What do you consider a good standard deviation?

[MattMcConaha]

Obviously one can expect some sort of deviation in solve times from a number of different variables, but how large should be expected for a "consistent" solver? I searched and saw something about <10% being ok and <5% being really good, but this was a pretty old post and I'm wondering how consistent a lot of people are now that times are getting faster and faster, down to 10 second averages.

I, for one, currently have a terrible SD with my times bouncing off the walls and all over the place. From a collection of 150 solves I have a best of 16.49, mean of 23.06, and worst of 30.71 with an SD of 2.84 (12.3%)

Another question is how likely it is for one's standard deviation to naturally fall through practice without attempting to be more consistent. Basically, for example, if I were to ignore my consistency and just keep doing what I am doing until I average 15 seconds, is it expected for my SD to stay around 12% or drop down to say 9%?

 

[Dene]

Interesting question. I don't have any data to look at, but my SDs would vary considerably depending on how well I'm solving on any particular day. For example, I started a cubing session about an hour ago and gave up on it after maybe 20 minutes because I couldn't even consistently sub15, whereas I aim for sub13 and usually average sub14. I was getting solves between 12 and 18 today; all over the show >.< . But if I had a good, low13 or sub13 average, the SD would be much lower, with times between 11 and 14.

I don't know whether you could expect your general SD to decrease with time, but what does it matter as long as your times steadily drop anyway?

 

[MaeLSTRoM]

From what I've seen with my times, your SD isn't related to you speed really. For example, I am quite consistent on Square-1 (SD~=5s) but I average about 1:10 (7.1%), however for 3x3 I'm much less consistent (SD~3, Avg=13,23%). If you kept practicing without working on consistentcy, I think it would stay around the same proportion, but that it would decrease, however, I wouldn't say that is guaranteed.

 

[cubernya]

I would say a good SD is around 10%. The time range should drop, but only because of SD. So say you average 30s, a 3s SD is pretty good. If you average 10s, then a 1s SD is good (albeit a bit harder to achieve).

 

[MalusDB]

I think SD is more affected by what you know than how fast you are. If you don't know full oll/pll then you will be inconsistent when you get a 1 look compared to a 2 look oll/pll. Interesting question though, my SD can be awful one day and great the next, it kinda depends on the OLLs and PLLs mostly though. My F2L SD is very low.

 

[Dacuba]

I am not sure, but I think you can estimate the probality of the next solve bein sub x if you take your normal average as expected value and the SD. If I'm wrong, please tell me, I have no idea, what I am talking about

So let's say I have a 14s average with a SD of 1.2, what's the chance for a solve to be sub9?

(edit: yeah slightly offtopic, but not totally )

 

[mDiPalma]

P[Z<(9-14)/1.2]=.0000155 or .00155%

 

[JonnyWhoopes]

I prefer a low SD, but I also find that when my SD suddenly spikes for a few days that I'm usually about to break through a barrier. It can show your potential for good solves, and you just have to tap into what makes those solves good to regain your consistency; albeit at a new, faster speed.

 

[MTGjumper]

P[Z<(9-14)/1.2]=.0000155 or .00155%

Assuming that your solves are normally distributed. I've actually looked into this a bit, and solves from my sessions don't follow a normal distribution very well.

 

[MattMcConaha]

I think SD is more affected by what you know than how fast you are. If you don't know full oll/pll then you will be inconsistent when you get a 1 look compared to a 2 look oll/pll. Interesting question though, my SD can be awful one day and great the next, it kinda depends on the OLLs and PLLs mostly though. My F2L SD is very low.

For some reason I never really thought of that. I'm using a 2 look OLL right now, but I'm not doing it right so sometimes I end up doing 3 look... I'm guessing my SD would probably drop a lot if I learned the rest of OLL, but I've just never been able to get myself to do it.

And before anyone asks how I sometimes get 3 look OLL, I'll explain. If I get all four edges oriented, then I can do it 1 look. If I get two edges (in a line) oriented I can usually do 1 look, but sometimes I need 2. If I get 2 edges (in an L shape) oriented I usually do 2 look, but sometimes 1.
However, if I get none oriented, I only know one of the algs, so I just do F U R U' R' F' to get the two edges in a line and usually it works out but sometimes it doesn't.