.png)
Background
Recently, I wrote my final piece on the subject of "super" armies in Art de la Guerre (ADLG). The main points were:
From a visual, and statistically naive, analysis I concluded that there are no "super" armies in ADLG. By "super" I mean armies that offer all players an advantage.
My underlying hypothesis was that as games increase, luck, player skill and list choice even out, so efficacy tends to 50%, leaving only inherent list differences (if any).
More rigorous testing showed that luck does even out. What is left behind is real and not chance, but that leftover is a tangle of list strength, player skill and popularity. The tests can only measure its size, not pull the tangle apart.
My hypothesis was only partially right. In a perverse way my initial conclusion still holds, but for totally different reasons. There still aren't any "super" armies because, given the data available, it's impossible to prove or disprove their exstance. It's case unproven.
One of the "byproducts" of the rigorous testing was a list of 20 (yes just 20) statistically different armies (good and bad) in ADLG V4. At first I didn't see it as particularly useful: the differences could be due to any combination of factors not just an army's inherent performance.
Now I've had time to reflect, I think the lists may be of more general interest; not least because they contain a few surprises and people do love a good list or two.
Caveat
In what follows please resist the temptation to ascribe army performance to any one factor. At the risk of repeating myself: an army's efficacy could be the result of any number of factors. The only factor eliminated is chance. So when drawing your own conclusions please use a hefty "pinch of salt".
Definitons & structure
As well as dividing the list of signifcantly different armies into good & bad they were further subdivided: those that passed rigorous statistical tests of difference and those that passed weaker tests (see full technical details).
This gives the four groups presented here. In each table the armies are listed in decreasing statistical certainty; specifically how sure we are that the army's efficacy differs from 50%, not how far it differs.
Some armies were effectively tied. 166, 110 and 245 differ by very little. Likewise for 251, 85 and 131. Their relative order isn't meaningful.
Good, strong evidence
Good, weak evidence
Bad, weaker evidence
Bad, strong evidence
Closing remarks
I think I've said enough on this topic, but I'd be very interested in your comments.
2 comments :
The obvious thing is there's no obvious thing - it's not a case of armies with Knights being good and ones with Spearmen being bad, or similar.
For armies with low play counts, just one or a few particularly good or bad players might explain the result. Harder to buy with e.g. the Nicephorians, but I suppose there might be one or two *very* active players who happen to favour them and be very good?
I have it on good authority that the Nikephorians have always been popular under lots of rules not just ADLG. There were a number of changes in V4 that favoured them (see previous post's comments). Plus Dark Age (or pre-Feudal) competitions are also popular.
It's easy to imagine how these factors (period, popularity & list advantages) could combine to produce a consistent run of victories above the average; especially for frequent players. I just can't prove it one way or another.
Post a Comment