logo

All times are UTC - 6 hours

Mark forums read


Post new topic Reply to topic  [ 126 posts ]  Go to page Previous  1, 2, 3, 4, 5, 6, 7  Next

Author Message
Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 9:31 am 
Name Calling Internet Bully
Name Calling Internet Bully
User avatar

Joined: Sat Sep 01, 2007 10:10 pm
Posts: 6172
Location: Gurnee, IL
I think a "Better win" is more important than a "Bad Loss".

as for Jason's two ideas, I like SoS removing the worst person. In fact I think that is incredibly simple, and could easily be inserted into the system with ease. It also would mean "full SoS" could still be used later as a further tie breaker, which I like.

SoS will take into account a Bad Loss. Obviously it's entirely possible someone someday could manage to lose to a 1-5, then beat two 5-1s or something crazy, but I don't particularly think we will need an additional increase to this in particular. In fact, I actually think removing the worst "win" from SoS should actually increase the effects of a bad loss. So I think Jason's idea actually does both things at once. Let me demonstrate.

Say someone did the following.
L 1-5
W 2-4
W 3-3
L 4-2
W 3-3
W 4-2
Sos = 17/36 or 47%
Modified with the removal of the worst win = 46%. It's actually a net drop in SoS because that huge loss takes on more importance.

Now let's look at someone who doesn't do that, but gets hit by a bad first round opponent.
W 0-6
W 3-3
W 4-2
L 6-0
W 5-1
L 5-1
SoS = 23/36 = 64%
MSoS = 77%

I think Jason has hit on the best solution.

_________________
Image

http://www.bloomilk.com/Squads/Search.aspx?UserID=29


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 10:12 am 
Major Tierce
Major Tierce

Joined: Sun Dec 09, 2007 11:21 pm
Posts: 4268
actually I was not considering full SOS minus worst win, but that could be great. I was considering SOS of your Wins minus your worst win. which is really the SOS of just your 3 best wins (assuming we are talking about 4-2s) being tied.

We will have to run some scnarios through it and see if it plays out where the most qualified person gets to the playoff (or wins the tie).

_________________
When I left you I was but the learner . . . now I am the master.


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 10:41 am 
Name Calling Internet Bully
Name Calling Internet Bully
User avatar

Joined: Sat Sep 01, 2007 10:10 pm
Posts: 6172
Location: Gurnee, IL
urbanjedi wrote:
actually I was not considering full SOS minus worst win, but that could be great. I was considering SOS of your Wins minus your worst win. which is really the SOS of just your 3 best wins (assuming we are talking about 4-2s) being tied.

We will have to run some scnarios through it and see if it plays out where the most qualified person gets to the playoff (or wins the tie).


Perhaps a providential reading then :)

_________________
Image

http://www.bloomilk.com/Squads/Search.aspx?UserID=29


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 10:43 am 
One of The Ones
One of The Ones
User avatar

Joined: Mon Oct 08, 2007 8:01 pm
Posts: 8402
Location: Chicago, IL
billiv15 wrote:
I think a "Better win" is more important than a "Bad Loss".


Totally fine with this

billiv15 wrote:
I think a "Better win" is more important than a "Bad Loss".

as for Jason's two ideas, I like SoS removing the worst person. In fact I think that is incredibly simple, and could easily be inserted into the system with ease. It also would mean "full SoS" could still be used later as a further tie breaker, which I like.


Every system can use full SoS as a later tie-breaker. I thought we always had that in mind.

I don't think it's any easier than simply counting to 1. With the system of the first tie-breaker being if someone beat an opponent with a better record, it is the quickest and easiest way to find that "better win".

More often than not, nobody at 4-2 will have beaten a 5-1. And if they have, they deserve to be elevated. In WI - Jonny would have been #4 before anything else had to be considered.

We could still use Jason's system, but only after we confirm that the other categories are tied.

A. # of games won vs opponents with a better record.

B. # of games won vs opponents with the same record.

C. Jason's modified SoS

D. Full SoS

billiv15 wrote:

SoS will take into account a Bad Loss. Obviously it's entirely possible someone someday could manage to lose to a 1-5, then beat two 5-1s or something crazy, but I don't particularly think we will need an additional increase to this in particular. In fact, I actually think removing the worst "win" from SoS should actually increase the effects of a bad loss.


