New Scoring System

[T Duquette]

Im hoping that my man Slythex or somebody else can help me out on this. I have a question regarding the averaging and the elimination of 'outliers' (a statistical term) in the new system:

After the two outliers are eliminated, are the scores of the remaining three judges rounded? For example if judge 1 and 2 are eliminated, and judge 3 has a score of 21 for the red corner, judge 4 has a score of 21 and judge 5 has a score of 23, the average for the red corner will be 21.66666667 Do they round this up to 22 or do they just keep it as 21... And why don't they just keep it as 21.66666667 or at least as 21.7?

As somebody who lost on a tie breaker to the eventual Olympian in the Olympic Trials I am very interested in this answer...

And if the average score of a bout is 22 to 22 and one boxer led on the pre-averaged score card by an actual UNAVERAGED score of 22>(j3=21, j4=21, j5=24) to 21.66666667 (j3=21, j4=21, j5=23, then how is it fair to go to some other criteria for deciding who the winner is when the winner should be the boxer with the highest averaged score?

Hope I didn't confuse anybody but I have to labor through a lot of math at my school and this has been something that has bothered me since the introduction of the new scoring system, any insight is appreciated...

To be honest, I guess I am really confused. I guess what I m really asking is, do they average the score of the three judges that they average together to the nearest tenth? And if they do, which I assume they do because every score that I have seen has been an even score, why??

[Dennis]

It is a good question, but unfortunately the info provided by AIBA and USAB to the public never answered it. USAB should have had a clinic during the day on how the scoring system works with demonstrations. They could have a sparring asession with judges scoring it and the coaches could see the computer tech's monitor on a big screen and see for themselves how it works in reality not just in theory. They could also have coaches and boxers sit down and try to score for themselves and quickly learn what punches they can see land and which ones they can't.

There is so much down time at tournaments that could be utilized to improve our sport.

[Slythex]

Ok, let's see what we have here:

After the two outliers are eliminated, are the scores of the remaining three judges rounded? For example if judge 1 and 2 are eliminated, and judge 3 has a score of 21 for the red corner, judge 4 has a score of 21 and judge 5 has a score of 23, the average for the red corner will be 21.66666667 Do they round this up to 22 or do they just keep it as 21... And why don't they just keep it as 21.66666667 or at least as 21.7?

First off, the outliers are eliminated on a round by round basis. So while it's possible to get 22 blows scored in a round, it's unlikely. You do NOT apply the outlier details to the total raw score. But for a given round, yes, the average is rounded. In theory, because the score is supposed to represent an actual scored blow. You can't really have 2.5 blows (I'm not sure I agree with this logic, but that's the rationale I believe).

But let me walk through an example for one round:
For red, we have 7,7,9,13,14
The outliers are 13,14 so we take the average of 7+7+9 or 7.66 repeating. This rounds to 8.
So for this round, red receives a score of 8

For blue, we have 5,8,11,12,15. Both 8, 11, 12 and 11,12, 15 are equally similar. So we apply the trimmed mean process. Therefore our outliers are 5 and 15. We get 8+11+12=31 which averages to 10.333 repeating, rounds to 10.

In this example, for this round, we have a score of 8 red, 10 blue.
Each end of round score is totalled for the final total. Then, and ONLY then, if there's a tie, they go to the tiebreaker.

To address the tiebreaker, you don't have outliers. ALL raw judge scores are used. I think that's where the confusion lies. To use your example above, judges 1 and 2 would not be eliminated from having their raw scores apply in the tiebreaker. Also, the raw scores are added and not averaged, so you don't have decimals anyway. Level 1 tiebreaker, would be J1+J2+21+21+23 for a total, no average, no rounding.

Moving on to address Dennis:

It is a good question, but unfortunately the info provided by AIBA and USAB to the public never answered it. USAB should have had a clinic during the day on how the scoring system works with demonstrations. They could have a sparring asession with judges scoring it and the coaches could see the computer tech's monitor on a big screen and see for themselves how it works in reality not just in theory. They could also have coaches and boxers sit down and try to score for themselves and quickly learn what punches they can see land and which ones they can't.

