Quite awesome how you predicted the rising of Yiheng! I'd be curious to see this analysis done again maybe in a couple of months, when the WC will be really close.
I absolutely love this idea and will try to calculate stuff myself! One thing I would disagree with though is putting the entire data through a linear regression, the times cubers got when they first started competing affects the slope way too much. Maybe take away the first year or so of data or only look at the past X years of data or something would make it better? Also I didn't know this database query existed! I'm going to have so much fun with this data, thanks for sharing!
Thanks! I tried playing around with the linear regression, but cubers who been to few competitions or haven't competed in a while got really skewed results so I left it as it is. Definitely agree though, might just get rid of the variable completely.
heres my list 1 / 2 = Tymon/Max (I think its too close) 3 - Matty 4 / 5 - Luke / Leo 6 - Ruihang 7 - Patrick 8 - Feliks Dont know after that, yiheng probably wont be there One person i really hope does will is Seung-Hyak-Nam (think thats what his name is spelled) who lost his title in worlds 2017 cuz of that timer reset. He was robbed of his title, and then couldnt come to 2019 cuz of his mandatory military service. I really hope he does well, he deserves it.
I'm really hyped to see what worlds has to offer us! I'm waiting for a lot of records to be broken, and who knows, even tymon winning his first world championship! (and hopefully the 3x3 single is broken until there lol)
1:33 The WCA will round the average you get. If you get a 6.129 average, it will be rounded to 6.13, since WCA only uses 2 decimals. This isn’t used for getting single solves though.
I think u gotta re-do this ranking after Singapore Championship 2023... We are seeing crazy records and improvements by the Chinese cubers🤯Yiheng became the third person ever to record an official sub-5 average, Ruihang getting a 5.19 average, 2 sub-4 solves in one minute.... wild.
I think yiheng is going to dominate as he did almost break the world record but it didn’t happen because a miss scramble but then a few weeks later he breaks the world record
Timestamps: 10th 1:32 9th 2:37 8th 3:08 7th 3:46 6th 4:22 5th 5:19 4th 6:16 3rd 6:50 2nd 8:03 1st 9:15 Timestamps if u just don't wanna watch the whole vid. But please watch this vid, this is very entertaining and also informative for all cubing fellas out there.
yo this is awesome! id love see something similar for all events (ik thats a lot of work tho lol so dont feel pressured) as i think this stuff is super interesting!
Really nice job with this. When trying to determine the winner of a given event, looking at minimizing standard deviation makes sense (improved consistency, etc). However, another interesting question I've tried to predict is which cuber is most likely to break the existing avg of 5 world record. For this, if we assume that cubers don't improve over time, a higher standard deviation actually helps. The results of the Monte Carlo show that Tymon and Max are the only cubers capable of breaking the record (not a surprise), and that it should be broken within the next 1.5 years. Tymon's odds are slightly higher than Max's. Tymon and Max also have a huge advantage over others in this analysis because of how many comps they attend per year. They get more "at bats", so more chances to have a lucky set of scrambles.
For predicting a world record, I wonder if an extreme value distribution would work, because we're not trying to predict the average, but an extreme event (like an actuary trying to predict a high cost event like a hurricane).
@@statmango you actually are trying to predict an average - an average of 5 where the high and low values are dropped. That’s what makes it such a complicated problem to tackle statistically (I’m not even sure if it’s theoretically possible to solve with pure math). That’s why I chose to approach it with Monte Carlo simulation with a million solve sample size.
good stuff dude - interested in learning about your model further - specifically how you 1) chose variables, 2) how you solved for the beta coefficients. To get "predicted world's final average", did you just take each previous world's final competitor result and backtest based on each of their 7 stats going into that world's final?
