Frankenstein
Joined: Aug 02, 2003
|
  Posted:
Apr 28, 2003 - 02:07 |
|
What's the formula for the MVP-distribution?
By no means are MVPs distributed randomly, I first wondered when I retired my 83-games player without a single MVP, studying my current roosters supports enough evidence that not every player is eligible for a MVP-award (so far it would appear that only players on the lower slots are available, whereas a high number excludes players from gaining a MVP, but there might be another reason for that phenomenon).
http://fumbbl.com/FUMBBL.php?page=team&team_id=5162
http://fumbbl.com/FUMBBL.php?page=team&team_id=4941
Combined with the recently retired 83 games lineman the players on the bottom slots made about 300 games without a single MVP. For teams which have rosters containing 11 players approximately. That's a probability of about 0.0000000000002%. Now, that would be an awesome coinincidence! |
Last edited by Frankenstein on %b %28, %2003 - %03:%Apr; edited 3 times in total |
|
Jigplums
Joined: Aug 02, 2003
|
  Posted:
Apr 28, 2003 - 02:44 |
|
I find whenever some dies he gets the mvp. i think it would be better if the player getting the most ssp's got the mvp or your opp could choose |
|
|
Frankenstein
Joined: Aug 02, 2003
|
  Posted:
Apr 28, 2003 - 03:04 |
|
Jigplums wrote: | I find whenever some dies he gets the mvp. i think it would be better if the player getting the most ssp's got the mvp or your opp could choose |
Hmmm, so far not one killed player of yours got a MVP th egame he died.
Can you provide any data which would support your curious assumption? |
Last edited by Frankenstein on %b %28, %2003 - %03:%Apr; edited 1 time in total |
|
Korhil
Joined: Aug 02, 2003
|
  Posted:
Apr 28, 2003 - 03:34 |
|
There is no formula for the MVP distribution since its just generated off 'random' numbers at the end of a match.
The 'random' seed used for each game is different, and it doesnt record who got the MVP in each game that you run the Java Client.
I've had a team that played 9 games and 1 player got the MVP 5 times during that, weird things happen.
Some people get FF 20, thats just as unlikely, if not more so, thats alot of rolling without getting 1's.
---Korhil |
|
|
Frankenstein
Joined: Aug 02, 2003
|
  Posted:
Apr 28, 2003 - 03:47 |
|
Korhil wrote: | There is no formula for the MVP distribution since its just generated off 'random' numbers at the end of a match.
The 'random' seed used for each game is different, and it doesnt record who got the MVP in each game that you run the Java Client.
I've had a team that played 9 games and 1 player got the MVP 5 times during that, weird things happen.
Some people get FF 20, thats just as unlikely, if not more so, thats alot of rolling without getting 1's.
---Korhil |
Sorry Korhil, but you obviously don't know what you're talking about. It's certainly not unlikely (it's actually almost inevitable) to get FF 20 with a good team and by no means is that chance anywhere near 0.0000000000002%, as you claimed in your posting. Getting 5 MVPs in 9 games (what's so remarkable about that, btw?), suffering 4 kills out of 5 casualities, multiple double-ones or -skulls, 5 consecutive FF-rolls of 1, all these occurences are well within reasonable statistical probabilities, but the MVP-thing is definitely not. Something must be wrong with the current formula or the generated random numbers. |
|
|
Mnemon
Joined: Aug 02, 2003
|
  Posted:
Apr 28, 2003 - 15:17 |
|
|
Frankenstein
Joined: Aug 02, 2003
|
  Posted:
Apr 28, 2003 - 18:33 |
|
Thanks, Mnemon, that roster is indeed interesting with regard to this subject. If it wasn't for the single MVP for Nr. 14 it would conribute further evidence to the initial suspicion. I very much assume that he got his MVP when all 13 preceding slots contained a player. Nevertheless, the overall pattern of that roster is intriguingly similar to the pattern of the 2 Wood Elf rosters posted above. 4 of the first 5 players (the 4 who played 20+ games) have 3 MVPs, the following 5 have 0-2 and the last 3 players only got 0, except for Nr.14 (who got one lonely MVP). |
|
|
AlphaX
Joined: Aug 02, 2003
|
  Posted:
Apr 29, 2003 - 00:31 |
|
I don't see a team with 300 games. Interperting :
http://fumbbl.com/FUMBBL.php?page=team&team_id=4941
is not so trivial; interpert the MVP as a collector's problem, and you should know that getting an MVP for all your players can take hundreds of games. Even if MVPs are distributed uniformly amongst the players present at the game. [is the rand() used in javabowl uniform?]
