Final seeding after DEs

[SJCFU#2]

Originally Posted by ViewtifulMisho

Does anyone know why the FIE/USFA use absolute numbers for indicators, instead of percentage of touches?

...

Maybe this wasn't done in the past because of cumbersome pencil-and-paper work, but with most things computerized today, it should be no problem.

Part of it may simply be inertia. Bear in mind that it wasn't that many years ago that seeding was determined by total victories rather than percentage, and that's a simple enough calculation to not require a computer.

[Greg_D]

Originally Posted by ViewtifulMisho

Does anyone know why the FIE/USFA use absolute numbers for indicators, instead of percentage of touches?

...

Maybe this wasn't done in the past because of cumbersome pencil-and-paper work, but with most things computerized today, it should be no problem.

I believe that the FIE is currently considering making this very reasonable change. The USFA would almost certainly adopt this modification shortly after the FIE.

Your suspicion matches mine for why it's not already in the rules.

Greg

[Rockstar44]

Does anyone know why the FIE/USFA use absolute numbers for indicators, instead of percentage of touches?

Exactly! You are Genius!

For those of you who were beating me up, this is what I was talking about.

[bunbury]

Simple solution: never lose, get first place. Amirite or amirite?

[epeemike81]

Originally Posted by ViewtifulMisho

Does anyone know why the FIE/USFA use absolute numbers for indicators, instead of percentage of touches?

This is a stupidly technical and minute point:

A fencer in a 7-person pool with all victories has the potential for a minimum of one more indicator than a fencer in a 6-person pool with all victories, but it's not reflective of their relative success.

To be more correct, the formula for indicators should be:

touches scored - touches received
---------------------------------
        touches possible

where touches possible is 5*number of fencers in pool.

Maybe this wasn't done in the past because of cumbersome pencil-and-paper work, but with most things computerized today, it should be no problem.

EDIT: \/\/\/ Good point. Meant to say 5 * number of opponents

I've raised this question myself... one little nitpicky thing: the factor of 5 is unnecessary. Just make it TS-TR/B and call it "Average Margin of Victory".

-m

[Rockstar44]

Originally Posted by bunbury

Simple solution: never lose, get first place. Amirite or amirite?

Of course you are, but how does this add to the discussion? How do comments such as this justify or discredit anyone's position?

Some of us believe that the system should be modified. If you disagree, tell us why you disagree.

If all of us never lost it would be a dull sport, indeed.

[peet]

Originally Posted by Rockstar44

Does anyone know why the FIE/USFA use absolute numbers for indicators, instead of percentage of touches?

Exactly! You are Genius!

For those of you who were beating me up, this is what I was talking about.

Actually, I don't think so.

CMIIW, but you were suggesting counting the indicator as more important than win/loss record, and ViewtifulMisho is suggesting a way to equalize the indicator advantage of being in the larger pool. Two different things.

-p

[epeemike81]

Originally Posted by Rockstar44

Does anyone know why the FIE/USFA use absolute numbers for indicators, instead of percentage of touches?

Exactly! You are Genius!

For those of you who were beating me up, this is what I was talking about.

uh, no, not so much...

Originally Posted by Rockstar44

Here's my issue:

Fencer A in a pool of five scores 5/4, 5/4, 5/4, 5/4. This gives him an indicator of +5 with 4 wins and no losses.

Fencer B in the same pool scores 5/0, 5/0, 5/0, 4/5. This gives her an indicator of +14 with 3 wins and one loss.

Under the current system, Fencer A is seeded higher than Fencer B, but since they all fenced the same people, who really performed better in this pool? And, who should be seeded higher?

I think that indicators should be given more weight than wins/losses in the pools.

ViewtifulMisho's change would still have Fencer A seeded higher than fencer B. The change he wants to make would fix the following situation:
Fener X in a pool of 6 wins all his bouts by a score of 5-0.

Fencer Y in a pool of 7 wins 4 of his bouts 5-1 and 2 of them 5-0.

Under the current rules, Fencer Y is seeded first (1.00 V/B, +26 indicator) and Fencer X is seeded second (1.00 V/B, +25 indicator). This is quite unfair, given that Fencer X was perfect in every bout he had an opportunity to fence. He scored the Max, allowed the Min. he shouldn't be seeded behind somebody who was imperfect.

Under VM's change, Fencer X is seeded first (1.00 V/B, +5.0 I/B) and Fencer Y is seeded second (1.00 V/B, +4.33 I/B).

Average margin of Victory would replace simple indicator, but would still be subordinate to victory percentage.

-m

[Rockstar44]

Epeemike81:

You are right. Thanks for helping me understand the math.

This is an interesting system and is worthy of consideration for seeding. It looks like it is better than the current way of doing things.

[pinkelephant]

It was considered at the last congress, along with HS/max possible HS for the third indicator, instead of just HS. They are in favour but have deferred the decision.

