The Statistical Case for Combined Finals

[Kirjava]

Anyone who thinks that someone setting three averages of 14 is better than someone setting averages of 9, 25 and 9 is extremely naive.

 

[reThinking the Cube]

Anyone who thinks that someone setting three averages of 14 is better than someone setting averages of 9, 25 and 9 is extremely naive.

Anyone who thinks that someone setting averages of 9, 25 and 25 is better than someone setting averages of 10, 10 and 10 is extremely naive.

 

[Kirjava]

I disagree. The former competitor would be ranked higher than the latter.

 

[reThinking the Cube]

Anyone who thinks that someone setting three averages of 14 is better than someone setting averages of 9, 25 and 9 is extremely naive.

Anyone who thinks that someone setting averages of 9, 25 and 25 is better than someone setting averages of 10, 10 and 10 is extremely naive.

I disagree. The former competitor would be ranked higher than the latter.

I disagree. The latter competitor *should* be ranked higher than the former.

 

[ExoCorsair]

Why did the numbers change from 14 to 10?

 

[PhillipEspinoza]

I think the point is that you do it when it counts. There are many people who do good at home but can't really do that good in competition. For one competition, right before the final round, I was practicing with Chris Dzoan and got two 8-second times in a row. I was called up shortly after that, competed, and received a DNF average which resulted in Michael Gottlieb winning the competition with a 13.67 average. This is when I averaged around high-10, low-11 seconds.

The point is you gotta do it when it counts.

Oh and I think in a competition we're measuring speed and not so much consistency. It's not the person with the lowest SD who should win but the person with the fastest average.

 

[Kirjava]

I disagree. The latter competitor *should* be ranked higher than the former.

Have you literally lost your ****ing mind?

Please stop trolling.

 

[reThinking the Cube]

I disagree. The latter competitor *should* be ranked higher than the former.

Have you literally lost your ****ing mind?
Please stop trolling.

That's very funny, I was thinking the same thing about you. Your use of some feeble-minded namecalling and *****ing reveals some things already quite lost.

What don't you get about a 10 sec. final round average beating a 25 sec. final round average? Which competitor will get the better rank for that event? Sorry to make you look so *******, but that 9sec. result in an earlier round with the current format, would obviously not affect the final rankings for that event.

9,25,25 > 10,10,10

Now, PLEASE stop *condoring*. I think just posting a picture of your most recent attempts at self-mutilation, would suffice.

 

[cmhardw]

Please keep this discussion civil, or we will resort to infractions and/or temp-bans.

Chris

 

[Kirjava]

final rankings

WCA ranking != final ranking

Now, PLEASE stop *condoring*

Could you define this for me? It doesn't appear to be a word.

I think just posting a picture of your most recent attempts at self-mutilation, would suffice.

Aside from the technique you've been using in this thread where you copy the post someone made with slight changes to it, this is one of the stranger things you do...

 

[qqwref]

[...]The former competitor would be ranked higher than the latter.

I disagree. [...]

You can't disagree. That's the way the overall rankings are set up.

I was called up shortly after that, competed, and received a DNF average which resulted in Michael Gottlieb winning the competition with a 13.67 average.

lol, sorry

Things like this would be much much less likely if we used the 'best average in any round wins' style. I think it is more accurate to say "A is better than B because A got a faster average in this competition" than "A is better than B because B got two DNFs in five consecutive solves". (Is it a good thing to let someone who is clearly not the fastest solver there have a non-negligible chance of winning? I am not sure what I think.)

That's very funny, I was thinking the same thing about you. Your use of some feeble-minded namecalling and *****ing reveals some things already quite lost.

You are a surprisingly devoted hypocrite.

What don't you get about a 10 sec. final round average beating a 25 sec. final round average? Which competitor will get the better rank for that event? Sorry to make you look so *******, but that 9sec. result in an earlier round with the current format, would obviously not affect the final rankings for that event.

9,25,25 > 10,10,10

This whole topic is about which result should be considered better. As you've somehow managed to figure out, a 25 sec. final round average would indeed lose under current rules. However, the question of whether someone who gets 9,25,25 is better than someone who gets 10,10,10 (and whether they should be ranked better in a given competition and overall) is nontrivial. This is especially true since it is a LOT easier in cubing for someone who is normally around 10 seconds to get a 25 second avg5 than for someone who is normally around 25 seconds to get a 10 second avg5. Now that you know this, the discussion should make more sense.

 

[reThinking the Cube]

I think the point is that you do it when it counts.