You're right about it increasing the weight of a bad loss, but it does not give much additional weight to beating someone better. I think that deserves greater weight.

billiv15 wrote:
Say someone did the following.
L 1-5
W 2-4
W 3-3
L 4-2
W 3-3
W 4-2
Sos = 17/36 or 47%
Modified with the removal of the worst win = 46%. It's actually a net drop in SoS because that huge loss takes on more importance.

Now let's look at someone who doesn't do that, but gets hit by a bad first round opponent.
W 0-6
W 3-3
W 4-2
L 6-0
W 5-1
L 5-1
SoS = 23/36 = 64%
MSoS = 77%


In my system we wouldn't have to have it come down to any form of SoS, since all we had to do was see that Player B beat a 5-1, and Player A did not.

They are tied in score and record, and did not play head to head - but player B beat a 5-1. Player B advances. Easy, simple and valid.

Here's a situation where Jason's system alone breaks down:

Player A
L 5-1
W 0-6
W 3-3
L 6-0
W 3-3
W 4-2
Sos = 21/36 or 63.88%
Modified with the removal of the worst win = 70%.

Player B
W 1-5
W 1-5
W 3-3
L 6-0
W 5-1
L 5-1
SoS = 21/36 = 63.88%
MSoS = 66.66%

So they had the same overall SoS, but player A wins with Jason's system. To me clearly player B was better, because player B beat a 5-1, and lost to the same two guys Player A did.

I really think extra weight should be given to beating a 5-1


Last edited by TimmerB123 on Fri Jun 24, 2011 10:50 am, edited 3 times in total.

Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 10:43 am 
One of The Ones
One of The Ones
User avatar

Joined: Mon Oct 08, 2007 8:01 pm
Posts: 8402
Location: Chicago, IL
billiv15 wrote:
urbanjedi wrote:
actually I was not considering full SOS minus worst win, but that could be great. I was considering SOS of your Wins minus your worst win. which is really the SOS of just your 3 best wins (assuming we are talking about 4-2s) being tied.

We will have to run some scnarios through it and see if it plays out where the most qualified person gets to the playoff (or wins the tie).


Perhaps a providential reading then :)


My example above is the same either way.


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 11:44 am 
Major Tierce
Major Tierce

Joined: Sun Dec 09, 2007 11:21 pm
Posts: 4268
Every system has it's breakdowns no matter what.

From a philisophical question which is better?

Beating 3 chumps and 1 really good person

or

beating 1 chump and 3 above average people

That is basically the question that we are asking and I am not really sure which is better.
Since multiple people have stated over the years that beating someone who goes 0-6 shouldn't really count then I would say that would tend toward the player who beat the 3 above average players and not the one who beat the really good player. Which in Tims scenario means that the player who had the best win would unfortunately get eliminated.

_________________
When I left you I was but the learner . . . now I am the master.


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 11:55 am 
Major Tierce
Major Tierce

Joined: Sun Dec 09, 2007 11:21 pm
Posts: 4268
I think the basic problem is that we have to either penalize or reward someone for every game they play. Since we don't want to reward someone for beating the 0-6 guy then we are penalizing him for it and I am not sure that makes sense.

It all boils down to the fact that our pairings are random (and not sure if not random pairings would help much) that some players play a tougher schedule than others. If we reward those players for playing a tougher schedule then we are penalizing those who do not. Currently SOS rewards you for playing tougher opponents throughout the day (whether you win or lose) which is as it should be is it not?

If I went 4-2 vs all players who finished up with good records shouldn't my 4-2 be better than someone who went 4-2 vs bottomfeeders?

Looking at WI it is true that Jonny had the best win, but he also had the worst loss. Is the best win that much better than than the worst loss?

It shouldn't be how you did against 1 particular opp, but how you did over the course of the day right? 1 rd shouldn't make or break t8 for you.

Also think about the fact that if person hadn't won a particular game (even against the 5-1) that he wouldn't be tied he would be even lower (at 3-3)

_________________
When I left you I was but the learner . . . now I am the master.


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 12:00 pm 
Major Tierce
Major Tierce

Joined: Sun Dec 09, 2007 11:21 pm
Posts: 4268
1 more thing.

In all these cases Jonny is the one who "would have made it" at WI.

If we just used SOS Jonny would have made it.