There is so much down time at tournaments that could be utilized to improve our sport.

First off, being one of the ESS techs, I've tried to publish as much information as I know. I know that USAB is not always viewed as the most transparent organization, and at least as far as I'm concerned, though, I refuse to be that way. I will gladly share any information I have. Specifically, I posted the details of the AIBA documentation (which was iffy but all we had) as soon as I had it on http://boxrec.com/forum/viewtopic.php?f=6&t=134089
This was also crossposted to the USAB web site. It included my contact information. Want to guess how many inquiries I got? Zero.

We received the software in late April, at which point in time I tested it extensively, although I lost some time to a license key issue requiring tech support from the vendor, Swiss Timing. USAB asked that I give a clinic of precisely the type you mentioned to the national coaches at their coaches retreat in early July, which I did. This was only a few weeks prior to Nationals, so there wasn't that much time in between. I was also asked to give a short demo at JOs and did so between the afternoon and evening Olympic Trial sessions that Monday. I've offered to share my presentation powerpoint for the national coaches to anyone who asked. Just send me a private message here on boxrec and I'll send it your way(oh hell, you can just email me at [email protected], that's fine too). I don't think there's a way to upload it here, but I'd do that if it were possible. I've posted extensively about the ESS on boxrec and try and answer any ESS questions that arise.

So I understand your frustration, but I've tried as much as possible to explain and be as open and forthcoming as possible regarding the computer scoring system. I'm a bit defensive about this, as you can tell, because I too get annoyed when I think USAB is not being transparent, and I've actively tried to NOT be like that.

As for a clinic during either Nationals or Olympic Trials? I guess I could have done that. No one asked. Although the life of an ESS tech has less down time than you might expect as there's a fair bit of administravia that happens both before and after the sessions of the day.

All of that said, I'm the technical lead for the ESS team, and I will be glad to answer any and all questions regarding how the system works at any time. Again, just email me at [email protected].

-John

[DCAmateurBoxing]

As somebody who lost on a tie breaker to the eventual Olympian in the Olympic Trials I am very interested in this answer...

I can tell you that we had a boxer who won on 4/5 judges cards (raw score) but lost the fight because of the accepted score. So if it was on paper, he would have won 4-1. But the accepted score had him losing 25-24. I'm sure that some of the 1pt and 2pt decisions and level I tiebreakers may have been a similar case. The system is not perfect but I think it's better than the previous ones. It leaves room for improvement and you just have to know how it works.

[T Duquette]

I can tell you that we had a boxer who won on 4/5 judges cards (raw score) but lost the fight because of the accepted score.

How is this possible?

The system is not perfect but I think it's better than the previous ones. It leaves room for improvement and you just have to know how it works.

I agree, I think its better but can be improved

[DCAmateurBoxing]

I can tell you that we had a boxer who won on 4/5 judges cards (raw score) but lost the fight because of the accepted score.

How is this possible?

Red corner (opponent) Blue corner (our boxer)
Judge 1:
1st round 9 red 6 blue
2nd round 11 red 20 blue
3rd round 12 red 10 blue
Total: 32 red to 36 blue BLUE WINS

Judge 2:
1st round 6 red 8 blue
2nd round 8 red 20 blue
3rd round 10 red 15 blue
Total: 24 red 43 blue BLUE WINS


Judge 3:
1st round 6 red 2 blue
2nd round 7 red 6 blue
3rd round 14 red 6 blue
Total: 27 red 14 blue RED WINS

Judge 4:
1st round 4 red 6 blue
2nd round 5 red 13 blue
3rd round 4 red 10 blue
Total: 13 red 29 blue BLUE WINS

Judge 5:
1st round 3 red 5 blue
2nd round 2 red 9 blue
3rd round 1 red 4 blue
Total: 6 red 18 blue BLUE WINS

But after the magical act of taking the most similar scores and then adding 2 pts becuase of the warning, our boxer loses 25 to 24. On paper if you added the 2pts, then it still wouldn't have made a difference. I also think that judge 3 should not be judging.

