| Poll |
| What direction should the E.L.F. take? |
| Go back to the way it used to be. |
|
19% |
[ 13 ] |
| Stay the course. Everything is just fine! |
|
19% |
[ 13 ] |
| Badges, you say... are they shiny? |
|
28% |
[ 19 ] |
| Will you listen to me for a change!!! |
|
7% |
[ 5 ] |
| The ELF is dead man. Just let it go already. |
|
24% |
[ 16 ] |
|
| Total Votes : 66 |
|
Mr_Foulscumm
Joined: Mar 05, 2005
|
 Posted:
Feb 22, 2009 - 18:08 |
 
|
| babelfilm wrote: | I am trying to understand exactly how you can cherry pick.
Can somebody explain what is considered cherry picking? |
This... hope it helps |
_________________
Everybody's favorite coach on FUMBBL |
|
westerner
Joined: Jul 02, 2008
|
| Posted:
Feb 22, 2009 - 18:27 |
|
| babelfilm wrote: | I am trying to understand exactly how you can cherry pick.
Can somebody explain what is considered cherry picking? |
I think the best definition is:
"Offering a game that you yourself would refuse" |
_________________
\x/es |
|
Jakob_olsen
Joined: Nov 20, 2005
|
| Posted:
Feb 22, 2009 - 18:45 |
|
| westerner wrote: |
I think the best definition is:
"Offering a game that you yourself would refuse" |
That is actually a very good definition !
But hard to build into a rules system  |
|
|
babelfilm
Joined: Dec 15, 2005
|
| Posted:
Feb 22, 2009 - 19:14 |
|
I asked because we all try to make the scoring system remove cherry picking.
But to be able to do that, we must have a numeric definition of the problem.
Where can I see the season XV winners matches? |
|
|
MadTias
Joined: Jun 19, 2004
|
| Posted:
Feb 22, 2009 - 19:24 |
|
| babelfilm wrote: | I asked because we all try to make the scoring system remove cherry picking.
But to be able to do that, we must have a numeric definition of the problem.
Where can I see the season XV winners matches? |
We already removed the incentive to picking. With the current baseline scoring system. Introduced in season XVI, as a response to season XV. Please read the season XV rules review. |
|
|
CircularLogic
Joined: Aug 22, 2003
|
| Posted:
Feb 22, 2009 - 19:33 |
|
babelfilm:
The most excessive cherrypicking would be playing matchups that are barely legal. If you challenge an orcteam that has 5 more TS than your elves and is coached by a competent coach and you manage a 3-0 victory causing 2 casualties, that`s an impressive feat. Now imagine how you would feel, when you see that I went and talked a CR132 coach (12 years old, barely knows the basics) into playing my elves, even though his orcs are TS 105 and my elves are TS 145. I then go on and use my DPs for 6 casualties, scoring a 7-0 victory. Is this an accomplishment for me? No. Yet I gain ALOT more points than you.
That is simply not right. Because it means in order to be good in ELF, you have to be sneaky and offer unfair matches. |
|
|
Tathar
Joined: Sep 29, 2006
|
| Posted:
Feb 22, 2009 - 20:36 |
|
| SillySod wrote: | | Personally I dislike the baseline. I would rather have some sort of system where we get .5 of a point per TS disadvantage and .3 of a point per CR over 155 (or some balanced equivalent). While we have some sort of cutoff (where its [R]-legal or the "baseline") we will always be presented with an optimal game which is very easy to determine..... i.e. it will be pretty obvious what the easiest way to get points it (either to pick at many TS or to find coaches at exactly 155 CR). Finding coaches who are "just baseline" is still cherrypicking in my eyes. |
Also with this type of scoring you could set the "Baseline" to quite a low level (i.e. 150, or 130 if you wanted) since the scoring will be relative to the bonuses from the match ups from your peers. This may encourage some coaches who were disheartened by the 155CR figure. |
|
|
Timppis
Joined: Jun 02, 2006
|
| Posted:
Feb 23, 2009 - 15:51 |
|
You know I already DID write down the ultimate scoring system that is not only excellent and fair but also totally awesome and even cool  |
|
|
babelfilm
Joined: Dec 15, 2005
|
| Posted:
Feb 24, 2009 - 01:44 |
|
First of all, I would like to apologize for my stubbornness and my last attempt at making a perfect scoring system.
| Mr_Foulscumm wrote: | | babelfilm wrote: | I am trying to understand exactly how you can cherry pick.