@@chungboislim2061 I was initially confused by this reply, but now I understand it quite well with your comment on this video about Feliks as an example
@@chungboislim2061 GOAT topic aside. I only made two claims in my comment, 1) Bofan faster than Yiheng at the same age is a fact, and 2) Bofan might very likely got top10 placing at WC2023. Feel free to make some arguments about these
pr average tell nothing sorry. what you should do is plot a linear regression model and predict the time they'll get in world champs. Then plot a linear regression for Standard Deviations and predict the standard deviation they have in world champs. Solely use these two factors to make predictions
this model definitely needs improvement as I won't believe that until this year the final of the world championship will have more than two sub-6 averages, I mean how can you expect everyone doing at their best? most of them will choke and struggle to even get sub-7, look at WC 2019, a lot of sub-7 level solvers actually got 8 sec avgs so i believe that the same thing will happen again
I aggre with all but the top 2. I think max park should be 1 because he improved his pr average from 5.08 to 4.86 I on month of tymon taking it. Also you should do squan, 2x2 and skweb even though everyone knows who is 1 for all of those. Max shau for squan and zayn khanani 2x2 and skewb
I can tell sm work was put into this! I'm glad you too into account over factors like standard deviation and improvement. This is great content
You're a great youcuber
Why did I read this like mario 💀
Quite awesome how you predicted the rising of Yiheng! I'd be curious to see this analysis done again maybe in a couple of months, when the WC will be really close.
yiheng came second by 0.01 to max park
10:54 "I don't really expect the 3x3 podium to be all sub 5.6" Insane how Tymon placed 3rd with 5.42 and Ruihang 4th with 5.59 😂.
"Yiheng wang is only 5th"
Now:
9:16 1st place
8:02 2nd place
6:50 3rd place
6:15 4th place
5:03 5th place
4:19 6th place
3:45 7th place
3:08 8th place
2:37 9th place
1:33 10th place
1 year later crazy how the stats predicted yihengs rise
I absolutely love this idea and will try to calculate stuff myself!
One thing I would disagree with though is putting the entire data through a linear regression, the times cubers got when they first started competing affects the slope way too much. Maybe take away the first year or so of data or only look at the past X years of data or something would make it better?
Also I didn't know this database query existed! I'm going to have so much fun with this data, thanks for sharing!
Thanks! I tried playing around with the linear regression, but cubers who been to few competitions or haven't competed in a while got really skewed results so I left it as it is. Definitely agree though, might just get rid of the variable completely.
heres my list
1 / 2 = Tymon/Max (I think its too close)
3 - Matty
4 / 5 - Luke / Leo
6 - Ruihang
7 - Patrick
8 - Feliks
Dont know after that, yiheng probably wont be there
One person i really hope does will is Seung-Hyak-Nam (think thats what his name is spelled) who lost his title in worlds 2017 cuz of that timer reset. He was robbed of his title, and then couldnt come to 2019 cuz of his mandatory military service. I really hope he does well, he deserves it.
Uhh, he actually did come to 2019
And the next worlds is in Korea. I really hope he does well.
Luke Garrett left the chat
@@chillguy07 he literally put luke at number 4/5
He probably won't be there... 4.69 average (I'm this was before but still he had a 2.99 single online)
love statistics videos like these! hope to see more like these in the future
I'm really hyped to see what worlds has to offer us! I'm waiting for a lot of records to be broken, and who knows, even tymon winning his first world championship! (and hopefully the 3x3 single is broken until there lol)
I really liked this! SQ1/Mega/5x5 I would want to see, particularly 5x5 to see Feliks vs. Max.
For 5x5 probably a landslide to Max lol. In my greatest of all time video though, Feliks does a lot better in 5x5
I think in 5x5, Max has a slight bit of competition from Tymon and Seung. But he'll probably win.
1:33 The WCA will round the average you get. If you get a 6.129 average, it will be rounded to 6.13, since WCA only uses 2 decimals. This isn’t used for getting single solves though.
Technically you can only get a .XX00, .XX33 or .XX67 average as it's mo3 but yeah
this is really cool! I like how it not only takes into account a cuber's ability and accomplishments, but also their potential
This is insightful! Especially when there are so many clickbait videos from others recently, this high quality video is very interesting to watch
thank you dr fang
I think u gotta re-do this ranking after Singapore Championship 2023... We are seeing crazy records and improvements by the Chinese cubers🤯Yiheng became the third person ever to record an official sub-5 average, Ruihang getting a 5.19 average, 2 sub-4 solves in one minute.... wild.