[byrdman66]

The scoring system sucks.. At the Trail there where way to many TIES// there was many questionable bouts..

THIS IS MY OPION : The BOTTOM LINE IS THAT USABOXING DOSE NOT WHAT ANY RULES BROKEN// BUT IF THEY BREAK THE RULES THEN IT OK..... LET PLAY COVER UP....

I

[DCAmateurBoxing]

Keep in mind this new system is coming down from AIBA. USAB can't control or change anything, so the only thing I think we can do is educate ourselves about how it works (or doesn't in some cases). The only problem I really have is that if judges scores are tossed during the accepted score calculation why are those same judges allowed to be included in the level one tiebreaker. That is dumb to me as you already said their scores shouldn't count towards determining who wins the bout. Also, unless I'm missing something, if they add the warning pts at the end of the totals, that is actually weighing the warning more than it would ordinarily if you added the 2pt in the round that it occurred. If I add 2 pts. to the red corner in the last round, then the blue corner wins even on the new system. . .

John, don't you agree my calculation? Am I missing something?

[Dennis]

Kullen,
I took a look at the scores you listed and I don't see how Red could win even with the new scoring system. Maybe I'm doing something wrong, but I have the scoring as follows:
Round 1:
Red: 9,6,6,4,3
Blue: 6,8,2,6,5
The most similar scores: Red: 6,6,4=16/3=5 Blue: 6,6,5=17/3=6

Round 2:
Red: 11,8,7,5,2
Blue: 20,20,6,13,9
The most similar scores: Red: 8,7,5=20/3=7 Blue: 20,13,9=42/3=14

Round 3:
Red: 12,10,14,4,1
Blue: 10,15,6,10,4
The most similar scores: Red: 12,10,14=36/3=12 Blue: 10,6,10=26/3=9

Total Score: Rd 1: Red-5 Blue-6; Rd 2: Red-7 Blue-14; Rd 3: Red-12 Blue-9
Total: Red-24 Blue-29 then add the two points to Red and the score is RED-26 BLUE-29 and Blue WINS.

[Slythex]

First this one:

The scoring system sucks.. At the Trail there where way to many TIES// there was many questionable bouts..

THIS IS MY OPION : The BOTTOM LINE IS THAT USABOXING DOSE NOT WHAT ANY RULES BROKEN// BUT IF THEY BREAK THE RULES THEN IT OK..... LET PLAY COVER UP....

I

Can't speak to general rule breakages or cover up or whatever, but to the best of my knowledge, there were no grievances with respect to the ESS scores. If there was some scoring system cover up, they did a good job, since even I don't know about it, and I'm the tech lead for the ESS group :-)

Not everyone was happy with the outcome, and yes, there were a lot of ties, but there was no cover up of any sort with these. Why were the scores so close? I don't know. I suspect a few factors: a) I think that with similar scores averaged the way they are, the overall effect is a normalizing effect so you get closer scores; b) to some degree, we had the highest caliber boxers, and therefore the disparity in skill was much less pronounced so given ANY scoring system, it'd be fairly close. Remember that we didn't have as many close scores at USAB Nationals, or the JOs (although with the JOs I think there was a much bigger skill difference). I wasn't at the Last Chance, so I can't speak to that, but I don't recall hearing about lots of close scores.

I can tell you that we had a boxer who won on 4/5 judges cards (raw score) but lost the fight because of the accepted score.

How is this possible?

DC explained how, but the short version is that remember, each ROUND is scored individually. So from a math perspective, the final totals are only relevant in a tiebreaker.