Good point. Currently, 5-solve trimmed average for the final round is arbitrarily defined as what *counts*. In 1982 only 1 solve counted. WCA can make whatever it wants count. Kirjava, and qqwref advocate making the best average in ANY round *count*. I really don't have a problem with that either, but if all the solves where treated as one round, and a 15-solve double trimmed average was used to decide the winner, then ALL the solves would *count*.

[...]The former competitor would be ranked higher than the latter.

I disagree. [...]

You can't disagree. That's the way the overall rankings are set up.

LOL. We are discussing a format change that would potentially affect the rankings - are we not? Reread it in the original post, and notice *would* and *should* are 2 different words. There is also the difference between overall WCA rankings, and the rankings for any single tournament event. Now that you know this, the discussion should make more sense.

I was called up shortly after that, competed, and received a DNF average which resulted in Michael Gottlieb winning the competition with a 13.67 average.

lol, sorry

Things like this would be much much less likely if we used the 'best average in any round wins' style. I think it is more accurate to say "A is better than B because A got a faster average in this competition" than "A is better than B because B got two DNFs in five consecutive solves". (Is it a good thing to let someone who is clearly not the fastest solver there have a non-negligible chance of winning? I am not sure what I think.)

@qqwref/kirjava: The key thing to realize is that the serious competitors of a major comp. are now too close in ability for any one 5-solve trimmed to decide it fairly. I think this has been presented rather convincingly in some earlier posts. Whether or not tournament results should be, or need to be fair, is another debate altogether, but there is about as much aberation in results for final round now as there is for a 1-solve only for Round#1 elimination, which isn't done for that same reason. Using the best average from any round would be an improvement, but it is (IMO) not as good as using all 15 solves with double trimming.

BTW - Phillip would have won (Bayview Hills Open 2010 3x3x3)http://www.worldcubeassociation.org/results/c.php?allResults=All+Results&competitionId=BayviewHills2010 if any of these proposed formats were used. By trimming 2 best and 3 worst times - Sikan Li would have made it to the final round, and Michael Gottlieb would have placed 5th overall in my book, instead of getting the win.

This whole topic is about which result should be considered better. As you've somehow managed to figure out, a 25 sec. final round average would indeed lose under current rules. However, the question of whether someone who gets 9,25,25 is better than someone who gets 10,10,10 (and whether they should be ranked better in a given competition and overall) is nontrivial. This is especially true since it is a LOT easier in cubing for someone who is normally around 10 seconds to get a 25 second avg5 than for someone who is normally around 25 seconds to get a 10 second avg5.

No ****Sherlock! What have I actually been suggesting as a nontrivial improvement for determining which result(s) should be considered better?

 

[hr.mohr]

Guys how can a discussion about statistics and competition format result in such tone?

There is a few things that got lost in this discussion and that's the other properties in the current format that makes organizing competitions more manageable.

Having more attempts takes more time and are imho less exciting for spectators to watch.

Also in the arguments about performing when it counts there is a difference. There is much more pressure when you know that the next few solves determine if you go thru to the next round or win the competition. Taking the best average or the mean of averages will not induce the same pressure because the next solve does not win or loose a competition. If your first rounds was much better than your main competitor then the last round should only be on par with the other guys result. Less pressure and less excitement for spectators.

Fairness is also a tricky thing to discuss. The current format only favor one cuber over the other in the first rounds because of the different scrambles. The final are fair as all competitors get the same scramble. While a shorter average enables cubers to perform much better than larger averages, it's still fair to determine a winner because all competitors get the same terms. I think that some people see competition differently. I see it as a chance to feel pressure and do my best when it matters and some see it as a way of determine the best overall. I find that pretty boring. I like that there is chance of big upsets. I like when you can discuss results from previous rounds and have thought experiments about who could have won instead.

 

[reThinking the Cube]

Guys how can a discussion about statistics and competition format result in such tone?

All right, you have earned my utmost respect.

There is a few things that got lost in this discussion and that's the other properties in the current format that makes organizing competitions more manageable.

Having more attempts takes more time and are imho less exciting for spectators to watch.

This is true, but I don't recall anyone suggesting increasing the number of attempts.

Also in the arguments about performing when it counts there is a difference. There is much more pressure when you know that the next few solves determine if you go thru to the next round or win the competition.

Yes, this is a very valid consideration. Trimmed running averages can also keep the pressure on, but using the best average of any round might not, since the result might have already been clinched.

Fairness is also a tricky thing to discuss. The current format only favor one cuber over the other in the first rounds because of the different scrambles. The final are fair as all competitors get the same scramble. While a shorter average enables cubers to perform much better than larger averages, it's still fair to determine a winner because all competitors get the same terms.

Well said. This (scramble equity) is one of the key objections to using the best average for any round format, but this shouldn't be much of a problem with a trimmed running average.

