Of course Ryan could fix the sub-50% issue by simply keeping track of the number of rolls per game. In fact, that would be a good warmup for fixing the scoring system.

well I don't even think these 1% differences are really that significant but while I was checking them, I thought, well why not post them for fun... maybe I could get a 51,5% luck for the second half of the month ;-)

I have the impression that Ryan did fix the sub-50% luck problem.

Sub-50 luck among the higher ranked players may indicate that they know when to make riskier rolls...when the payoff is worth it.

That's not right Skrum.. if the "luck" were based simply on win/loss percentage than most people would be considerably over 50%. Luck shouldn't be controllable as you say.

I don't really know what he does, probably compares the average of all your dice rolls to the expectated average of 3.5. A really good measure of "luck" (which has no certain definition) would be tricky. You could roll a 5 every time, and lose to your opponent's 6s. All 5's is lucky, but you still lost....

Riskier rolls are fewer in number when they are made only when the payoff is big. It takes more games for the odds to "even out" when a particular roll is used less often, and it is unlikely that enough games have been played yet for the rarer rolls to have evened out to 50%. The people who have made risky rolls more often will have evened out to 50%, but they will not be in the top 10.

The "luck" statistics is pure and utter nonsense altogether and you all know that. I just had a game losing 2 times 4v2 attacks in 1st turn, defended only once the entire game and got 4th - luck 46%...

On my wall I have a game with not a single win at all and 20.7% luck...

Not to touch the point of starting position, restack and so on - this value is just bullshit.

I concur that this value has no use. It is there because there are enough people who want it.

I did a quick t-test using graphpad.com of the rank of the 5 luckiest players and the 5 unluckiest players. Guess what. The difference in means is not statistically significant.

luckiest 5: mean rank 1505 SD 1034
unluckiest 5: mean rank 2481 SD 2050

t = .8983
df = 8
p = .3913

Difference is insignificant.

ooops, I really didn't want to start yet another thread about the meaninglessness of luck%. I just checked it and thought it looked funny that I had the lowest luck of top10.

Luck% will only be a meaningfull parameter if it takes into account the difference from the expected values of all roll%. What I mean by that: 8v8 is rougly 44% chance of winning so if you have >44% for your month 8v8 average win%, your luck% should get influenced +; <44% should make it lower. Luck % should give a true measure of how lucky you are and for that it would somehow have to sum all differences from the expected value for every AvB roll possible.

Ideally I would also give a triple weight for every roll in the first two rounds and a double weight for your rolls in round 3 and 4 since they usually are more important than later in the game. Rolls that would get you connected if won would have to be weighed x 2 as well for adding 2 or 3 extra lands to your main group and x 3 if it would connect 4 or more extra lands to your main land mass by that roll.

The more accurate the luck stat is, the more boring it is. Imagine everyone having 50% luck. Yawn...

Rolls are counted along with the chance of winning each roll. The luck stat is an average of the odds you made with each roll.

The original sub 50% problem was because when I was doing a daily sum I was averaging luck per game instead of per rolls per game so short games, typically with few rolls bad luck got a boost in the luck calc. Now daily and month sums are averaged based on rolls and this is why we see a 1-2% increase in luck this month.

The goal however is to reach 50% as it should. Actually the real goal of this stat is to show people that they don't have considerably worse rolls than others so as long as everyone is relatively equal its doing its job. But still, I wonder where the missing % is.

It also should be noted that you'll never see games where people have considerably higher luck than 50%. Typically when you get really good luck you play out the game making a much larger number of rolls and your average creeps down closer to 50% (in that 8v8 stage probably). It's normal to see many <50% games and only a few >50%. But the few >50% should have so many rolls that they bring the average up to 50%.

Maybe the missing percent is just a rounding error or precision error in my probabilities table.