Mr-Klipp
Joined: Aug 02, 2003
|
 Posted:
Aug 23, 2003 - 03:09 |
 
|
| Mirascael wrote: | | Mr-Klipp wrote: | | Mirascael wrote: | Christer, Klipp, no offense meant, but a formula which rates a worse player (Halfling with Catch and MA+1) 25% higher than a better player (Goblin with Catch, has AV7, Fling has AV 6, else they are identical) is pointless.
I appreciate what you have tried to accomplish, but I think the formula is somewhat random. |
First off, a halfling with catch and + move is debatably better than a goblin, as he only has one point less av, and has a usefull skill the goblin lacks. Besides, what you are saying is that you can find a few questionable values and think that makes the whole system pointless, while all it means at most is that there are a few special cases that need closer looking at. Following your logic, I could point to a handfull of cases like Enron and the savings and loan scandals of years past and say that that is more than enough proof that capitalism is a flawed system and should be tossed out. What is needed is examination of the few remaning cases where there are odd values, not a rewrite because you find a few odd values. |
Mr_Klipp, the Goblin here has Catch like the halfling. The only difference is that the Goblin is just better due to the higher armour. Thus, a functional STR-formula would rank him higher. |
I missed that, but as pointed out above the comparison using normal tr rules still rates the halfling higher.
| Mirascael wrote: |
but the claim that it would be superior to TR is not justified (with the exception of leaving out mng's and cash and neglecting CL's and AC's, that is). |
Sorry, but you are wrong. Look at the actual stats instead of complaining about a few inconsistancies and you will see that str is doing a damn fine job of predicting game outcomes. |
|
|
MickeX
Joined: Aug 02, 2003
|
| Posted:
Aug 23, 2003 - 03:47 |
|
Sure, the STR rating needs tuning in. But it's way better than TR.
I think one step further would be to use some kind of multiplier function between MA, ST and AG. The stat combo is what makes Luthor and Griff such superstars. The MA 8 of a skink is bvery nice, but MA 8 is fantastic on an AG4 player or a strong player like Griff.
MA and AG decides how much use you will have of the players stats and skills, since it determines how often and how easily you will be able to put the player to use at the critical tasks. |
|
|
Mr-Klipp
Joined: Aug 02, 2003
|
| Posted:
Aug 23, 2003 - 03:58 |
|
| MickeX wrote: | Sure, the STR rating needs tuning in. But it's way better than TR.
I think one step further would be to use some kind of multiplier function between MA, ST and AG. The stat combo is what makes Luthor and Griff such superstars. The MA 8 of a skink is bvery nice, but MA 8 is fantastic on an AG4 player or a strong player like Griff.
MA and AG decides how much use you will have of the players stats and skills, since it determines how often and how easily you will be able to put the player to use at the critical tasks. |
Take a look at the formula str is based on, this is already taken into account. |
|
|
EvolveToAnarchism
Joined: Aug 02, 2003
|
| Posted:
Aug 23, 2003 - 21:02 |
|
| Quote: | | Following your logic, I could point to a handfull of cases like Enron and the savings and loan scandals of years past and say that that is more than enough proof that capitalism is a flawed system and should be tossed out. |
Mr-Klipp, if you wanted to prove the OPPOSITE side of the argument you seem to support, you've done it, with this brilliant piece of rhetoric! 
Enron and co. are not examples of capitalism. They are example of state-sponsored capitalism. Very different beasts but with equally tragic flaws. None the less, the system in the American Empire should be thrown out very soon before it results in the end of human evolution via extinction.
I'll throw in a comment about the STR-formula so I'm not completely off-topic. What statistical data (specific examples would be appreciated) are you using to justify that STR is a more accurate predictor of winning % than other measures (like TR-modifiers for niggles and MNGs)? |
|
|
Mr-Klipp
Joined: Aug 02, 2003
|
| Posted:
Aug 23, 2003 - 21:21 |
|
| EvolveToAnarchism wrote: | | Quote: | | Following your logic, I could point to a handfull of cases like Enron and the savings and loan scandals of years past and say that that is more than enough proof that capitalism is a flawed system and should be tossed out. |
Mr-Klipp, if you wanted to prove the OPPOSITE side of the argument you seem to support, you've done it, with this brilliant piece of rhetoric!