Are we sure that we aren't just running around in circles making sure Jonny makes it when if we just stood still Jonny would have made it?

Maybe all we really need is to just define when h2h is used (only when person has beaten all other tied) and leave it at that.

It seems that so far everyone thinks that Jonny should have made it and under SOS (full SOS) he would have made it. I am not sure we could easily find an example (using our current 3/2 system) that SOS tiebreak winner would be different than some new theoretical tiebreak winner.

_________________
When I left you I was but the learner . . . now I am the master.


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 12:03 pm 
Name Calling Internet Bully
Name Calling Internet Bully
User avatar

Joined: Sat Sep 01, 2007 10:10 pm
Posts: 6172
Location: Gurnee, IL
The only thing I'd say Tim, is that I'm not positive beating a 5-1 is really that much better than beating a 4-2. I could easily demonstrate this problem by showing the only good win a player got was against a 5-1, whereas plenty of good players do end up 4-2. Take a look at the people who were 5-1 and those that were 4-2 last year. I'm sure if you didn't know the records, and I asked which list was list, you'd choose the 4-2s as the better group, and the more likely to have been 5-1 than the list of 5-1s. I'm not saying this to devalue that group, but simply to say that I don't agree that a single win alone unless it was against the tied person, shouldn't necessarily mean you had the best day. I do recognize that a poor person who has to play the 0-6 and then a 1-5 person would have trouble, but that is much more rare than you give it credit. And my system would eliminate about 95% of the situations, maybe higher than that actually.

_________________
Image

http://www.bloomilk.com/Squads/Search.aspx?UserID=29


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 12:09 pm 
One of The Ones
One of The Ones
User avatar

Joined: Mon Oct 08, 2007 8:01 pm
Posts: 8402
Location: Chicago, IL
urbanjedi wrote:
Every system has it's breakdowns no matter what.

From a philisophical question which is better?

Beating 3 chumps and 1 really good person

or

beating 1 chump and 3 above average people

That is basically the question that we are asking and I am not really sure which is better.
Since multiple people have stated over the years that beating someone who goes 0-6 shouldn't really count then I would say that would tend toward the player who beat the 3 above average players and not the one who beat the really good player. Which in Tims scenario means that the player who had the best win would unfortunately get eliminated.


You're right on what the base question is.

Honestly - my answer to your hypothetical is that the guy who beat the 5-1 is better. Bad luck matched him up against 3 other chumps. I truly believe that since they are tied in record and score, and H2H cannot be counted because they did not play, simply the fact that one beat a 5-1 means they should be elevated. I will stand by that firmly an confidently.

It seems from Bill and you (and I am fine with it) that beating someone good should count more than losing to someone bad.

So ESPECIALLY in light of that, then greater wins should have greater weight.

It simply doesn't weigh enough in this system IMO. I do like it as the a 3rd or 4th tie-breaker after H2H, but I still think simple wins vs better record opponents deserves greater weight.


If you accept the fact (Jason I think you mentioned it before in this thread) that in general you should lose to guys with better records, and you should beat guys with worse records, then someone should be elevated for doing something above their level. The opposite conclusion can be drawn to punish someone for losing to someone worse, but it seems we have more or less reached the consensus that greater wins should weigh more than lesser losses.

The easiest and simplest way to do this is simple counting.

# of wins vs opponents with better records.

urbanjedi wrote:
Every system has it's breakdowns no matter what.


I think this is probably true, but I easily came up with a reasonable and realistic scenario where the "wrong" player advances.

I am sincerely racking my brain to come up with a scenario where my system comes up with the "wrong" guy winning. The only scenarios I can come up with are outrageously improbable. (losing to a 1-5 record and a 2-4 record, but beating a 5-1). But once again that is considering losses, which it seems like we are leaning against.

So if we have two systems and one can break-down is moderate and reasonable circumstances, but the other breaks down in only the most extreme bizarre cases - which system is better?




Bottom line is this - all wins vs opponents with lesser records should be ELIMINATED from the equation completely. At least until the 5th or 6th tie-breaker. There are several ways to do this, but this is a strong belief of mine that the first tie-breaker after H2H needs to not have wins vs lesser opponents in the equation AT ALL.


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 12:16 pm 
One of The Ones
One of The Ones
User avatar

Joined: Mon Oct 08, 2007 8:01 pm
Posts: 8402
Location: Chicago, IL
urbanjedi wrote:
1 more thing.