Just a ridiculous example to illustrate the point.
Rd1-
Red: j1,j2,j3,j4,j5 100, 100, 5, 5, 5 --> similar score is 5
Blue j1,j2,j3,j4,j5 0, 0, 6, 6, 6 --> similar score is 6
blue up 6-5 for round 1

Rd2-
Red: j1,j2,j3,j4,j5 5, 5, 100, 100, 5 --> similar score is 5
Blue: j1,j2,j3,j4,j5 6, 6, 0, 0, 6 --> similar score is 6
rd2 totals, 6-5 for blue, cumulative round totals after 2: red 10, blue 12

Rd3 (you can probably see where this is going now)-
Red: j1,j2,j3,j4,j5 5, 5, 5, 6, 100 --> similar score is 5
Blue: j1,j2,j3,j4,j5 6, 6, 6, 0, 0 --> similar score is 6
rd3 total, 6-5 for blue, cumulative round totals after 3, red 15, blue 18, blue wins.

BUT the totals for ALL FIVE judges is for red:
J1: 110 - 12 red
J2: 110 - 12 red
J3: 110 - 12 red
J4: 111 - 6 red
J5: 110 - 12 red

Blue wins (and should, since those scores are ridiculous)
There are even non-ridiculous examples where this could happen, but it does require sort of extreme scores from some judges.

On the level 1 tiebreaker and not taking similar scores: I do know that an earlier version of the software dropped the high and low. This was used in one of the Pan Am qualifiers. I have no idea why it changed, but it's reasonable to assume that just as we found some weird cases here, they had some strange ones there prompting the change. That change coupled with the law of unintended consequences, and we get some weird (but legitimate according to the mathematics procedure) results.

As far as warnings, they are added to the end-of-round score for the round in which the warning occurred. So if the raw score for round 2 was 7-3, with a warning to red, the end-of-round score would actually be 7-5 (3+2 for the warning = 5). They do not affect the individual judge scores (i.e. not factored in the similar score process) nor their raw scores if used in a level 1 tiebreaker.

Makes sense? (not saying you should agree, but does this at least explain how the system works).

Once again going to add the disclaimer that I'm a tech, and I understand the system, but that doesn't mean I necessarily like it. I think you get the result in a large majority of the cases, but there are some weird possibilities. I don't think AIBA bothered to test this in practice, really. Just came up with rules, programmed them, and ran some test bouts. This is one of the reasons I made my excel simulator. I was able to take a bunch of scorecards from my LBC, enter them to get real data, and see how they'd work out under the new scoring system. (and I'll gladly make this excel simulator available as well. I've been sending it with the powerpoint)

[Slythex]

By the way, where's Mel? I keep waiting for her to chime in and bust my chops over something or other :-)

[T Duquette]

John, since you obviously know the most about the new system, what if any changes would you make to it if you could?

[donnellon]

First this one:

The scoring system sucks.. At the Trail there where way to many TIES// there was many questionable bouts..

THIS IS MY OPION : The BOTTOM LINE IS THAT USABOXING DOSE NOT WHAT ANY RULES BROKEN// BUT IF THEY BREAK THE RULES THEN IT OK..... LET PLAY COVER UP....

I

Can't speak to general rule breakages or cover up or whatever, but to the best of my knowledge, there were no grievances with respect to the ESS scores. If there was some scoring system cover up, they did a good job, since even I don't know about it, and I'm the tech lead for the ESS group :-)

Not everyone was happy with the outcome, and yes, there were a lot of ties, but there was no cover up of any sort with these. Why were the scores so close? I don't know. I suspect a few factors: a) I think that with similar scores averaged the way they are, the overall effect is a normalizing effect so you get closer scores; b) to some degree, we had the highest caliber boxers, and therefore the disparity in skill was much less pronounced so given ANY scoring system, it'd be fairly close. Remember that we didn't have as many close scores at USAB Nationals, or the JOs (although with the JOs I think there was a much bigger skill difference). I wasn't at the Last Chance, so I can't speak to that, but I don't recall hearing about lots of close scores.

I can tell you that we had a boxer who won on 4/5 judges cards (raw score) but lost the fight because of the accepted score.

How is this possible?

DC explained how, but the short version is that remember, each ROUND is scored individually. So from a math perspective, the final totals are only relevant in a tiebreaker.

