How do you treat screw-ups in big averages?

[amostay2004]

There's probably no accurate way of determining my "true" average so these seem reasonable to me..

1. No I don't throw out fast, lucky times
2. I won't throw out bad, screwed up, unlucky (if unlucky solves exist), yet non-popping, times either
3. I'd limit DNFs to 5 out of 100 as stated earlier which is a pretty reasonably rate IMO since in an official competition you only do 5 solves in an average. If I pop 5 or less times in 100 solves it's pretty reasonable to call them 'abnormal' times that would distort the average. Of course, less than 5 is always better

I've changed my cube to a less-popping one so I'm pretty sure it's achievable

Or does anyone else have a better suggestion to determine one's "true" average, if there is actually such a thing?

 

[qqwref]

What about this: just trim off the best 25% and worst 25% of solves. Sure, you'll end up discarding every decent time you have, but no especially good/bad times will affect your mean (unless you are unbelievably inconsistent).

 

[blah]

Maybe I should add a function in my timer to trim off as many of the best times as the number of DNFs and calculate a DNF-less session average. Or maybe you could do that to qqTimer, Michael, I don't have the time to do anything right now, there are so many bugs at the moment it's unbelievable

It's a thought, no? :rolleyes:

 

[amostay2004]

@qqwref: Well it is possible and sounds reasonable..only thing is it takes some effort to manually cancel out the times, which comes to...

@blah: That'd be cool..but first please fix your SD calculation

 

[mazei]

How about a more of modal class of times function since that kind of seems more fitting in the situation qqwref is asking for?(Well in my very shallow pool of knowledge, it seems the most fitting. Any better, please suggest)

 

[miniGOINGS]

Just calculate the median of your times. That's what I would do, sounds pretty accurate to me.

 

[hawkmp4]

I like the idea of leaving everything and trimming off as many good times as DNFs. Seems to be the fairest way to go if you don't want bad solves in there. Then again...
I think that a POP raises your average much more than one lucky solve lowers it. For example, if I'm averaging 30s right now (which I am) and I pop and try to finish the solve, lets say I get a minute twenty. That's gonna raise my average much more than even an LL skip could.

 

[mazei]

When we POP, we DNF.

 

[Mike Hughey]

When we POP, we DNF.

What does that mean?

 

[amostay2004]

@ hawkmp4: Yeap. True

@ mazei: Well in official competitions you do get the choice to DNF the solve or fix it back and continue

 

[rjohnson_8ball]

I propose a big session be in multiples of 12 solves, with 1/12 of the best removed, 1/12 of the worst removed, and the rest averaged. For example, 120 solves with 10 best removed, 10 worst removed, and the remaining 100 averaged. An alternative method would be to just do ten separate sets of 12, removing 1 best and 1 worst from each, averaging the remaining 10 in each, then averaging the ten averages.

 

[Anthony]

I propose a big session be in multiples of 12 solves, with 1/12 of the best removed, 1/12 of the worst removed, and the rest averaged. For example, 120 solves with 10 best removed, 10 worst removed, and the remaining 100 averaged. An alternative method would be to just do ten separate sets of 12, removing 1 best and 1 worst from each, averaging the remaining 10 in each, then averaging the ten averages.

I really like that idea. In an average of 100 it's really easy to get a few bad pops that can mess you up. But I guess some would argue that doing it the current way better represents your "real" average.

 

[Meisen]

In my humble opinion the best would be to count all times, even the ones where you screw up.

If you pop once, solve the cube 102 times, remove the pop and the fastest time.
If you pop twice, solve the cube 104 times, remove the two pops and the two fastest times, and so on and so on.

The same principle applies to DNF's ofcourse.

This will give you a true average of 100.

 

[Mike Hughey]

I prefer the "real man's" approach. A DNF means your average is a DNF and you get to start over.

(But I'm guessing almost everyone could have guessed that about me.)

Almost all of my averages of 100 have included at least one time over 40 seconds.

 

[amostay2004]

I propose a big session be in multiples of 12 solves, with 1/12 of the best removed, 1/12 of the worst removed, and the rest averaged. For example, 120 solves with 10 best removed, 10 worst removed, and the remaining 100 averaged. An alternative method would be to just do ten separate sets of 12, removing 1 best and 1 worst from each, averaging the remaining 10 in each, then averaging the ten averages.

That agrees with what some people here are thinking (including me)..only thing is that it's quite a hassle to go through 100 times and cancel out the best and worst times..(or does CCT have a function to customize that?) If popular timers like CCT would add in a 'true average' feature where it calculates in a similar way (best and worst n number of solves removed)..a lot of people would start using that figure as their average I think

I prefer the "real man's" approach. A DNF means your average is a DNF and you get to start over.

(But I'm guessing almost everyone could have guessed that about me.)

Almost all of my averages of 100 have included at least one time over 40 seconds.

It can be quite demotivating though. Especially when you're doing really great and all of a sudden a huge pop happens and 8 pieces fly around the room

 

[shelley]

1. No I don't throw out fast, lucky times
...
3. I'd limit DNFs to 5 out of 100 as stated earlier which is a pretty reasonably rate IMO since in an official competition you only do 5 solves in an average. If I pop 5 or less times in 100 solves it's pretty reasonable to call them 'abnormal' times that would distort the average. Of course, less than 5 is always better

If you allow yourself to discard 5 DNFs out of 100 you also need to discard your 5 fastest times when calculating your average.

 

[amostay2004]

Well..fast, lucky times do not always mean the best times. They can be fast but I might have more than 5 non-lucky faster times in an average.

And yes I do agree with discarding the 5 fastest times but they don't necessarily have to be lucky..sorry if I was unclear about that (but then again the 'throw out both fastest and slowest times' hasn't come into discussion yet when I made that post )

 

[hawkmp4]

I prefer the "real man's" approach. A DNF means your average is a DNF and you get to start over.

(But I'm guessing almost everyone could have guessed that about me.)

Almost all of my averages of 100 have included at least one time over 40 seconds.

+1
POPs are not just bad luck, they're a result of a combination of poor cube set up and sloppy turning. I think that's important to note in any reflection of a cuber's skill.

 

[Tomarse]

You treat a screw up by throwing your puzzle in a downward motion towards the nearest hardest bit of concrete you can find and then sit down for 30 minutes sulking like a 4 year old Oh wait, thats just me?