|
First, regardless of any opinions on the merits of capitalism, a small handfull of examples is far from enough to consider the complete system flawed, and that is the point I was making.
| EvolveToAnarchism wrote: |
I'll throw in a comment about the STR-formula so I'm not completely off-topic. What statistical data (specific examples would be appreciated) are you using to justify that STR is a more accurate predictor of winning % than other measures (like TR-modifiers for niggles and MNGs)? |
I don't have specific examples handy, but for those who forget, str used to be based on simply tr - cash with modifiers for niggles and MNG, and we had just as many complaints then saying people didn't like it. What I am coming to realize, is that no matter what system is put in place there will be those to complain about it. No system is perfect, and no one with any sense would argue that str is less accurate than the TR system it is being used to replace. While I mentioned that I don't have comparative figures, looking purely at the statistics for how STR predicted win percentage compares with actual win percentage should leave no doubts that the system is working very well overall. |
|
|
Mezir
Joined: Aug 02, 2003
|
| Posted:
Aug 23, 2003 - 21:35 |
|
For the statistics you need, contact Christer, he has the page somewhere.
Using win probability of certain strs playing each other, the games that have been played correspond almost exactly to the calculated win probability. When using TR, they are horrible (naturally), and when using STR = TR - (cash + mng + niggles) they aren't nearly as close to the probability values.
As such, these are statistics speaking greatly in favour of the current str system. |
_________________
Build a man a fire and he's warm for a day; set a man on fire and he's warm for the rest of his life. |
|
neverdodge
Joined: Aug 02, 2003
|
| Posted:
Aug 23, 2003 - 22:30 |
|
For those who don t like the STR formula, just challenge some teams in your TR range ... After all, TR is the LRB base to determine teams power.
Personnaly, i take a look at both TR an STR, and try to play team taht are approximatively in my range (about 20 pt diff) What i look too, is the number of game they ve played. A team with STR 150 and 15 game, i find it stronger than one say with STR 150 and 8 games . |
|
|
HoboJed
Joined: Aug 02, 2003
|
| Posted:
Aug 23, 2003 - 22:57 |
|
| Propmaster wrote: | | Looking ahead one skill choice, for example, the +MA mummy can get block, while the block mummy will likely get another regular skill. |
Why look ahead one skill choice? If the formula is designed to show the likelyhood that a team as it is now will win a match, how is the fact that a movement 4 mummy will be more likely to get block than another is to get a movement increase relevant? The fact one can get +1 MA and block should be taken into account when it has +1 MA and block, and not before.
Not that I think the formula is bad (or even works in the way you described), I'm just saying that player's value should represent what they are, and not what they potentialy can be, and using their potential as an excuse for higher value isn't justified. |
|
|
EvolveToAnarchism
Joined: Aug 02, 2003
|
| Posted:
Aug 23, 2003 - 23:01 |
|
| Quote: | | First, regardless of any opinions on the merits of capitalism, a small handfull of examples is far from enough to consider the complete system flawed, and that is the point I was making. |
Whether it is a small handful of examples or systemic flaws is open to debate in both cases. I'd to think that anyone who gives some serious thought on the matter would realize that Enron and co. are not just rare exceptions but symptomatic of a systemic flaw that merits state-sponsored capitalism's rejection. When a system repeatedly rewards/punishes people based largely on their luck of the draw in the parent lottery, I think the system is fatally flawed (I could go on for hours about the flaws in the system but I'll bring it back to BB soon enough). The rags to riches story of the American Dream is largely that a dream. An exception that proves the rule that it's not what you do but who your parents are that determines your social class.
| Quote: | | For the statistics you need, contact Christer, he has the page somewhere. |
I sure hope it's not just this one page that he's relying on to justify the STR formula. Because I've seen the one you are referring to (It might be useful to post a link to this page) and it allows ample room for systemic flaws to enter the formula. It would only be fair if any of the systematic flaws are made public.