From the FIE website:

http://www.fie.ch/download/letters/2...ONS%202007.pdf



E) PROPOSALS DEFERRED
Fencing Federation of Brazil (BRA)
Proposal 3
3.1. NEW CRITERIA FOR THE INDICES IN CASES OF EQUALITY OF INDICES V/M IN THE POOLS.
Article o.19 , 2 d) Organisation Rules:
d) In cases of equality of this first indicator, and to separate fencers with equal first indicators, a second indicator will be established { (HS-HR) / maximum possible HS-HR } in the pool of each fencer, in the bouts actually recorded at the end of the pool. The fencer with the highest indicator { (HS-HR) / maximum possible (HS-HR)} is seeded highest.
e) In cases of equality of the two indicators V/M and { (HS-HR) / maximum possible (HS-HR) }, a third indicator will be established { HS / maximum possible HS } in the pool of each fencer, for the bouts recorded at the end of the pool. The fencer with the highest indicator { HS / maximum possible HS } is seeded highest.
f) In cases of absolute equality of the three indicators above, the seeding order of the classification table will be decided by drawing lots.
Opinion of the Rules Commission: the Commission is in favour of the proposal:

d) in cases of equality of this first indicator, and to separate fencers who are equal, a second indicator will be established { (HS-HR) / maximum possible HS-HR} for each fencer in the pool, in the bouts recorded at the end of the pool. The fencer with the highest indicator { (HS-HR) / maximum possible HS-HR} is seeded highest.
e) in cases of equality of both indicators V/M and { (HS-HR) / maximum possible HS-HR}, a third indicator will be established {HS / maximum possible HS } for each fencer in the pool, in the bouts recorded at the end of the pool. The fencer with the highest indicator {HS / maximum
possible HS } is seeded highest.
f) In cases of absolute equality between two or more fencers, their position in the ranking will be
drawn by lots.

[epeemike81]

Originally Posted by pinkelephant

It was considered at the last congress, along with HS/max possible HS for the third indicator, instead of just HS. They are in favour but have deferred the decision.

that's a weird change for the third one... I'm of the opinion that if one person goes 5-0, 5.0 I/B (25 TS) and the other goes 6-0, 5.0 I/B (30 TS), the one who scored 30 SHOULD be seeded ahead of the one who scored 25.

-m

[Chris]

Originally Posted by ViewtifulMisho

Does anyone know why the FIE/USFA use absolute numbers for indicators, instead of percentage of touches?
<snip> [description of alternative solution]</snip>

When I first started fencing (back in the <cough> 70's </cough>), I was just doing it as a social activity: a bunch of my friends were trying it, so I tried it...Obviously, a page full of big fat 0's isn't much fun no matter what the format: at the time, the standard format for most events was to promote 50-60% out of each pool, and do more pools (there's more on that in another thread). So, I was always trying to win, but until I started winning a few, it really didn't matter much, and I really didn't pay much attention to details of running the events...

HOWEVER, I do remember seeing big (several pages of) charts (I thought they were printed into the 1974 USFA rulebook, but not sure...) which HAD a broad range of decimal number ratio's so that you could just look this up. (i.e they had already done the math for the range of numbers covering what happened in competitions, and just written down all the possible combinations.)

The primary thing was still winning bouts, bout, I think seeding into advance rounds usually used 2 rounds of results (whether or not it was feeding into a rare DE table).

I think that they had been using the ratio of touches as an indicator, but it may have been the third indicator; when fencers tied on the difference.
But again, I am not sure...

[pinkelephant]

Originally Posted by Chris

When I first started fencing (back in the <cough> 70's </cough>), I was just doing it as a social activity: a bunch of my friends were trying it, so I tried it...Obviously, a page full of big fat 0's isn't much fun no matter what the format: at the time, the standard format for most events was to promote 50-60% out of each pool, and do more pools (there's more on that in another thread). So, I was always trying to win, but until I started winning a few, it really didn't matter much, and I really didn't pay much attention to details of running the events...

HOWEVER, I do remember seeing big (several pages of) charts (I thought they were printed into the 1974 USFA rulebook, but not sure...) which HAD a broad range of decimal number ratio's so that you could just look this up. (i.e they had already done the math for the range of numbers covering what happened in competitions, and just written down all the possible combinations.)

The primary thing was still winning bouts, bout, I think seeding into advance rounds usually used 2 rounds of results (whether or not it was feeding into a rare DE table).

I think that they had been using the ratio of touches as an indicator, but it may have been the third indicator; when fencers tied on the difference.
But again, I am not sure...

No, this was just for the victories indicator - back then calculators were new, and not everybody could do sixths or sevenths in their head. (Some couldn't do fifths either, but don't get me started on the state of maths education - in teh UK at least).

[Chris]

Originally Posted by pinkelephant

No, this was just for the victories indicator - back then calculators were new, and not everybody could do sixths or sevenths in their head. (Some couldn't do fifths either, but don't get me started on the state of maths education - in the UK at least).

Well, I don't know - the tables seemed pretty big, although, I suppose they were covering pools of up to 10...(although, I am not sure how they analyzed when combining 2 rounds...), though, there's not much reason to bother doing division; if you can't analytically determine which is better/larger, it's easier to do Least Common Multiple: most people can do multiplication faster, and more accurately

[pinkelephant]

The tables were big to accommodate competitions run by poule unique.