I think that some people see competition differently. I see it as a chance to feel pressure and do my best when it matters and some see it as a way of determine the best overall. I find that pretty boring. I like that there is chance of big upsets. I like when you can discuss results from previous rounds and have thought experiments about who could have won instead.

I can relate to this pressure/challenge/chance aspect. Using a trimmed running average should place that aspect on every solve.

 

[Kirjava]

but if all the solves where treated as one round, and a 15-solve double trimmed average was used to decide the winner, then ALL the solves would *count*.

This will never happen.

[...]The former competitor would be ranked higher than the latter.

I disagree. [...]

You can't disagree. That's the way the overall rankings are set up.

LOL. We are discussing a format change that would potentially affect the rankings - are we not? Reread it in the original post, and notice *would* and *should* are 2 different words. There is also the difference between overall WCA rankings, and the rankings for any single tournament event. Now that you know this, the discussion should make more sense.

A 9 average will currently be ranked higher than a 10 average. This is a fact you cannot disagree with.

You're still doing that childish thing where you mirror other people's posts.

Also, I'd still love to know what you think the word 'condoring' means.

 

[Pedro]

Fairness is also a tricky thing to discuss. The current format only favor one cuber over the other in the first rounds because of the different scrambles. The final are fair as all competitors get the same scramble. While a shorter average enables cubers to perform much better than larger averages, it's still fair to determine a winner because all competitors get the same terms.

Well said. This (scramble equity) is one of the key objections to using the best average for any round format, but this shouldn't be much of a problem with a trimmed running average.

Not really. Easy scrambles may affect your average by a big margin, and someone who didn't get the same scrambles but has about the same speed would be much slower than you.

This is the main problem I see with couting every solve (or 10 out of 15 or whatever) or the best average of the competition: different scrambles. I know that for the top people, the scramble doesn't matter that much, but as Macky said, bad LL or F2L cases can slow you down 0.5 or so, and in a world where sometimes competitors are 0.05 apart, that's gotta be considered.

EDIT:

Just to illustrate:

Suppose 3 rounds.
Competitor A and B compete in different groups in the first 2 rounds.

1st round:
A does 10,11,10,(11),(9) ("easy" scrambles)
B does 11,12,12,(11),(12) (not that easy)

2nd round:
A: (9),10,10,(11),10 (easy again)
B: (10),11,11,(12),10 (not so easy)

Final: (all in the same group)
A: (10),10,11,12,(12)
B: (10),11,10,11,(12)

In the current format, B wins (10.67 x 11.00).

If we took the best average, A wins (10.00 avg on 2nd round)

If we average all of them (13 out of 15), A gets 10.38 avg and B gets 11.08 avg.

Suppose they were in the same group for the 2nd round and A kept the same results, but B got
B: 10,10,(9),10,(11)

Now B's average is 10.85.

My conclusion (not claiming it's the absolut thruth):
First round scrambles were so easy that A got much better than what he actually is. B seems to be faster to me (two 10.00 x 10.00 and 10.67 x 11.00 when they had the same scrambles).

 

[cmhardw]

Guys there is one other thing to bring up here. If we use 15 solves to rank every competitor, how do we eliminate people into smaller and smaller rounds? I can tell you that at some of the smaller competitions, with fewer judges, and also at the much larger competitions, with MANY competitors, it will not be feasible to give every competitor 15 solves. Someone, and in fact many people, must be eliminated at some point to make the running of the competition feasible. This would mean that not every competitor gets 15 solves, and yet you must still fairly rank the ones that do not get all solves.

I know that, from a theory standpoint, everyone getting all 15 solves is best. But, try giving every competitor at a small competition 15 solves when you're already short on judges for the current format. It will not be easily done without either eliminating other events or running over time.

Chris

 

[CubesOfTheWorld]

I think that the winner should be decided by whoever had the best average in any of the rounds. There should still be elimination to move onto the next round.

 

[Daniel Wu]

If the winner is decided by the best average, then it would seem that it would really take away from the entire concept of the finals.

 

[riffz]

If the winner is decided by the best average, then it would seem that it would really take away from the entire concept of the finals.

I disagree. By the time finals roll around, each competitor knows the mark that they need to aim for to jump into 1st place. It's comparable to watching olympic speed skating, where you know the current first place time, and I still think that's very exciting to watch.

Each competitor also knows what they averaged in previous rounds, so if one of their previous averages was already fast, they can really put it on the line and go all out, instead of holding back a bit because they want to secure a spot on the podium.

And let's not forget that even if a really fast average has been set in previous rounds that no one feels they have a chance of beating in the finals, people are still striving to beat their PBs. I think most people would agree that they care more about their WCA ranking in an event than how they placed at a specific competition.