Hi,

as the title says, could you explain the behaviour of scores in this tournament??

I'll rewrite them here in case some game will finish and thus change them.

At this point, the three of us had played exactly two games against each other with perfectly balanced results and in this order:

Blackwood (10000) defeated flyingbelgium (10000) ====> 10500 - 9500

flyingbelgium (9500) defeated Tesctassa II (10000) ====> 10026 - 9474

Tesctassa II (9474) defeated Blackwood (10500) ======>  10028 - 9945

I reported the starting scores between brakets and the respective scores I expected in the right-hand side.

As you can see, they're different and I don't understand why. Is it something I'm missing? I used this formula to compute the variation in scores:

variation = loser's score / winner's score * 500

new loser's score = loser's score - variation

new winner's score = winner's score + variation

Thanks in advance for any explanation!

:)

All looks correct.
  Blackwood (10000) defeated flyingbelgium (10000)
    var = (10000/10000)*500 = 500
    BW=10000+500=10500,
    FB=10000-500=9500.

  flyingbelgium (9500) defeated Tesctassa II (10000)
    var = (10000/9500) *500 = 526
    FB = 9500+526 = 10026
    T = 10000-526 = 9474

  Tesctassa II (9474) defeated Blackwood (10500)
    var = (10500/9474)*500=554
    T = 9474+554 = 10028
    BW = 10500-554 = 9946

So blackwood's score is off by 1 which is strange but otherwise correct.  The order of the wins / loses is important and not the overall record of the players.

Emh.. Alpha... the table show the actual scores, which are different from those you and I computed! =D

Basically you found out the same thing, that is, scores are not those you and I expected.

=)

Quick update. I just won against Bleda the Hun who had 10000 points. I had 9972. Here's the table before the end of the game (I canceled players between me and Bleda just for clarity):

and here it is after score updates:

So something isn't going as it should or maybe it's different from what I think it should be. Either cases, can someone explain this? Thx!

(=

Ok, I think I got it. It looks like scores are updated so that at the end of the tournament they resemble the order by which games are scheduled (i.e. following round order) and not actually played (like, for example, round 3 first then round 1 and so on).

Which, by the way, is the right way, so nothing to complain about!

Am I correct? =)

Tesctassa II wrote:

Ok, I think I got it. It looks like scores are updated so that at the end of the tournament they resemble the order by which games are scheduled (i.e. following round order) and not actually played (like, for example, round 3 first then round 1 and so on).

Which, by the way, is the right way, so nothing to complain about!

Am I correct? =)

I didn't think this was the way calculations occurred, so this would be news to me or a new change.  I thought scores were calculated at the time of game completion, regardless of round (although it's been suggested to do some different stuff).

Yertle wrote:

I didn't think this was the way calculations occurred, so this would be news to me or a new change.  I thought scores were calculated at the time of game completion, regardless of round (although it's been suggested to do some different stuff).

Then I think we have to wait tom to have the final statement about this

:)

Tesctassa II wrote:

Yertle wrote:

I didn't think this was the way calculations occurred, so this would be news to me or a new change.  I thought scores were calculated at the time of game completion, regardless of round (although it's been suggested to do some different stuff).

Then I think we have to wait tom to have the final statement about this

:)

I checked the completion time of the 3 first games. T2 is right and I don't understand the calculation neither. It's weird.

Something else, related to that, is that the notification you receive by e-mail shoud give the calculation details of the tournament score and the global tournament ranking score (just like it is done for normal games). It would allow to check the calculations more easily.

Tesctassa II wrote:

Ok, I think I got it. It looks like scores are updated so that at the end of the tournament they resemble the order by which games are scheduled (i.e. following round order) and not actually played (like, for example, round 3 first then round 1 and so on).

Which, by the way, is the right way, so nothing to complain about!

Am I correct? =)

After I won against you, and after doing some calculations, I believe you are right.

Tesctassa II wrote:

Emh.. Alpha... the table show the actual scores, which are different from those you and I computed! =D

Basically you found out the same thing, that is, scores are not those you and I expected.

=)

oh, minor details aren't that important, but sleep is.  Good job and I am glad that I could verify your numbers.