Can somebody explain what is considered cherry picking? |
This... hope it helps |
After reading this, and still not seing the problem, I Plotted all the games of the winners of the last 5 ELF tourney into a spreadsheet (I know, I have no life ') including TR, TS and CR at the time. And finally I saw it. Its not that I think that the coach was cherry picking, as much as trying manage his team to victory, in a broken system.
The results were quite astonishing. If anyone is interested, PM me and I will be happy to email the 11 page spreadsheet. Here I will just give you the conclusions.
With all the numbers in the system, it was easy to try out different point systems. If you look at the chart, you will see 6 different point systems. The X axis are the results of the 5 winners named after the season they won. The Y axis is points.
<img src="http://fumbbl.com/teams/260943.jpg">
<B>-XV </B>
The old point system used up until season XV. It heavily favors finding weak opponents, due to easily getting a lot of TD and CAS, but also the Shutout bonus and the 5+ TD bonus.
<B>XVI-XIX</B>
The new system put in place after a lot of complaining about some teams not playing "fair" matches. As you can see, the winner of season XV is way behind. This may or may not have been a factor in ELF losing a lot of active coaches. I know [B] might have something to do with it, but as some people have pointed out, the drop in completed cycles started before [B] was invented. This system, I believe, does not encourage new and/or lower ranking coaches to join ELF.
<B>Opp. CR > 155</B>
This is more or less the same as the new system, but without the TS factor and the following points:
<I>Baseline: 5 for games vs. CR 155 or more.
Win: 10
Draw: 5
TD for: 1
TD against: -1
CAS for: 1
5+ TD: 0
Shutout: 0
Category A: x2</I>
This seems to take out the cherry picking advantages. But it may still scare people away, if they find it daunting to play vs. “hard” opponents. And some people say that finding CR 156 coaches, is still a kind of cherry picking.
<B>TS > Opp.</B>
This is more or less the same as the new system, but without the CR factor and the following points:
<I>Baseline: 5 for games vs. teams with equal or higher TS.
Win: 10
Draw: 5
TD for: 1
TD against: -1
CAS for: 1
5+ TD: 0
Shutout: 0
Category A: x2</I>
This gives almost the same result as above. But has the problems of some teams having a notoriously low TS (halflings, ogres) and some having high TS (dwarves, undead) and thus effectively removing them from the list of viable opponents.
<B>Fractions</B>
This is the suggestion of SillySod, who had the idea of using fractions of coach rating differences to make a more fluent system.
<I>Baseline: 0,3 points per CR over 155 and 0,5 points per opponent TS advantage.
Win: 10
Draw: 5
TD for: 1
TD against: -1
CAS for: 1
5+ TD: 2
Shutout: 2
Category A: x2</I>
This performs much the same as the above 2 systems, but calculating your own points for a match is a bitch.
<B>Table</B>
This is my suggestion for a balanced point system. From looking at the above systems, we know that TS is not an ideal basis for points, but we want to acknowledge the fact that a game against a high CR coach is much harder than against a newbie. Therefore we made a sliding table system, where you can easily see how many points you will get.
<B><I>Baseline:
Opp. CR - Points
=>150 - 1
=>155 - 2
=>160 - 3
=>165 - 4
=>170 - 5
Win: 10
Draw: 5
TD for: 1
TD against: -1
CAS for: 1
5+ TD: 0
Shutout: 0
Category A: x2
I would also suggest a 10 point reward for playing the game, so a lost game gives more points, than not playing at all. </I></B>
This gives the super cherry picking team with 8-0 victories almost the same points as the much lower 3-1 victories over much harder coaches. The average opponent CR for the season XVI-XIX winners is 10 higher than the season XV winner. Remember that season XVII and XIX both have a draw in their cycle.
One of the important changes is to remove the 50 TD and shutout bonuses, as they give a lot of points for high scoring teams, thus making cherry picking necessary.
After trying about a million different combinations, this was the system most fair for all, and easiest to understand. |
Last edited by babelfilm on %b %24, %2009 - %11:%Feb; edited 1 time in total |
|
pizzamogul
Joined: Jun 13, 2005
|
| Posted:
Feb 24, 2009 - 02:08 |
|
Wow, I leave for the weekend and come back to five pages for discussion... I've got some reading to do! |
_________________
"Don't expect mercy."
-Woodstock |
|
CircularLogic
Joined: Aug 22, 2003
|
| Posted:
Feb 24, 2009 - 08:42 |
|
| babelfilm wrote: | <B><I>Baseline:
Opp. CR - Points
=>150 - 1
=>155 - 2
=>160 - 3
=>165 - 4
=>170 - 5
Win: 10
Draw: 5
TD for: 1
TD against: -1
CAS for: 1
5+ TD: 0
Shutout: 0
Category A: x2
I would also suggest a 10 point reward for playing the game, so a lost game gives more points, than not playing at all. </I></B>
This gives the super cherry picking team with 8-0 victories almost the same points as the much lower 3-1 victories over much harder coaches. The average opponent CR for the season XVI-XIX winners is 10 higher than the season XV winner. Remember that season XVII and XIX both have a draw in their cycle.