Just a ridiculous example to illustrate the point.
Rd1-
Red: j1,j2,j3,j4,j5 100, 100, 5, 5, 5 --> similar score is 5
Blue j1,j2,j3,j4,j5 0, 0, 6, 6, 6 --> similar score is 6
blue up 6-5 for round 1

Rd2-
Red: j1,j2,j3,j4,j5 5, 5, 100, 100, 5 --> similar score is 5
Blue: j1,j2,j3,j4,j5 6, 6, 0, 0, 6 --> similar score is 6
rd2 totals, 6-5 for blue, cumulative round totals after 2: red 10, blue 12

Rd3 (you can probably see where this is going now)-
Red: j1,j2,j3,j4,j5 5, 5, 5, 6, 100 --> similar score is 5
Blue: j1,j2,j3,j4,j5 6, 6, 6, 0, 0 --> similar score is 6
rd3 total, 6-5 for blue, cumulative round totals after 3, red 15, blue 18, blue wins.

BUT the totals for ALL FIVE judges is for red:
J1: 110 - 12 red
J2: 110 - 12 red
J3: 110 - 12 red
J4: 111 - 6 red
J5: 110 - 12 red

Blue wins (and should, since those scores are ridiculous)
There are even non-ridiculous examples where this could happen, but it does require sort of extreme scores from some judges.

On the level 1 tiebreaker and not taking similar scores: I do know that an earlier version of the software dropped the high and low. This was used in one of the Pan Am qualifiers. I have no idea why it changed, but it's reasonable to assume that just as we found some weird cases here, they had some strange ones there prompting the change. That change coupled with the law of unintended consequences, and we get some weird (but legitimate according to the mathematics procedure) results.

As far as warnings, they are added to the end-of-round score for the round in which the warning occurred. So if the raw score for round 2 was 7-3, with a warning to red, the end-of-round score would actually be 7-5 (3+2 for the warning = 5). They do not affect the individual judge scores (i.e. not factored in the similar score process) nor their raw scores if used in a level 1 tiebreaker.

Makes sense? (not saying you should agree, but does this at least explain how the system works).

Once again going to add the disclaimer that I'm a tech, and I understand the system, but that doesn't mean I necessarily like it. I think you get the result in a large majority of the cases, but there are some weird possibilities. I don't think AIBA bothered to test this in practice, really. Just came up with rules, programmed them, and ran some test bouts. This is one of the reasons I made my excel simulator. I was able to take a bunch of scorecards from my LBC, enter them to get real data, and see how they'd work out under the new scoring system. (and I'll gladly make this excel simulator available as well. I've been sending it with the powerpoint)

Do I understand correctly that a warning ceases to have any effect on accepted scores?

[DCAmateurBoxing]

Kullen,
I took a look at the scores you listed and I don't see how Red could win even with the new scoring system. Maybe I'm doing something wrong, but I have the scoring as follows:
Round 1:
Red: 9,6,6,4,3
Blue: 6,8,2,6,5
The most similar scores: Red: 6,6,4=16/3=5 Blue: 6,6,5=17/3=6
CORRECT

Round 2:
Red: 11,8,7,5,2
Blue: 20,20,6,13,9
The most similar scores: Red: 8,7,5=20/3=7 Blue: 20,13,9=42/3=14
CORRECT RED, BLUE IS 6,13,9=28/3=9

Round 3:
Red: 12,10,14,4,1
Blue: 10,15,6,10,4
The most similar scores: Red: 12,10,14=36/3=12 Blue: 10,6,10=26/3=9
CORRECT

Total Score: Rd 1: Red-5 Blue-6; Rd 2: Red-7 Blue-14; Rd 3: Red-12 Blue-9
TOTAL SCORE: RD 1: RED-5 BLUE-6; RD2: RED-7 BLUE-9; RD 3: RED-12 BLUE-9

Total: Red-24 Blue-25 then add the two points to Red and the score is RED-26 BLUE-29 and Blue WINS.
TOTAL: RED-24 BLUE-24 THEN ADD THE TWO POINTS TO RED AND THE SCORE IS RED-26- BLUE 24. RED WINS

ALSO TO SHOW YOU THE RD TWO BLUE MOST SIMILAR SCORE CALCULATION, YOU HAVE TWO SETS OF NUMBERS S1(20,13,9) AND S2(13,9,6) THAT AT FIRST GLANCE LOOK LIKE THE MOST SIMILAR. THE COMPUTER ACTUALLY TAKES ALL COMBINATIONS OF THREE OUT OF THE FIVE SCORES AND DOES THIS.