Sorry to weigh in late on this one - yes the scores are based on round scheduling not by the order of game completion. So do you think this is a problem?

The downside I can see of having it done based on game completion time is that you might see players starting to game the system by deliberately advancing / delaying the time of their defeat / victory to minimize / maximize the effect on their tournament score.

tom wrote:

Sorry to weigh in late on this one - yes the scores are based on round scheduling not by the order of game completion. So do you think this is a problem?

The downside I can see of having it done based on game completion time is that you might see players starting to game the system by deliberately advancing / delaying the time of their defeat / victory to minimize / maximize the effect on their tournament score.

I tom! Thx for answering.

I don't think this is a problem at all!!! Instead I think it's the right way of using this score, since as you pointed out, some could "control" the progress of his/her game to gain advantage. Actually I've been one of the supporter for the implemented method and I've been asking for it (just for curiosity, when was it implemented?).

Thx again for answering
:)

Tesctassa II wrote:

tom wrote:

Sorry to weigh in late on this one - yes the scores are based on round scheduling not by the order of game completion. So do you think this is a problem?

The downside I can see of having it done based on game completion time is that you might see players starting to game the system by deliberately advancing / delaying the time of their defeat / victory to minimize / maximize the effect on their tournament score.

I tom! Thx for answering.

I don't think this is a problem at all!!! Instead I think it's the right way of using this score, since as you pointed out, some could "control" the progress of his/her game to gain advantage. Actually I've been one of the supporter for the implemented method and I've been asking for it (just for curiosity, when was it implemented?).

Thx again for answering
:)

Like T2, I believe it's better that way. I was surprised too as I did not know it had changed. May be a few words in the e-mail notification and on the Leaderboard should say that the score is a temporary calculation (as your score might change even without playing).

Aye, when did this change?  It hasn't always been that way has it?

Yes it's always been that way.

tom wrote:

Yes it's always been that way.

I'm not sure we knew that... http://www.wargear.net/forum/showthread/1064p1/Tournament_tiebreaker

At least according to Hugh's initial post:

CiscoKid and Viper both posted 6-1 records.  It's a round robin tournament, and among the 5 opponents they commonly defeated, Ciscokid raked due to early wins, beating many of them at scores around 10000.  Viper, upon defeating the same 5 people, beat most of them at lower scores than CK did.  For example, Alpha was 1-0 when CK beat him, and 1-3 when Viper beat him.  Alpha finished 1-5, so CK got him at his maximum score. 

That matches with date/time and not rounds of the tournament http://www.wargear.net/tournaments/view/133

Good to know!  I think it's been suggested that it goes by rounds rather than date/time (which is what it has the appearance of doing).

Yertle wrote:

I'm not sure we knew that...

I did not know that. Rounds or not, the time sensitivity remains. When you beat a particular player and how they were doing at that time will affect the outcome of score. It is difficult to describe the criterion according to which it breaks ties. The best I can come up with is that it follows "parameter estimation logic" (the right thing to do for rankings over a long period of time, but idiosyncratic over small periods of time). I maintain that this is a problem, and that the solutions mentioned in the thread

http://www.wargear.net/forum/showthread/1392p4/Solution_for_Tournament_Tiebreaker_Method

give clearer criteria according to which ties are broken. 

Hugh wrote:

Yertle wrote:

I'm not sure we knew that...

I did not know that. Rounds or not, the time sensitivity remains. When you beat a particular player and how they were doing at that time will affect the outcome of score. It is difficult to describe the criterion according to which it breaks ties. The best I can come up with is that it follows "parameter estimation logic" (the right thing to do for rankings over a long period of time, but idiosyncratic over small periods of time). I maintain that this is a problem, and that the solutions mentioned in the thread

http://www.wargear.net/forum/showthread/1392p4/Solution_for_Tournament_Tiebreaker_Method

give clearer criteria according to which ties are broken. 

Similarly, I didn't know that scoring worked this way, but I remember talking about it early on.  I think it lessens the need for a tie breaker as now the tie breaker is luck of schedule as no seeding is used for RR tournaments.  With that said, I still think a new tie breaker system is a good idea.

Yep I still need to implement the new tie breaking system as previously suggested / agreed upon.