In all these cases Jonny is the one who "would have made it" at WI.

If we just used SOS Jonny would have made it.

Are we sure that we aren't just running around in circles making sure Jonny makes it when if we just stood still Jonny would have made it?

Maybe all we really need is to just define when h2h is used (only when person has beaten all other tied) and leave it at that.

It seems that so far everyone thinks that Jonny should have made it and under SOS (full SOS) he would have made it. I am not sure we could easily find an example (using our current 3/2 system) that SOS tiebreak winner would be different than some new theoretical tiebreak winner.


In this example if you or Bill had a greater SoS then Jonny would not have. But in several other of our suggestions Jonny wins. So this is why it needs to change. Imagine if Bill had the best SoS - then he advances? In no other scenario we have would he advance, and the arguement was always between Jason and Jonny for that day. Jason lost to better opponents, but Jonny beat a better one. But Bill advances? This is why SoS should be a lower level tie-breaker.

Giving one example of why something works doesn't make it valid. But all you need is one example to de-bunk it.


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 12:19 pm 
One of The Ones
One of The Ones
User avatar

Joined: Mon Oct 08, 2007 8:01 pm
Posts: 8402
Location: Chicago, IL
billiv15 wrote:
I do recognize that a poor person who has to play the 0-6 and then a 1-5 person would have trouble, but that is much more rare than you give it credit.

Happened to me in 2007 Championships, admittedly why I am so passionate about it. Especially frustrating when I beat a really good player who then drops (Ben G) and would have SURELY won games had he kept playing.


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 12:24 pm 
Name Calling Internet Bully
Name Calling Internet Bully
User avatar

Joined: Sat Sep 01, 2007 10:10 pm
Posts: 6172
Location: Gurnee, IL
TimmerB123 wrote:
billiv15 wrote:
I do recognize that a poor person who has to play the 0-6 and then a 1-5 person would have trouble, but that is much more rare than you give it credit.

Happened to me in 2007 Championships, admittedly why I am so passionate about it. Especially frustrating when I beat a really good player who then drops (Ben G) and would have SURELY won games had he kept playing.


That is incorrect, Ben was 0-2, didn't hurt you nearly as bad as an 0-6. What you lost was that he might well have went 4-2 or 3-3, but 2-4 is nearly the same as 0-2 (probably 1% point difference between the two at most in total SoS). Now, take away your worst win (probably the 1-5) and see if that changed your rank. I'm pretty sure this would actually have solved that particular issue, and you would have made the top 8 as you should have.

_________________
Image

http://www.bloomilk.com/Squads/Search.aspx?UserID=29


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 12:34 pm 
One of The Ones
One of The Ones
User avatar

Joined: Mon Oct 08, 2007 8:01 pm
Posts: 8402
Location: Chicago, IL
urbanjedi wrote:
Looking at WI it is true that Jonny had the best win, but he also had the worst loss. Is the best win that much better than than the worst loss?


I am confused. I agree with this statement, and I tried to weigh them equally with the equation of:
Average records of:
a players opponents with lesser records who they lost to
AND
a players opponents with better records that they beat


Then you disagreed and said greater wins should weigh more than lesser losses.

Have we reached the circular arguments stage?


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 12:47 pm 
Hall of Fame Member
Hall of Fame Member
User avatar

Joined: Sun Sep 02, 2007 7:22 pm
Posts: 4994
urbanjedi wrote:
From a philisophical question which is better?

Beating 3 chumps and 1 really good person

or

beating 1 chump and 3 above average people



the second one IMO. In a game where luck plays such a large role consistency should be rewarded.
So he beat 3 crap guys and got lucky (possibly) against the good dude. The other guy performed consistently well against 3 strong opponents, i would give the nod to the 2nd guy.

_________________
Image
Image


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 12:50 pm 
One of The Ones
One of The Ones
User avatar

Joined: Mon Oct 08, 2007 8:01 pm
Posts: 8402
Location: Chicago, IL
billiv15 wrote:
TimmerB123 wrote:
billiv15 wrote:
I do recognize that a poor person who has to play the 0-6 and then a 1-5 person would have trouble, but that is much more rare than you give it credit.

Happened to me in 2007 Championships, admittedly why I am so passionate about it. Especially frustrating when I beat a really good player who then drops (Ben G) and would have SURELY won games had he kept playing.


