Is there ever any sort of cap or record on luck differentials? At the moment in a current game I'm at
32.32 (Opponent) to -39.35 (Me). Anyone had it worse?
The longer a game runs, the more likely you are to have a greater differential, so on a really big board with lots of units that kind of luck is not too extraordinary. There has been a lot of discussion on how to accurately compare luck, but I don't think there is any consensus, or at least no way to do it easily.
Ozyman wrote:There has been a lot of discussion on how to accurately compare luck, but I don't think there is any consensus, or at least no way to do it easily.
Because we danced around this for so long, I consulted one of my professional stats friends who nudged me in the right direction. I made a post at one point and I'm very convinced by the technique. One part of the implementation is a calculation as easy as luck stats. The other part is harder, but it's a standard calculation, so there should be code out there. I'll try to dig up the post.
http://www.wargear.net/forum/showthread/1950/Including_measured_luck_in_the_luck_stats
Summary: The technology to calculate luck in terms of percentiles exists. It allows us to compare the large board bad luck to the small board bad luck. The technique works across dice types. The results will only be inaccurate/uninterpretable with a small number of rolls.
Hugh wrote:http://www.wargear.net/forum/showthread/1950/Including_measured_luck_in_the_luck_stats
Summary: The technology to calculate luck in terms of percentiles exists. It allows us to compare the large board bad luck to the small board bad luck. The technique works across dice types. The results will only be inaccurate/uninterpretable with a small number of rolls.
So based on your numbers how would you rate a - 43.77 out of about 1500? I know it feels like a kick to the nutz, but it's probably not "probability of being hit by lightning" sucks.
If it's 1500 rolls of standard three 6-sided dice versus two 6-sided dice, then we get a standard deviation of around 31, or about 1.4 standard deviations below average.
This is about the 15th percentile. You are rolling better than about 15% of people rolling 1500 dice. Not lightning, but not good.
The estimate would be better if we had full information about your 3v1 rolls.
Here's the screen shot
itsnotatumor wrote:Here's the screen shot
I don't see your screen shot - perhaps retry.
Hugh wrote:If it's 1500 rolls of standard three 6-sided dice versus two 6-sided dice, then we get a standard deviation of around 31, or about 1.4 standard deviations below average.
This is about the 15th percentile. You are rolling better than about 15% of people rolling 1500 dice. Not lightning, but not good.
The estimate would be better if we had full information about your 3v1 rolls.
That's awesome Hugh! I would love to see percentile as the luck stat. Much more interesting.
Agreed.
Alpha and Mongrel were the driving force behind getting luck stats here.
They haven't been around for awhile, so, Hugh, I think it is up to you -
The Luck stats are pretty popular to look at (I believe) so I think making them more meaningful would be, as Ozy said, awesome.
+1
Amidon37 wrote:Alpha and Mongrel were the driving force behind getting luck stats here.
They haven't been around for awhile, so, Hugh, I think it is up to you -
The Luck stats are pretty popular to look at (I believe) so I think making them more meaningful would be, as Ozy said, awesome.
I actually met them at a conference earlier in the year. They've been busy with research and other things. But, I did ask them what they thought of this approach and they agreed it was the best way to do it. (i.e. the details given in that long prior post)
I'm glad you all like the idea. Percentiles are easy to interpret so I think it'd be cool. And I'd be happy to produce some code, but I'd need a little bit of info from tom regarding the LS implementation.
I actually met them at a conference earlier in the year.
Was it at a Nerd Symposium??? :)
I like to think it was a wargear conference. I'd love to see those proceedings, although why wasn't I invited!
Ozyman wrote:I like to think it was a wargear conference. I'd love to see those proceedings, although why wasn't I invited!
A WarGear conference! What a great idea. But... hmmm, how am I going to get there from the Lyra Constellation?
Hugh wrote:itsnotatumor wrote:Here's the screen shot
I don't see your screen shot - perhaps retry.
I took a screen shot (186k), but the "Image Upload" won't select the jpeg. What do I need to fix to make it work?
M57 wrote:A WarGear conference! What a great idea. But... hmmm, how am I going to get there from the Lyra Constellation?
let me count the ways...
tardishyperspacewarpspeedoutsidewormholejumpdriveftlsliptreamantimatterdrivephiloticweb
weathertop wrote:M57 wrote:A WarGear conference! What a great idea. But... hmmm, how am I going to get there from the Lyra Constellation?
let me count the ways...
tardishyperspacewarpspeedoutsidewormholejumpdriveftlsliptreamantimatterdrivephiloticweb
Ooh Ooh! I like this game. Infinite improbability drive, Holtzman Drive, trans-dimensional teleporter, Alcubierre drive, Conjoiner drives, and if you wanna take the scenic route: reactionless drive, beam drive, ram jet, solar sail... =D
Which one best deals with temporal issues? I can only spare a weekend.