He did
Whatching in July 2023 and seeing yiheng Wang ranked 5...
I think yiheng is going to dominate as he did almost break the world record but it didn’t happen because a miss scramble but then a few weeks later he breaks the world record
YOU PREDICTED YIHENG WANG'S FUTURE
Timestamps:
10th 1:32
9th 2:37
8th 3:08
7th 3:46
6th 4:22
5th 5:19
4th 6:16
3rd 6:50
2nd 8:03
1st 9:15
Timestamps if u just don't wanna watch the whole vid. But please watch this vid, this is very entertaining and also informative for all cubing fellas out there.
yo this is awesome! id love see something similar for all events (ik thats a lot of work tho lol so dont feel pressured) as i think this stuff is super interesting!
Then YiHeng got two WR averages.
You should also remake this
Really nice job with this. When trying to determine the winner of a given event, looking at minimizing standard deviation makes sense (improved consistency, etc).
However, another interesting question I've tried to predict is which cuber is most likely to break the existing avg of 5 world record. For this, if we assume that cubers don't improve over time, a higher standard deviation actually helps.
The results of the Monte Carlo show that Tymon and Max are the only cubers capable of breaking the record (not a surprise), and that it should be broken within the next 1.5 years. Tymon's odds are slightly higher than Max's.
Tymon and Max also have a huge advantage over others in this analysis because of how many comps they attend per year. They get more "at bats", so more chances to have a lucky set of scrambles.
This is very interesting too! Would love to look into this more some time
For predicting a world record, I wonder if an extreme value distribution would work, because we're not trying to predict the average, but an extreme event (like an actuary trying to predict a high cost event like a hurricane).
@@statmango you actually are trying to predict an average - an average of 5 where the high and low values are dropped. That’s what makes it such a complicated problem to tackle statistically (I’m not even sure if it’s theoretically possible to solve with pure math). That’s why I chose to approach it with Monte Carlo simulation with a million solve sample size.
9:17 😂
Thank you Mr. Fang!
Wow this video is amazing and its clear that you put a lot of time and effort into it great job
2023 just got started…
good stuff dude - interested in learning about your model further - specifically how you 1) chose variables, 2) how you solved for the beta coefficients. To get "predicted world's final average", did you just take each previous world's final competitor result and backtest based on each of their 7 stats going into that world's final?
2:28 Feliks got 2nd so he did podium but didn’t manage to pull of the expected win like he did in the first three rounds
You're a legend sam, I love predictive statistics like this. Keep it up!
The Martín Telésforo (El feo) joke cracked me up... haha
Ruihang Xu 5.48 didn't really come out of nowhere.
He was smashing times like his 2.68 and had already had great times at Monkey League.
u could do square 1 next, it will be pretty interesting for you to analyse your own stats
It’s time to make a new video for this after Singapore Chanpionship 2023. The Chinese Cubers should make a significant climb.
Best cubing video of 2023 so far
Fantastic. Would love to see the analysis for blindfolded
try lukes rs33m super
How about Bofan Zhang (2021ZHAN01), I think he should be in top10 now, he got even better potential than Yiheng (faster than Yiheng at the same age)
@@chungboislim2061 I was initially confused by this reply, but now I understand it quite well with your comment on this video about Feliks as an example
@@chungboislim2061 GOAT topic aside. I only made two claims in my comment, 1) Bofan faster than Yiheng at the same age is a fact, and 2) Bofan might very likely got top10 placing at WC2023. Feel free to make some arguments about these
Quality content!
This is the best video I’ve seen in a long time
The best cuber of all time:
A non-cuber's cousin
Could you do this for bigger cubes, such as 5x5 or 7x7? That would be amazing
0:04 beard tymon
How to access the database?
This is a little outdated, Yiheng has the WR average and it isn't even Worlds yet.
Can I do this myself? (without the massive formula)
what r ur formulas for avg/sd slope/growth factors
i loved this! ty
You are so underrated! Keep up the good work❤
Hey Sam! Could you make another one of these before worlds? Cause these are really outdated lol.
yiheng ruihang bofang podium
That Martin joke though 💀
Tymon is my favorite.