That is incorrect, Ben was 0-2, didn't hurt you nearly as bad as an 0-6. What you lost was that he might well have went 4-2 or 3-3, but 2-4 is nearly the same as 0-2 (probably 1% point difference between the two at most in total SoS). Now, take away your worst win (probably the 1-5) and see if that changed your rank. I'm pretty sure this would actually have solved that particular issue, and you would have made the top 8 as you should have.



0-2 is still 0%. It does weigh less than a 0-6, that is true. But a 2-4 is still better than a 0-2. Even if he had played 1 game and won it would have been better. If they would have won ANY games, it would have been better for you.



Anyway - I am not trying to be obstinent.

I think dropping the worse record you beat from SoS is a step in the right direction. Exactly 1 step, since you are dropping the lowest win. I want to go further and drop all lower record wins until a later tie-breaker, but that being said - it is better than what we have.

1 step better


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 12:52 pm 
Major Tierce
Major Tierce

Joined: Sun Dec 09, 2007 11:21 pm
Posts: 4268
Id have to do research to see if this happened as I don't know who sepcifically the t8 lost to, but take last year's Gencon. I lost to 2 5-1 players and beat 2 other 4-2 players a 3-3 and a 2-4.

You are saying tim that if some other 4-2 had beaten one of the 5-1s but their other wins were against 1-5s or 2-4s that they should have been in over me just because they had a better win? (even though their other wins weren't as good?) This actually would have happened had JAmes won the last rd as he had beaten daniel in rd 1 or 2 but didn't have any other quility wins throughout the day.

And yes, I think Jonny should have made T4 at WI (because of SOS). He had the hardest time getting to 3-2 out of all the 3-2 players.

_________________
When I left you I was but the learner . . . now I am the master.


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 12:57 pm 
One of The Ones
One of The Ones
User avatar

Joined: Mon Oct 08, 2007 8:01 pm
Posts: 8402
Location: Chicago, IL
urbanjedi wrote:
Id have to do research to see if this happened as I don't know who sepcifically the t8 lost to, but take last year's Gencon. I lost to 2 5-1 players and beat 2 other 4-2 players a 3-3 and a 2-4.

You are saying tim that if some other 4-2 had beaten one of the 5-1s but their other wins were against 1-5s or 2-4s that they should have been in over me just because they had a better win? (even though their other wins weren't as good?) This actually would have happened had JAmes won the last rd as he had beaten daniel in rd 1 or 2 but didn't have any other quility wins throughout the day.

And yes, I think Jonny should have made T4 at WI (because of SOS). He had the hardest time getting to 3-2 out of all the 3-2 players.


Do you think Bill should have made T4 at WI if he had the best SoS? Even though he lost to you, (a lower record than anyone you lost to) and Jonny beat a better record?


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 1:28 pm 
Name Calling Internet Bully
Name Calling Internet Bully
User avatar

Joined: Sat Sep 01, 2007 10:10 pm
Posts: 6172
Location: Gurnee, IL
TimmerB123 wrote:
0-2 is still 0%. It does weigh less than a 0-6, that is true. But a 2-4 is still better than a 0-2. Even if he had played 1 game and won it would have been better. If they would have won ANY games, it would have been better for you.

Not really. Here's the math. That was a 7 round event, but looking forward, let's assume a 6 round event. Let's say your SoS ended up at 43.75%, so your opponents were a total of 14/32. If the guy went 0-6 instead of 0-2, it would be 14/36 = 39%, a loss of 5% points. But if the guy was a 2-4 instead of 0-2, that's 16/36 = 44.4% which also rounds to 44%. It's an incredibly small percentage difference between 0-2 and 2-4. But a big difference between having an 0-6, or a 1-5 than not having it. 5% points in this case was simply an addition of 4 losses (adding in the additional 0-4 from and 0-2 guy finishing the event and losing out). That's huge, and should allow a person who gets screwed in the first round to overcome other tied opponents with what we all agree is an overall equal day because SoS is similar (but the guy who played the slightly harder overall day by virtue of playing one low rank, and then 5 opponents slightly higher than the other guy with the same or similar SoS) move ahead. That's what our system should do I believe. Reward the person who played the better day.


TimmerB123 wrote:
Anyway - I am not trying to be obstinent.