PUBLIC SERVICE ANNOUNCEMENT: STR is completely broken in the STUNTY LEEG! 
| Quote: | | No system is perfect, and no one with any sense would argue that str is less accurate than the TR system it is being used to replace. |
Which is not something I'd advocate. I'm interested in seeing if there are any systemic flaws in the system. Mirascael, seems to have pointed out some. And I wouldn't be suprised if there were many more to still discover.
| Quote: | | As such, these are statistics speaking greatly in favour of the current str system. |
| Quote: | | While I mentioned that I don't have comparative figures, looking purely at the statistics for how STR predicted win percentage compares with actual win percentage should leave no doubts that the system is working very well overall. |
This is a point that irks me. People seem to be convinced that this one stats pages proves that the STR system is working very well. I think you could both benefit at looking at the issue with a little Socratic humility. I'd agree that it definitely works better than TR. But you'll need to show me much more data to convince me that there are no systemic flaws. I'm not saying that I know that there are systemic flaws, just that I heard anything that has showed me that these flaws have been hunted down by a good statistical analysis of the data. That one page summary of the data is the kind of evidence that is evident in the worst popularizations of scientific research.
I know the above may be a little abstract for some people to follow so I'll try to flesh it out with some hypothetical examples. Someone who's actually gives a damn about rankings might want to do some concrete research.
Hypothetical Data (You can access TR100 data if you really want to do this boring work on real data):
TR100 Halfling Teams average STR is 110 which is clearly above average for a TR100 team.
TR100 Halfling Teams almost always play vs teams within 10 STR.
TR100 Halfling Teams win % is 20/10/70 which is clearly below average.
With STR110 Halfling teams should be winning more than they are losing but they don't.
Question: Is this one page statistical summary influenced at all by this systemic flaw?
Answer: NO! All these matches get lumped in together with all the matches that are played between teams who are within 15 STR (I could be wrong about the specific spread but insert the appropriate spread form this one page of stats that "shows the STR system is working very well overall"). According to this page it really doesn't matter who wins these games,that will go into the middle of this stats page, as the probability SHOULD be the same that either team wins. This probably is compounded throughout this "justifying" stats pages as a lot of games get played within that STR spread where it doesn't matter who wins.
Has anyone done a statistical analysis of the data to look for such bugs as the hypothetical one I've proposed above?
As Always,
Evolve To Rigorously Logical Thinking
P.S. I do have a background in stats and logic. I never found a need to master the more rigorous computer-assisted statistical programs that would be required to do the analysis that I'm suggesting (Although stats students might enjoy the challenge). But from those countless hours of boredom in my stats class, I did retain the ability to detect statistical evidence that just doesn't prove what it claims to prove. |
|
|
Mr-Klipp
Joined: Aug 02, 2003
|
| Posted:
Aug 24, 2003 - 02:34 |
|
| EvolveToAnarchism wrote: | | Quote: | | First, regardless of any opinions on the merits of capitalism, a small handfull of examples is far from enough to consider the complete system flawed, and that is the point I was making. |
Whether it is a small handful of examples or systemic flaws is open to debate in both cases. |
No it's not, as that was totaly irrelevant to the point I was making. That point being, a few problem examples do not prove the complete failure of a system.
| EvolveToAnarchism wrote: |
PUBLIC SERVICE ANNOUNCEMENT: STR is completely broken in the STUNTY LEEG!
|
And it will contine to be so, as stunty leeg is crazy rules breaking mess. Just the way we like it.
| EvolveToAnarchism wrote: |
But you'll need to show me much more data to convince me that there are no systemic flaws. I'm not saying that I know that there are systemic flaws, just that I heard anything that has showed me that these flaws have been hunted down by a good statistical analysis of the data.
|
As for availablity of data, the complete formula is open, and you are more than welcome to analyze how it works to spot any flaws you might beleive it to contain.
| EvolveToAnarchism wrote: |
Has anyone done a statistical analysis of the data to look for such bugs as the hypothetical one I've proposed above?
|
As mentioned above, both the formula and all sorts of game statistics are available to anyone wanting to attempt said analysis. |
|
|
MickeX
Joined: Aug 02, 2003
|
| Posted:
Aug 24, 2003 - 03:35 |
|
| Mr-Klipp wrote: | | MickeX wrote: |
I think one step further would be to use some kind of multiplier function between MA, ST and AG. The stat combo is what makes Luthor and Griff such superstars. |
Take a look at the formula str is based on, this is already taken into account. |
It is? I read it again but couldn't find anything but the first table, which is only adding up the bonuses.
Thinking about it, if we could get a statistician (?) on it, you could probably compute a pretty exact formula using the vast statistics produced by FUMBBL. Could be a nice project for some mad mathematician out there...
Micke |
|
|
|
|
| |
Secret League Americ...
Search
