration. It also gives us a useful way to think about faction or list diversity. And techniques from machine learning can be adapted to ask whether the collection of units in a list tells us anything about how strong that list is likely to be.
None of those applications requires inventing a new branch of statistics. The interesting part is the translation. A wargaming problem may feel unique, but someone in economics, sports analytics, law, finance or data science has often wrestled with a structurally similar question already. That is a big part of what I try to do with Data & Dice: find those ideas, understand what they actually measure, and then see whether they survive contact with the tabletop.
Finding the edge
There is also a competitive reason I find all of this interesting. If everyone evaluates the game correctly, there is no edge. If everyone thinks a unit is excellent, and it actually is excellent, everyone can take it. If everyone understands the strongest factions and matchups, everyone can prepare accordingly. The opportunities appear where perception and reality separate.
Maybe a faction with an ordinary overall win rate happens to have good matchups into the armies dominating the current tournament field. Maybe players are paying heavily for raw damage while undervaluing Unit Strength and late-game scoring. Maybe a unit that the community dismissed early in an edition has a role that was not obvious when everyone first read the rules. Maybe a supposedly terrible matchup is closer to even than players think. None of those ideas has to be true. That is exactly why they are interesting. The useful question becomes: where might the community be wrong?
Why Data & Dice exists
I started Data & Dice because I enjoy Kings of War, and I enjoy data. But the connection between those two things goes back much further than this game–it goes back to sitting in my dad’s office reading Bill James. The lesson I took from those books was not that numbers are smarter than people. Bill James himself was far too interested in baseball, its history and the people who played it for that interpretation to make much sense. The lesson was to stay curious about things people have stopped questioning:
When someone says a faction is overpowered, I want to know how much.
When someone says the meta has become less diverse, I want to know whether we can see it.
When someone says a particular style of army cannot win, I want to look for the people winning with it.
Sometimes the data confirms the conventional wisdom. Sometimes it complicates it. Occasionally it turns it completely upside down.
So when I hear someone say, “Everybody knows…”, my first instinct usually is not to disagree but to open the spreadsheet (or, more accurately, the python script).
And that is probably as close as Data & Dice gets to a mission statement.