I think dropping the worse record you beat from SoS is a step in the right direction. Exactly 1 step, since you are dropping the lowest win. I want to go further and drop all lower record wins until a later tie-breaker, but that being said - it is better than what we have.

1 step better
I disagree. The reason is as I said before. Just because someone ends up 3-3 or 4-2 does not mean they were easier to defeat than a single 5-1. We are looking for the better overall day, SoS does this best. Dropping the 1 outlier at the bottom end fixes SoS's worst offense. Shrinking farther actually decreases SoS from being a relevant issue, and inserts more randomness. As Deri said, we want the guy who played the better overall day, not the guy who lost one random game to a crappy player because of 3 20s, or the guy who lucked into 4-2 because he was decent, beat a bunch of bad players, and rolled 3 20s in his one tough game of the day.

I really think we have the solution in hand. I'm not sure I'm willing to do more than simply limiting H2H and inserting this modified SoS before the full SoS.

_________________
Image

http://www.bloomilk.com/Squads/Search.aspx?UserID=29


Top
 Profile  
Reply with quote  

Offline
 Post subject: Re: The tie-breaking system, and how to handle "head-to-head"
PostPosted: Fri Jun 24, 2011 1:57 pm 
One of The Ones
One of The Ones
User avatar

Joined: Mon Oct 08, 2007 8:01 pm
Posts: 8402
Location: Chicago, IL
billiv15 wrote:
TimmerB123 wrote:
0-2 is still 0%. It does weigh less than a 0-6, that is true. But a 2-4 is still better than a 0-2. Even if he had played 1 game and won it would have been better. If they would have won ANY games, it would have been better for you.

Not really. Here's the math. That was a 7 round event, but looking forward, let's assume a 6 round event. Let's say your SoS ended up at 43.75%, so your opponents were a total of 14/32. If the guy went 0-6 instead of 0-2, it would be 14/36 = 39%, a loss of 5% points. But if the guy was a 2-4 instead of 0-2, that's 16/36 = 44.4% which also rounds to 44%. It's an incredibly small percentage difference between 0-2 and 2-4. But a big difference between having an 0-6, or a 1-5 than not having it. 5% points in this case was simply an addition of 4 losses (adding in the additional 0-4 from and 0-2 guy finishing the event and losing out). That's huge, and should allow a person who gets screwed in the first round to overcome other tied opponents with what we all agree is an overall equal day because SoS is similar (but the guy who played the slightly harder overall day by virtue of playing one low rank, and then 5 opponents slightly higher than the other guy with the same or similar SoS) move ahead. That's what our system should do I believe. Reward the person who played the better day.


TimmerB123 wrote:
Anyway - I am not trying to be obstinent.

I think dropping the worse record you beat from SoS is a step in the right direction. Exactly 1 step, since you are dropping the lowest win. I want to go further and drop all lower record wins until a later tie-breaker, but that being said - it is better than what we have.

1 step better
I disagree. The reason is as I said before. Just because someone ends up 3-3 or 4-2 does not mean they were easier to defeat than a single 5-1. We are looking for the better overall day, SoS does this best. Dropping the 1 outlier at the bottom end fixes SoS's worst offense. Shrinking farther actually decreases SoS from being a relevant issue, and inserts more randomness. As Deri said, we want the guy who played the better overall day, not the guy who lost one random game to a crappy player because of 3 20s, or the guy who lucked into 4-2 because he was decent, beat a bunch of bad players, and rolled 3 20s in his one tough game of the day.

I really think we have the solution in hand. I'm not sure I'm willing to do more than simply limiting H2H and inserting this modified SoS before the full SoS.



So is this simply:
SoS minus the worst record you beat,

or as Jason originally meant:
SoS of ONLY the players you beat, minus the worst record.

I'm not sure which is better. I think I actually like the first, because then it holds you accountable for who you lose to. For our purposes Jason's method reduces it down to 3 games (4 wins minus 1), which is a much smaller sample, which is also what we're trying to avoid, right?


Top
 Profile  
Reply with quote  
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 126 posts ]  Go to page Previous  1, 2, 3, 4, 5, 6, 7  Next

All times are UTC - 6 hours

Mark forums read

Who is online

Users browsing this forum: No registered users and 51 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Jump to:  
cron
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
Jedi Knights style by Scott Stubblefield