BunnyPuncher
Joined: Aug 02, 2003
|
 Posted:
Jan 23, 2005 - 16:34 |
 
|
I ran a regression analysis on 63,000 games played in 2003 here on fumbbl.
Luck Differential (Player1 luck - Player2 luck) was a better predictor of who won than Coach Ranking Differential (CR1-CR2), TS Differential, Cas Differential, Race and a few others.
Luck Differential was not practically corellated to Coach Ranking Differential which surprised the heck out of me (it is a significant statistical corellation but only 0.04).
The entire analysis will be in the next edition of the GLN |
|
|
PWillis
Joined: Dec 28, 2004
|
| Posted:
Jan 23, 2005 - 16:41 |
|
Good job BunnyPuncher! I look forward to reading that one. Suggested title - "Being lucky - cheesy exploit or part of the game?" |
|
|
origami
Joined: Oct 14, 2003
|
| Posted:
Jan 23, 2005 - 17:16 |
|
I have a suspicion that the entire analysis will boil down to "Lucky dice rolls are an important factor in winning any individual game."
Not exactly a huge surprise, but it should still be interesting reading. |
|
|
Azurus
Joined: Aug 02, 2003
|
| Posted:
Jan 23, 2005 - 17:44 |
|
I'm not really convinced by the results from the first post.
The luck indicator in JBB is actually a 'success' indicator. Put this way, that post reads 'if you're more successful in game than your opponent you will probably win'.
That and the fact that better coaches will usually have higher average luck (JBB 'luck', that is) than poor coaches, due to being more likely to find the least risky way to achieve something.
I'm still pretty sure the 'luck' in game is pretty meaningless, since it's a horribly biased statistic. All it seems to be there for is to give people something else to moan about. (Not that they need it, most people can find something quite well on their own)
Hmm...think I've wandered a bit again...forgotten what I meant to say when I started typing. |
_________________
*This is a public safety announcement. Azurus is a cynical, sarcastic idiot. Please ignore any and everything he may say. Thank you for your attention.* |
|
BunnyPuncher
Joined: Aug 02, 2003
|
| Posted:
Jan 23, 2005 - 17:58 |
|
| origami wrote: | I have a suspicion that the entire analysis will boil down to "Lucky dice rolls are an important factor in winning any individual game."
Not exactly a huge surprise, but it should still be interesting reading. |
Heh.. yeah in the order of "Blind folded drivers are worse drivers than those who are not blindfolded" but there are a few little things in it that will be interesting (like how relatively unimportant cas are).
| Azurus wrote: |
That and the fact that better coaches will usually have higher average luck (JBB 'luck', that is) than poor coaches, due to being more likely to find the least risky way to achieve something.
|
This according to statistics is an old wives tale, that while it has high theoretical appeal (it makes sense) it was not supported in the analysis. See my post above yours. |
|
|
banana_fish900
Joined: Oct 13, 2003
|
| Posted:
Jan 23, 2005 - 19:35 |
|
| BunnyPuncher wrote: | I ran a regression analysis on 63,000 games played in 2003 here on fumbbl.
|
Which approach for regression did you use? I'm also curious about your final R squared. I guess I can wait for the GLN though. |
|
|
BunnyPuncher
Joined: Aug 02, 2003
|
| Posted:
Jan 24, 2005 - 01:58 |
|
| banana_fish900 wrote: |
Which approach for regression did you use? I'm also curious about your final R squared. I guess I can wait for the GLN though. |
Yeah, it won't be a year wait this time and I'd rather not give away the goods before hand. |
|
|
mstrchef13
Joined: Dec 11, 2004
|
| Posted:
Jan 24, 2005 - 03:03 |
|
Luck is a representation of relative success in rolling dice, is it not? I wonder if results would show that races who roll relatively few dice, such as dwarfs, are as a general rule "luckier" than races such as elves who in general roll many more dice per turn. I wonder if there is a way to capture "dice rolled per game" as a statistic. |
|
|
Shepherd
Joined: Oct 28, 2004
|
| Posted:
Jan 24, 2005 - 05:20 |
|
But by that logic, wouldn't it scale? If I roll 1000 dice, my chances of being "lucky" or "unlucky" are no better or worse than if I roll 100 dice. If the odds of success are favourable, i.e. elves are mostly dodging out on 2+ rolls, then your chances of being "lucky" are higher if you roll more dice. Right?
