Global Average?

[Rocky0701]

I have seen many people describe their "global averages" many different ways lately, but i have not seen a real definition of a "global average". I know what it means to me, and the basic concept, but how would you guys define it, and/or calculate it? I have seen people do ao1000's to figure it out, i have seen someone do and ao5 of 5 ao100's, i have seen another person do the same with an average of 12. How would you define it?

 

[jeff081692]

Maybe we could formally define it in the wiki or something. Right now it's really just a guess of what you think you could always achieve doing an average of any number. I've done like 25 megaminx solves now and I could say that my global average is that average but really if I were to do another 25 solves I would guess my global average to be much faster since I am in that beginner phase when each solve is faster than the previous.

People who are stuck at a time barrier can easily tell their global average since they see the same session average range each day and that's what I went with in the past.

 

[kcl]

For events like 3x3 I usually take the general average of ao100's I'm getting. For 2x2 I basically call it from stuff like ao1000.

 

[GuRoux]

i think for global average is just what you think you average based on the solves you've done.

 

[brian724080]

I think it's the expected value (time) of an event

 

[Kit Clement]

This gets down to the heart of frequentist statistics - how do we estimate the true value/measure of something? Of course, the easy way to get an idea is to look at data, but data never represents the true value, it can always fluctuate. So what do we do to measure our Rubik's Cube times? We take "averages;" specifically, truncated averages, as we always drop the best and worst X%, depending upon the average.

I see many people define it as an expected value, but this is NEVER done in practice. First, how do we deal with DNFs in expected values (when we take an average of all the solves)? Even in our sample averages, we usually take out the best and worst X solves. When we do this trimming, this is taking the true definition of average (not the cuber definition) and moving it more towards the median. The median is actually a trimmed average, although it's a bit trivial since it's doing the maximum amount of trimming. Cubing is also a very luck-driven activity, especially on the bad luck side of things. Pops, lockups, wrong algorithms, there's so many things that can go drastically wrong, and this is often why we employ the trimmed average, as this allows us to make a mistake without shooting ourselves in the foot. It also better reflects what our usual time is, as the average would be typically a bit higher than our usual time with our terrible solves pushing the average up. So taking an average or expected value of your times isn't a good approach, as it doesn't get our usual times, and it doesn't account for DNFs.

Also, looking at a sample average isn't a good way either, as this can vary greatly depending upon the time of day, amount of warm up solves, etc. We want to get our true skill level, and no matter how many solves you do, you can never really get the exact true value, although you will get close. Thus, I think it's best to look at global "averages" truly as global medians. Ask yourself this: what time can I anticipate to beat 50% of the time? I'd think that best describes your solving skill the most out of any measure.

 

[Mollerz]

Generally I consider Global Average to be: If you sit down right now and do an average of 100 solves, what would be the expected value. My PB average of 100 is low 13 right now, but I wouldn't expect to sit down and get a 13.20 average of 100, I'd expect it to be just above 14, so I say my global average is about 14.

 

[szalejot]

Generally I consider Global Average to be: If you sit down right now and do an average of 100 solves, what would be the expected value. My PB average of 100 is low 13 right now, but I wouldn't expect to sit down and get a 13.20 average of 100, I'd expect it to be just above 14, so I say my global average is about 14.

I agree with that.
Basically: When you sit down now and do average what is most expected value you will get?

 

[GuRoux]

Generally I consider Global Average to be: If you sit down right now and do an average of 100 solves, what would be the expected value. My PB average of 100 is low 13 right now, but I wouldn't expect to sit down and get a 13.20 average of 100, I'd expect it to be just above 14, so I say my global average is about 14.

the big gap between your pb average of 100 and global average seems strange.

 

[Rocky0701]

the big gap between your pb average of 100 and global average seems strange.

He was probably just really warmed up when he got that PB and/or he hasn't been practicing much lately.

 

[Ninja Storm]

the big gap between your pb average of 100 and global average seems strange.

I feel like it's somewhat normal. My PB avg100 is 9.97, but I average closer to ~10.2-3 at the moment.

 

[DeeDubb]

I've been doing an Ao100 pretty much every day for the past couple of weeks. It's weird, because when I feel like I'm doing awful, the difference is usually only 1~1.5 seconds from my PB Ao100. I think any given Ao100 is going to be a fairly accurate reflection of your global average, with a little margin for error.

 

[Rocky0701]

I've been doing an Ao100 pretty much every day for the past couple of weeks. It's weird, because when I feel like I'm doing awful, the difference is usually only 1~1.5 seconds from my PB Ao100. I think any given Ao100 is going to be a fairly accurate reflection of your global average, with a little margin for error.

For people like you and I, our PB ao100s are pretty much our global averages, because i have been improving constantly and have noticed that you have been too. For someone that is like sub 15 though, i would think that it would be around a second higher than their PB ao100.

 

[GuRoux]

I feel like it's somewhat normal. My PB avg100 is 9.97, but I average closer to ~10.2-3 at the moment.

That's really only a gap of .2-.3, not much. 1 second or more seem strange at sub 15 times.

 

[GuRoux]

For people like you and I, our PB ao100s are pretty much our global averages, because i have been improving constantly and have noticed that you have been too. For someone that is like sub 15 though, i would think that it would be around a second higher than their PB ao100.

I think my pb average of 100 is somewhere around 10.9 but my global average is probably 11.1. But that me just be because i don't do much average of 100.

 

[kcl]

I think my pb average of 100 is somewhere around 10.9 but my global average is probably 11.1. But that me just be because i don't do much average of 100.

No, I'm like that as well. My PB ao100 is 9.74 but I average around 9.9

 

[ChickenWrap]

I only do 7x7, but my global average is whatever the WORST possible average time I could expect for 30-50 solves. Right now, that is 5:30 for me, but on many days, I will get 5:15-5:25 avg of 12 with occasional days of 5:30. So I take the worst possible times I could expect and that is my global average.