[Help Thread] Improving Consistency Discussion

[Hiero]

Well, in Feliks' last competition he had a 10.36 and a 6.77. The 10.36 is 53% larger than the 6.77. Collinbxy's 38 sec solve was only 45.6% larger than his 26 sec solve. So on single solves Collinbxy is more consistent. Feliks' average was 31.1% larger than his world record. To be similar Collinbxy's average should be about 34.71 sec.

 

[freshcuber]

Well thanks Hiero. I expected a similar answer but didn't figure it would be such a large percentage.

 

[cyoubx]

Well, in Feliks' last competition he had a 10.36 and a 6.77. The 10.36 is 153% larger than the 6.77. Collinbxy's 38 sec solve was only 145.6% larger than his 26 sec solve. So on single solves Collinbxy is more consistent. Feliks' average was 131.1% larger than his world record. To be similar Collinbxy's average should be about 34.71 sec.

153% means that it was more than twice as long which it clearly was not...The 10.36 was only 34.7% longer than the 6.77.
Similarly, for Collinbxy, 38 is only 46.1% longer than 26 seconds.


Also, you're using a different number of significant figures. You're using two for Collinbxy but three for Felik's. If you were to use 10. and 7.0 you would get 43% difference.

Thus, really, there's only a 3% difference which is statistically insignificant.

 

[collinbxyz]

Well, in Feliks' last competition he had a 10.36 and a 6.77. The 10.36 is 153% larger than the 6.77. Collinbxy's 38 sec solve was only 145.6% larger than his 26 sec solve. So on single solves Collinbxy is more consistent. Feliks' average was 131.1% larger than his world record. To be similar Collinbxy's average should be about 34.71 sec.

Wow I never thought about that! I bet you could make a whole video on that...

 

[freshcuber]

153% means that it was more than twice as long which it clearly was not...The 10.36 was only 34.7% longer than the 6.77.

Depends on how you're measuring it. If you're measure the difference then yes you are correct. If you're comparing the total times of both solves then Hiero is correct because the first 100% is simply the time of the 6.77 solve.

 

[cyoubx]

Depends on how you're measuring it. If you're measure the difference then yes you are correct. If you're comparing the total times of both solves then Hiero is correct because the first 100% is simply the time of the 6.77 solve.

He said "larger" which implies a difference. He would have been correct if he said 153% of ___

 

[freshcuber]

If we're nitpicking then yes you are the correct one here but he is easily understood anyway.

 

[Hiero]

153% means that it was more than twice as long which it clearly was not...The 10.36 was only 34.7% longer than the 6.77.
Similarly, for Collinbxy, 38 is only 46.1% longer than 26 seconds.


Also, you're using a different number of significant figures. You're using two for Collinbxy but three for Felik's. If you were to use 10. and 7.0 you would get 43% difference.

Thus, really, there's only a 3% difference which is statistically insignificant.

My wording was off, but the slowest solve was 153% of the fastest one. The percent of change would be 53% from the fastest time to the slowest. From the slowest to the fastest it would be 34.7%. Collin's change from fastest to slowest would be 45.6%. If you go from percent of change from the slowest to the fastest, then Collin's would be 31.79%. I did use Collin's digits to the hundredths place as I did Feliks'. When I said it I just said 26 and 38, however I did use the 26.46 and 38.79.

You have to be careful if you are basing your percent of change off the slowest time or the fastest, you mixed them up. The percent from 6.77 to 10.38 is 53.3323486% not 34.7% so you are wrong too. Statistically I'm telling him not to worry about it.

Anyway, I changed the 153% to 53% in the post.

 

[cyoubx]

My wording was off, but the slowest solve was 153% of the fastest one. The percent of change would be 53% from the fastest time to the slowest. From the slowest to the fastest it would be 34.7%. Collin's change from fastest to slowest would be 45.6%. If you go from percent of change from the slowest to the fastest, then Collin's would be 31.79%. I did use Collin's digits to the hundredths place as I did Feliks'. When I said it I just said 26 and 38, however I did use the 26.46 and 38.79.

You have to be careful if you are basing your percent of change off the slowest time or the fastest, you mixed them up. The percent from 6.77 to 10.38 is 53.3323486% not 34.7% so you are wrong too. Statistically I'm telling him not to worry about it.

Anyway, I changed the 153% to 53% in the post.

Yep yep, that makes sense. I divided by the wrong number in my haste. Good catch.

 

[Hiero]

Do I get a prize or anything?

 

[collinbxyz]

Do I get a prize or anything?