FIRST YOU TAKE THE AVERAGE OF EACH SET:
S1(20,13,9) = 42/3 = 14
S2(13,9,6) = 28/3 = 9.33

THEN YOU TAKE THE ABSOLUTE (IGNORE POSITIVE OR NEGATIVE) DIFFERENCE OF EACH NUMBER WITHIN THE SET FROM THE AVERAGE:
S1
20-14= 6
13-14= 1
9-14 = 5

S2
13-9.33= 3.66
9-9.33=.33
6-9.33=3.33

SO OUT OF THE TWO SETS OF NUMBERS S1 HAS THE SCORE WITH THE HIGHEST DIFFERENCE FROM THE AVERAGE (I.E. 6) SO S2 ARE THE MOST SIMILAR SCORES TO BE USED.

JOHN, I THINK WE ARE BACK TO SQUARE ONE WITH THESE SCORES. ROUND 2 RED CORNER.

[Slythex]

Do I understand correctly that a warning ceases to have any effect on accepted scores?

Not exactly. A warning is not included with the individual judge scores.
It is added at the end of the round to the calculated scores.
By this, I mean, you do the similar scores/trimmed mean and get, say Red 7, Blue 4 for round 2.
THEN because Red got a warning, the end-of-round score is actually Red 7, Blue 6 (4+2).
Does that make sense?
Warnings are added on AFTER and are not reflected in each judge's raw numbers.

Round 2:
Red: 11,8,7,5,2
Blue: 20,20,6,13,9
The most similar scores: Red: 8,7,5=20/3=7 Blue: 20,13,9=42/3=14
THE MOST SIMILAR SCORES: RED: 8,7,5=20/3=6.66 BUT WHEN I REVIEWED WITH JOHN THIS WAS LISTED AS 6, BLUE IS 6,13,9=28/3=9

Let's go through the math again. I know you did this as well, but let me go through it manually as sort of a 'thinking aloud' exercise.

We have two potental cases for Red (without going through all possible combinations since some are obviously not relevant).

First, for any grouping of 3 scores, you take the average
Then you take the greatest deviation (NOT standard deviation)
from the average and that's the deviation for that set.

Then you take the set with the least deviation score and that's the most similar.

Set 1:
(11,8,7) -> 8.66 avg.
11 - 8.66 = 2.33
8.66 - 8 = .66
8.66 - 7 = 1.66

So the deviation for this set is 2.33

Set 2:
(8,7,5) -> 6.66 avg
8 - 6.66 = 1.33
7 - 6.66 = .33
6.66 - 5 = 1.66
Deviation for this set is 1.66

1.66 is lower than 2.33, so we use this as our similar score, and Red gets 7 for this round.
What you're saying is, at the Trials, we came up with 6 not 7?

I don't remember what the numbers were, so I'm assuming you have the correct ones, and you can read my chicken scratch which you took a camera photo of.

As for the blue corner
Blue: 20,20,6,13,9

Again looking at the reasonable sets:
We have (20,13,9) twice, which means if this set is the most similar, there are two combinations that get us there. This could be important if that's most similar because then we have more than one equally similar combination, and we'd apply the trimmed mean process. I don't think that applies here though.

Doing the math we get
set 1 (20,13,9) -> 14 avg, deviation 6 (20-14)
set 2 (6,9,13) -> 9.33 avg, deviation 3.66 (13-9.33)

So set 2 is our most similar.

So we'd have 7 to 9.

I don't know how we arrived at 6 for the red corner in round 2. Can you email me that picture? I want to make sure I wrote things correctly. I don't have a copy of that bout, so I can't double check the numbers. But assuming the original numbers are correct, your math is accurate.