This was a very interesting video on cubing. You deserve way more subs than this
and here we are 2 months later Yiheng gets WR average😂
This is a very interesting ranking or forecasting system.
im glad luke is on this list!
Yehang now: 🗿
What did the graph oh SD mean?
How does Matty end up 7th when he's ranked better than 7th in every category?
That confused me too, but it kinda makes sense when looking at the final chart. Everyone above him either improve really quickly, or have a better PR
Great video!!!
Yiheng Wang all day.
I'm wondering what's the most and the least important factor among the seven according to the coefficients of your formula?
PR is most important, although I think it should be weighed the most. Standard deviation was probably the least important.
update it wr avg has been broken
This is actually the actual ranking of speed cubers in the world 2023
what is yiheng on the list after his 4.91 average?
End of this video if you haven't seen it yet :)
th-cam.com/video/PForYJrBATI/w-d-xo.html
you should do this for square one
Good video!
Now that Yiheng has the wr average, how high would he be now
Now max park is the wr holder 3.13
@@speedcuber__AarinSingle
pr average tell nothing sorry.
what you should do is plot a linear regression model and predict the time they'll get in world champs. Then plot a linear regression for Standard Deviations and predict the standard deviation they have in world champs. Solely use these two factors to make predictions
This is so detailed.....! Osm
sadly the world championships is not anywere close to where i live, but im happy to see these people pottentioly get there prdidcted avrages
Maybe before worlds do this for 20 best cubers in every event that will be competing
Great video
5:15 and it ended up being SHOCKING that he didn't win worlds
this model definitely needs improvement as I won't believe that until this year the final of the world championship will have more than two sub-6 averages, I mean how can you expect everyone doing at their best? most of them will choke and struggle to even get sub-7, look at WC 2019, a lot of sub-7 level solvers actually got 8 sec avgs so i believe that the same thing will happen again
Yep agreed it needs improvement, as I mentioned towards the end. However, I think the rankings themselves are pretty accurate!
Sub-6 averages in world finals. I do think it will happen. Crazy
The fact that yiheng wang has a 4.91 Average now
can you pls do 3BLD. It is my main event.
Very impressive video
1. Tymon
2. Max
3. Feliks
4. Matty
5. Ruihang
6. Leo
1. Max
2. Yiheng
3. Tymon
4. Ruihang
5. Luke
6. I forgot
Yiheng got 4.91 average !!!!! 🎉
recent average be like for yiheng
4:42 those people clearly dont watch ruiman yan
I aggre with all but the top 2. I think max park should be 1 because he improved his pr average from 5.08 to 4.86 I on month of tymon taking it. Also you should do squan, 2x2 and skweb even though everyone knows who is 1 for all of those. Max shau for squan and zayn khanani 2x2 and skewb
he included max's avg... watch the whole video
Martin telesforo 🐐
9:17 💀
Luke Marvin, 2022MARV01, Pyraminx. I've been practicing it a lot. Thanks for the good content!
I watched this video 2 hours before i found out yiheng wang took ao5 wr. If you were to do the analysis over again, where would this then put him?
End of this video has the rankings after Singapore Championships th-cam.com/video/PForYJrBATI/w-d-xo.html
As of now, his predicted average is sub-5 :O
@@CubeMasterYT sheeee
If Yiheng manages to not fuck up because of the pressure, he's gonna destroy Max and Tymon. I'm calling it.
he choked on solve 5
very nice vid
But Max Park is more than just 3x3? He dominates big cubes
OH
Mega
5x5
3BLD
Statistics are cool, but clutch time is different.😊
Member gang
hi im member
also just a quick comment
consistency isn't really accuraate when you consider the fact that cubers can improve a lot in 2 years
I’m not 100% sure of the formula used but wasn’t this addressed in the video and accounted for? Not completely sure though
@@Yurio do you know which program he used to calculate
martin telesforo for the win lets goooo
Sq1, it would be so close