playing 46 games and not getting an MVP ? well if the team had 11 players on it - that is (10/11)^46 - which according to my windows calculator is about .012% - very unlikely... you cannot pick groups which match what you are looking for and call the result a valid statistic (that is fishing.) pick teams at random to come up with your 300 games.
Empirically speaking, there seem to be many more low numbered players with lots of MVPs than high numbered players...
This thread belongs under the java client ??? The webpage, fumbbl.com, doesn't assign the mvp. |
|
|
SkiJunkie
Joined: Aug 02, 2003
|
  Posted:
Apr 29, 2003 - 03:47 |
|
Hmm...that could be the problem. I'll have to take a closer look. I think the game rolls a random number 1 - #players on the team. If the player for that number is not eligible then it rolls again. But, when I wrote that I don't think the game allowed you to skip roster number. Now that it can, if you have 12 players but they are numbered 4-16 then it is possible that player 13-16 will never get an mvp because the die is rolling 1-12. I'll have to take a look later. |
_________________ --SkiJunkie
________________________________
Java BBowl - http://www.geocities.com/javabbowl
Mirror - http://javabbowl.no-ip.org
Mirror - http://home.austin.rr.com/javabbowl |
|
Korhil
Joined: Aug 02, 2003
|
  Posted:
Apr 29, 2003 - 03:56 |
|
Just generating a number between 1-16 and re-rolling when needed would be easier, I figured thats how it would be done anyway.
---Korhil |
|
|
Dragoonkin
Joined: Aug 02, 2003
|
  Posted:
Apr 29, 2003 - 07:05 |
|
I thought only players who actually played in the game got the MVP...and because once you tend to get up to around 16 players you have 8-9 "always on the pitch" guys and then some other ones...that would tend to lend itself to the same people always being in the MVP distribution.
My Khemri team, for instance, has 4 Mummies, 2 Blitz-Ras, 2 Thro-Ras and 2 skeletons that are usually on the pitch...the other guy tends to be random, and so my MVPs 90% (made up number) of the time go to the position players. |
|
|
Dragoonkin
Joined: Aug 02, 2003
|
  Posted:
Apr 29, 2003 - 07:06 |
|
Whoops, I remember now that the rules review said that everyone could get the MVP whether they played in the game or not, but you never know what accidentally isn't working as intended... |
|
|
cataphract
Joined: Aug 02, 2003
|
  Posted:
Apr 30, 2003 - 02:39 |
|
players who died can't get the mvp same as players who missed the game due to injury |
_________________ "the eunuch should not take pride in its chastity" |
|
Frankenstein
Joined: Aug 02, 2003
|
  Posted:
Apr 30, 2003 - 03:01 |
|
SkiJunkie wrote: | Hmm...that could be the problem. I'll have to take a closer look. I think the game rolls a random number 1 - #players on the team. If the player for that number is not eligible then it rolls again. But, when I wrote that I don't think the game allowed you to skip roster number. Now that it can, if you have 12 players but they are numbered 4-16 then it is possible that player 13-16 will never get an mvp because the die is rolling 1-12. I'll have to take a look later. |
That explanation is more than satisfactorally and explains all the available data. Stupid me, should have exploited this discovery. |
|
|
Frankenstein
Joined: Aug 02, 2003
|
  Posted:
Apr 30, 2003 - 03:06 |
|
cataphract wrote: | players who died can't get the mvp same as players who missed the game due to injury |
However, at least fumbbl-players can. |
|
|
|