I think that's the explanation Firi.

my god, i think firinneach is right. has the luck problem finally been solved?

lets say i play two games:
game 1: 50 rounds, i win 100/198 rolls, and end up with around a 50.5% luck stat.
game 2: 1 round, lose both rolls i take in a horrible way, and end up with about a 29.5% luck stat.

the question is, how are these two averaged together? if the games are properly weighted according to number of rolls, then i win 100/200 rolls, and get a 50% luck stat. however, if the luck from each game is simply averaged without a normalization-to-rolls, then i get a 40% luck stat.

i really think firinneach might be onto something here. any chance that we can normalize the luck to number-of-rolls-per-game before each game's luck is averaged together?

Monte, I understand what you want to do. But I don't think it would give a more "true" luck statistic. After all, those few rolls in the short game probably mattered a lot more than the rolls in the longer game on average.

My suggestion is rather to normalise the overall luck score to 50% on average for the whole community.

wiggin, what they are trying to say is that the average luck IS 50% if the outcome of the rolls didn't affect the length of your game.

If you play a long game, say 35 rounds, and it ends in an 8v8 battle against an even sized opponent, then your luck will most likely be at or around 50%.

If you play a game where you finish 7th thanks to losing a 4v3, 3v2, 2v2, you're game is done and you don't get the opportunity to recoup your luck with more rolls. You flag out 7th and it's over. Luck will be around 40% I'd guess.

Over time we all take our fair share of 5th's, 6th's, and 7th's and these are the games that drag down our luck. This is why it is VERY rare to ever see a player with a large number of games played to finish at or above the 50% mark.

I understand that Thraxle.

However, on second reading I don't think I actually understand what it is Monte suggests.

Ryan is good at writing code that truncates (not rounds) numbers. That may account for the less than 50%.

I think what monte is suggesting is that the luck % be weighted based on the number of attacks attempted in a game. He believes that the current calculation gives the 40% luck you receive for three rolls in one game the same value as the 50% luck you receive for 100 rolls in another game. This would give a player 45% luck when averaged together.

Monte is suggesting that the 40% be weighted with the 3 rolls and the 50% be weighted with 100 rolls.

((.40 x 3) + (.50 x 100))/103 = .497 or 49.7%

Having a luck of 49.7% between the two games is far different than having 45%. Of course, monte is only speculating that luck is calculated in this unweighted fashion. I'm afraid only Ryan, and maybe skrum, know for sure.

In that case, I understood him correctly. And I don't think it's a good idea for the reasons mentioned.

I think in your case, 45% luck is actually more "correct". The bad luck you had in those few rolls in the beginning of the game, probably mattered a lot more than the average individual roll in the longer game.

My suggestion is to "articifially" make the the luck score 50% by calculating the average luck score for the whole site and set

luck = ½ * your average luck / the average luck of all players

luck = 1/2 * your average luck / the average luck of all players

We have point values to determine how our games turn out, luck should be less biased towards game placement and value the rolls on their own.

But again, I have no idea how the dom is actually calculated, so this coversation is moot without Ryan's input.