I wish I could blame all of my losses on bad luck, but I doubt that is the case.


When I review the graphs after a game, cummulative luck seems to always end up as a negative number. So does my opponent's.

Any idea what luck number would be considered "neutral" luck? Cannot imagine it being 0.

I'm curious about this myself. I also have generally found that the more aggressive I am and the more attacking I do in a game, the lower my luck tends to be. I'm wondering if maybe the luck calculation is influenced by the 60% vs. 70% attacking v. defending chances in some fashion.

Just roll over your attacks in any game and you'll get a pretty good idea how luck is calculated. If you attack a 2 with 3 and each kill 2 that would be defense luck +.60 (.7x2=1.4 and you killed 2 so 2-1.4=.60 luck) and attack luck +.20 (3x.6=1.8 and you killed 2 so 2-1.8=.2 luck). Luck is a measure of how close the results were to the expected result if there were no luck.

Measured over 100s of attacks, the cumulative luck number should mostly even out. If you've been making a lot of lower probabilty attacks or defenses (a result ending ending in .1-.4 when multiplied by .6 or .7 respectively) then you're more likely to have a lower cumulative number. Of course this too should balance out over enough turns.

If luck is 75% than you get much greater swings and cumulative luck willl jump up and down more.
____________________________

I just checked some attacks with huge stacks to see how they affected luck and the result was very interesting. I attacked a 1 with 585 and the result was -4 offensive luck and -.3 defensive luck. I obviously killed the one defender and didn't lose any armies. So offensive luck calculations ignore the actual results in the game where there are insufficient armies for the luck factor to make a difference. I went -4 because I should've killed 351 (585x.6) and the random generator came out 8 points lower and then averaged against the norm to come out four points lower. Luck is set to 22% so the range of possible results was huge: anythign from 273.78 to 428.22, although with averaging against the norm the actual range is half that: 312.39 to 389.61. The important thing to recognize here is that disproportionate stack attacks can skew luck as a measure of how you're actually performing in a game. MAYBE RANDY SHOULD CONSIDER ADJUSTING LUCK RESULTS SO THAT THEY IGNORE RESULTS THAT DON'T MEASURE ACTUAL ARMIES KILLED {assuming he doesn't do this already in the graphing section).

I checked the cumulative defense and offensive luck in a few of my bigger games (50+ turns) and they're remarkably close to 0 cumulative luck. I've probably received and used up 2000 armies in this game and my cummulative luck is 1.5. My defense is better than my offense on average.

Excellent write-up, Duke. Your analysis is correct. I agree the luck should ignore results where the additional luck didn't help your attack, but it does not currently work this way.

Good analysis Duke. One thing I want to ask about is this:

"I attacked a 1 with 585 and the result was -4 offensive luck and -.3 defensive luck. I obviously killed the one defender and didn't lose any armies."

Wouldn't that be -.7 defensive luck? 0 armies were killed and 0.7 should have been killed for a net of -.7? Or am I missing something?

impressive stuff Duke.

yeah you're right Impaller -- it was -.7 luck for the defender because I didn't lose any armies -- good catch