Don't look at me, I failed Grade 11 Finite and never looked back.  |
_________________
A super-hero that always fails his Bonehead roll: MAN-MAN, the averagest hero alive!
Buy Dead Eyes Open, starting July 2005 from Slave Labor Graphics! |
|
Wizard
Joined: Jul 09, 2004
|
| Posted:
Jan 24, 2005 - 05:38 |
|
i hid the the title bar above the screen. |
_________________
"As long as one person lives in darkness then it seems to be a responsibility to tell other people." |
|
tassel
Joined: May 04, 2004
|
| Posted:
Jan 24, 2005 - 07:43 |
|
Actually if this is really what the "luck" was about, it _does_ make an interesting stat. I was under the impression it was a simple meter of wether you succeeded in a roll or not - I thought it didn't even take into account that different rolls have different chances to success. At least I thought I saw this happening as my elves always had a bigger "luck" than ag3 teams. |
|
|
Vero
Joined: Dec 30, 2003
|
| Posted:
Jan 24, 2005 - 15:57 |
|
The luck meter is pretty much perfect. Try it out: Dodge six elves out and repeat until you fail exactly once, your luck is at 50% (if a reroll is used then the there's two dodges counted, one for the failed and one for the reroll). Same for ag3 team and ag2 team. Luck is at 50%. Same goes for blocking (1d blocks only). Haven't checked the ball handling. |
|
|
nazerdemus
Joined: Nov 02, 2004
|
| Posted:
Jan 24, 2005 - 16:10 |
|
| Vero wrote: | | The luck meter is pretty much perfect. Try it out: Dodge six elves out and repeat until you fail exactly once, your luck is at 50% (if a reroll is used then the there's two dodges counted, one for the failed and one for the reroll). Same for ag3 team and ag2 team. Luck is at 50%. Same goes for blocking (1d blocks only). Haven't checked the ball handling. |
This seems odd to me because most games end with both teams above the 50 % mark and hardly ever finish below this |
|
|
mutescreamer
Joined: Apr 09, 2004
|
| Posted:
Jan 24, 2005 - 16:23 |
|
I'm not convinced that it is the most accurate figure, let alone one that is worth spending time assessing. However respect to bunnypuncher for putting the research in....I look forward to reading the results of your analysis (however cynically i will approach it)
good job dude |
|
|
CorporateSlave3
Joined: Feb 07, 2004
|
| Posted:
Jan 24, 2005 - 16:30 |
|
| nazerdemus wrote: | | This seems odd to me because most games end with both teams above the 50 % mark and hardly ever finish below this |
Well let me preface this by saying I know nothing about any of this, but from reading the posts so far it seems that this my be due to how 2 and 3-die blocks are handled by the luck meter?
I am guessing this since I have seen IG, that a POW and SKULL on a 2 die block will make the Luck meter jump - and since most blocks tend to be 2 die blocks, this would lead to the meter reading more success than failure in this case.
Also, RR seem to repalace the 'bad roll' which then doesn't get added to the luck score - dubble skulls on your first block, RR to double POW will show a luck reading of 100%. So since several 'bad' rolls will get replaced each match by rerolls (including Dodge, Pass, Catch, etc RR) this ought to lead to a higher luck meter score on average, shouldn't it?
Ofc, heck if I know, just guessing from what I've observed, which is after all a very imprecise and often inaccurate way of analyzing things!
 |
|
|
|
|
Gnomes are trash
Search