One of the important changes is to remove the 50 TD and shutout bonuses, as they give a lot of points for high scoring teams, thus making cherry picking necessary.
After trying about a million different combinations, this was the system most fair for all, and easiest to understand. |
So this system will favor the one that can find the most opponents that play his elves with a 30TS points disadvantage. Nice. Plus I would still try to find my opponent exactly in the CR150-157 range, as there I get 1 or 2 points but the skill-difference is the most worthwhile. Or I go and pick the complete noob, as it doesn`t matter if I play CR149 or CR129. |
|
|
Jakob_olsen
Joined: Nov 20, 2005
|
| Posted:
Feb 24, 2009 - 12:08 |
|
| CircularLogic wrote: |
Plus I would still try to find my opponent exactly in the CR150-157 range, as there I get 1 or 2 points but the skill-difference is the most worthwhile. Or I go and pick the complete noob, as it doesn`t matter if I play CR149 or CR129. |
But the same is true of the current ruleset. You would want to cherrypick coaches with exactly 156CR, since it does not matter if you play against CR 156 or 176... That is not fair in my mind.
I like the idea of an increasing bonus as you play better and better coaches. And you could always extend the range to start from 140, or whatever. I don't think that is a big problem.
In a system with an increasing bonus with CR, you could even end in a situation where people "reverse cherry-pick" and onlyu play against really good coaches. That would be cool wouldn't it ? |
|
|
SillySod
Joined: Oct 10, 2006
|
| Posted:
Feb 24, 2009 - 13:36 |
|
No, I think circ is correct in that that is a relatively non-incremental scale and more importantly misses out the whole issue of playing against substantially weaker teams rather than just weaker opponents.
| babelfilm wrote: | | Quote: | | This... hope it helps |
After reading this, and still not seing the problem, I Plotted all the games of the winners of the last 5 ELF tourney into a spreadsheet (I know, I have no life ') including TR, TS and CR at the time. And finally I saw it. Its not that I think that the coach was cherry picking, as much as trying manage his team to victory, in a broken system. |
The easiest way to see whether a team is picking or not it to click the "view stats" button next to the roster.... this shows the number of times played against each race but also the average TS difference against each race. If you look at the old winners Mr_Foulscumm linked you to you'll see that they played most of their games with an average advantage of 20TS! If you only look at the scoring results it probably looks even worse  As far as I can tell... none of your propsed systems actually deal with this.
As far as see it pizza has three rough options:
- KISS, back to basics with points being given for all matchups. Maybe you could give out prizes for completing the cycle vs high CR coaches or doing the Synnful Sprint or whatever.
- Baseline, keep a baseline system to prevent picking from running riot or being terribly successful. Alternatively award bonus points for all the scoring matches being vs an average of 160CR or whatever (so people have a little more flexibility to accept that 154CR game or whatever).
- Complicated, some kind of system that lets people play the games they want which is relatively resilient to optimisation. This would suit me fine but might put other people off if they feel they dont understand it or whatever so it might not be the best way forward for the group. |
_________________
Putting the "eh?" back into Sexeh.
"There are those to whom knowledge is a shield. There are those to whom it is a weapon. Neither view is balanced." |
|
babelfilm
Joined: Dec 15, 2005
|
| Posted:
Feb 24, 2009 - 17:54 |
|
| CircularLogic wrote: |
So this system will favor the one that can find the most opponents that play his elves with a 30TS points disadvantage. Nice. Plus I would still try to find my opponent exactly in the CR150-157 range, as there I get 1 or 2 points but the skill-difference is the most worthwhile. Or I go and pick the complete noob, as it doesn`t matter if I play CR149 or CR129. |
That may be true. But after extensive testing, it was the most fair, easy and balanced system. The only way to make it more fair, is to make a very advanced system that takes all factors into account. But then it would be hard to find out how to better your score.
If you want the spreadsheet, PM me. Then you can play around with some different combinations.
Actually the only way to find the best coach, is to look at the coach rating, and say that FischerKing won with CR 187.33...  |
|
|
babelfilm
Joined: Dec 15, 2005
|
| Posted:
Feb 24, 2009 - 23:47 |
|
There has been a few suggestions for new badges/medals.
So what would be cool? |
|
|
|
|
| |
Chaos Draft League R...
Search