Haha, maybe u can ask that to cyoubx

 

[cyoubx]

Do I get a prize or anything?

Haha, no.

I agreed with your latter comment. I still disagree with the other comment you made.
Using percentage is actually a pretty bad way to measure uncertainty for this simply because there are too many variables. With slower solves, a lock-up or two only effects the time by a small percentage. As times decrease, the lock-ups/otherthingsthatslowyoudown represent a larger percent of the solve.

 

[Hiero]

Haha, no.

I agreed with your latter comment. I still disagree with the other comment you made.
Using percentage is actually a pretty bad way to measure uncertainty for this simply because there are too many variables. With slower solves, a lock-up or two only effects the time by a small percentage. As times decrease, the lock-ups/otherthingsthatslowyoudown represent a larger percent of the solve.

You are absolutely correct. While fun to look at, the numbers are probably pretty meaningless when comparing people who are at different speeds.

Funny thing is that as a teacher we compare all these detailed standardized test results and try to read something out of them all the time when the majority of them are really meaningless. So I just learn to bite my tongue and pretend they do indeed mean something.

 

[collinbxyz]

You are absolutely correct. While fun to look at, the numbers are probably pretty meaningless when comparing people who are at different speeds.

Funny thing is that as a teacher we compare all these detailed standardized test results and try to read something out of them all the time when the majority of them are really meaningless. So I just learn to bite my tongue and pretend they do indeed mean something.

=D

 

[Thecuberlife]

How do I stop bipolar solves!

Ok so right now I really kinda suck at 3x3 and cant break the sub 40 barrier but i have a bad problem. And i have no idea what causes it. Bipolar solves. I hear or other people say they have bipolar solves because their solves change by about 2-3 secs but for me one solve can be 38 sec solve which is hard for me then after that ill get like 50 secs to a min. idk whats wrong please help my averages are really messed up because of this.

 

[JustinTimeCuber]

Although the important parts are in bold, I recommend that you read the whole thing, because I personally have had this problem and am getting much better at overcoming it.

A lot of people do get this problem. It is just more pronounced in beginners, because they are slower, and they have more parts of their solve that they can get lucky on or unlucky on. If anything, practicing in general will help improve your consistency. I myself have this problem, but it doesn't seem nearly as bad because the deviation is only like 3 seconds instead of 15. Although these solves don't happen nearly as often for me, but without the sup-16 solves in my averages I would easily average 12.xx. However, my last average of 50 was 13.57, and it was slower tham my global average just because after around 30 solves I started getting slower solves, and had to work harder to get faster solves. This does get better with practice. My number of sub-15/16 solves is dramatically decreasing, and I may break the sub 13 barrier soon. It isn't all bad though. Sometimes an inconsistent person will get a really good solve out of nowhere. In the spoiler is one of my favorite ever moments in cubing. At the time I averaged 17-18, although I generally do very well in comps, this is the second round, and since my first round I got a low 16 average that was what I was expecting, and I was hoping for a sub-16 average. I almost ended up with a sub-15. The end is worth watching though, because of how amazingly inconsistent it was (almost PB).

Spoiler

 

[PenguinsDontFly]

This doesnt only happen to slower cubers. It still happens to me, and I have no idea how to stop it. Slow turning helps sometimes (prevents mistakes, easier lookahead), but then you miss the opportunity to get a really good single on an easy scramble. Lol ao 5:

1. (10.93)
2. (6.00) (pb)
3. 8.34
4. 10.61
5. 9.40

average was 9.45 which is exactly what I average

 

[Isaac Lai]

I get this on skewb a lot. It'll be like cat 1 or cat 2 for 10 solves, then suddenly I get pure peanut + U/Z/H perm 4/5 times in an average. So my times will be like 3, 3, 4, 4, 4, 3... 3, 6, 5, 6 and then back to 3s and 4s. Sometimes I also screw up for no apparent reason so it's kinda annoying.

 

[dboeren]

Try to identify what's taking the extra time. Are you losing track of a piece and having to stop and hunt for it? Is it because you got wound up, tried to turn to fast, and locked up a couple times throwing off your rhythm?

Usually when one of my solves (I'm also a beginner) takes significantly longer than usual it's because I can't find a piece I'm looking for and have to go hunting for it.

 

[JustinTimeCuber]

Usually when one of my solves (I'm also a beginner) takes significantly longer than usual it's because I can't find a piece I'm looking for and have to go hunting for it.

that is what makes me inconsistent at 4x4 lel