As for my recommendations? First off, these are off the top of my head, so I didn't test them for mathematically validity or anything. But I'd like to see the total similar scores rather than an average. Just sum them. We'd get higher scores (which means they don't necessarily relate to X blows for red any more). Using the example above we'd have Red 20 vs Blue 28 (adding, not taking averages). I think I'd also look at a better way for the tiebreaker. Maybe apply the similar mechanism to the total raw scores. Maybe just take out a tiebreaker and just poll the judges directly.

[donnellon]

What I mean to say is do the warnings come in to place when it's a countback?

[Slythex]

What I mean to say is do the warnings come in to place when it's a countback?

Forgive my ignorance, but what's a countback?

And if you mean do they come into effect during a level 1 tiebreaker, the answer is no.
Accepted end-of-round scores are totalled. If there's a tie, then it takes the raw sum of all judges for red, and raw sum of all judges for blue. Raw scores are not modified by a warning.

[donnellon]

Yeah, that's a c/b or countback over here in Europe!
That appears to be a terrible anomaly to me, a warning can lead to a count back(or tie-breaker) but then gets ignored in calculating the c/b itself?
That needs to be remedied. The new system seems to work well so far in my opinion and the absence of a running score helps big-time in that regard as we don't have boxers, running, falling, clinching to hold on to a narrow lead.

[boxmel]

By the way, where's Mel? I keep waiting for her to chime in and bust my chops over something or other

Hey! You're the freaking expert! My lips are sealed. As I've told you before, I don't like the "similar" scores as I think they cause a whole bunch of confusion, especially when the "cards" say red won but the similar scores have blue as the winner. However, it's on AIBA to fix.

[moorser]

Michael O'Reilly of Ireland recently lost in the final of the European youths even though after 3 rounds of boxing he was ahead on 3 judges score cards and level on 2 , somehow the Russian got the decision by a point but the new scoring system has me baffled . I mean how can someone win a fight when 3 judges have him losing and 2 a draw its bizarre . Can anyone care to explain ?

[boxmel]

I mean how can someone win a fight when 3 judges have him losing and 2 a draw its bizarre . Can anyone care to explain ?

Moorser - several times in this thread this type of scoring has been discussed. Please look back and you'll see this exact problem explained using real scores from a US tournament.

[Slythex]

Moorser- the short version is that the overall individual judge totals are not directly related to the outcome. The scores are calculated round-by-round based on which judges are most similar for that given round.

The problem is, and you have a perfect example, is that people aren't comfortable with a result where judges' final scores has one corner winning, but the ESS has the other corner winning. It's mathematically legitimate, but both frustrating and counterintuitive.

Read this thread from the beginning if you want to see the details of how it all works, and if you have specific questions, or want to run through a specific example, let me know.

[moorser]

Moorser- the short version is that the overall individual judge totals are not directly related to the outcome. The scores are calculated round-by-round based on which judges are most similar for that given round.

The problem is, and you have a perfect example, is that people aren't comfortable with a result where judges' final scores has one corner winning, but the ESS has the other corner winning. It's mathematically legitimate, but both frustrating and counterintuitive.

Read this thread from the beginning if you want to see the details of how it all works, and if you have specific questions, or want to run through a specific example, let me know.

Thanks, I read over it and I think I understand now although I do have one question .At the end of each round is it the same judges whos scores get knocked off for both boxers or is it different judges scores for both boxers get knocked off ??

I think there will be a lot of confusion at the next olympics when casuals will be watching and will not understand the scoring system at all if these type of results happen .

[Slythex]

It is not necessarily the same judges for each round.

So for example, the red corner may have J1, J2, J4 as the most similar.
The blue corner may have J2, J4, J5.

Does that answer your question?

[moorser]

It is not necessarily the same judges for each round.

So for example, the red corner may have J1, J2, J4 as the most similar.
The blue corner may have J2, J4, J5.

Does that answer your question?

yeah thanks