The best 3x3 method that you probably don’t use

Joined
Jul 19, 2024
Messages
6,022
WCA ID
2022VALE15
For the entirety of speedcubing’s existence, every method out of the dozens of them always have atleast one flaw. CFOP has high TPS potential, but a higher movecount. Roux has a lower TPS potential, but a lower movecount. And ZZ… it’s kinda like CFOP but with better(?) ergonomics. These methods all follow a certain “rule”. If X(TPS) increases, Y(movecount) increases as well. And if X decreases, Y decreases too. But what if i told you that it doesn’t have to be this way? What if X increases and Y increases? What if i told you that the best method already exists among the dozens of methods, and has been presented to the public for 5 years? I’ll be explaining the story and overview of what is, in my opinion, the best 3x3 method created: APB.

Overview
APB is a heavily algorithmic method in it’s full form, and averages around 47-50 moves. The method is divided into 5 steps
1.) 2x2x3
This step is intuitive. The most common way cubers solve this step is by forming a 1x2x3 (as in roux), then solving the DFDB edges intuitively. Other methods of block building may be used, but 1x2x3 -> DFDB is the most popular. This step averages around 12-15 moves. This is a large step of the method and is one of the main areas of improvement for APB users.
2.) BR Pair
This step is usually done intuitively, but there is also an alg set for it if you want to learn it. For most solvers, they will be able to figure it out with their knowledge from the methods they already use (assuming you’re switching from a method). This step takes around 6 moves on average.
Optional Step: EOPair
This used to be the second step of the APB method, however, cubers realized that this alg set did not benefit the solver significantly, saving a shocking amount of 0.3 moves. It was decided that doing BR Pair and EO separately was just as good as EOPair. This alg set is not recommended.
3.) EO
This is done through one of eleven EO algs, averaging around 5 moves (rounded from 4.94 moves). These algs are easy to memorize, and most people can probably learn this alg set in a day.
4.) LXS (Last Extended Slot)
This is regarded as the best step of APB. Since EO is already solved, you can solve the rest of the cube with just RU gen moves. This step averages around 9 moves (including AUF). There is also an algorithm set for this step, which has around 116 algorithms. However, these algs are all easy to learn and are intuitive, making the 116 algorithms less intimidating. This step can still be done using your knowledge from other methods.
5.) ZBLL
Since we oriented the edges at step 3, that means the cross on top will be solved. ZBLL requires you to do 1 of 493 algorithms. On average, this step takes around 15 moves (including AUF) to solve.

In total, the full version of APB uses 620 algorithms. However, this algorithm count can be reduced to 11 algorithms. BR Pair and LXS can be done similarly to F2L in CFOP. Instead of ZBLL, you can use OLL and PLL, which you probably already know. All in all, the only algs you need to end up learning are the EO algs, which are easy to learn. However, do note that using OLL+PLL instead of ZBLL will significantly increase your movecount (in my experience, it seemed to have made my average movecount around 55-60 moves instead of 48-50).

Short Biography of the Inventor of the APB method (James Straughan)
James Straughan, also known as @Athefre, is the inventor/developer of APB. He’s made many meaningful contributions to the cubing community, from methods used by top solvers such as LEG-1 and EOLR, to recognition methods like Straughan Recognition and DFDB recognition for Roux. His method development journey started in 2006, where he developed two methods that you may or may not know about: MI1 and Nautilus (which would later be inspiration for other methods of his). He continued to pursue this passion, creating dozens of steps, recognition methods, and speedsolving methods. And in 2021, he proposed the APB method. Needless to say, James Straughan is arguably one of the most influential cubers in the community.

What makes APB the best method
Being an algorithm-based method, it ensures that you get the best (or close to the best) solution for a solve. Algorithms are usually faster than solving intuitively, unlike other methods. In CFOP, the speed of the solve is most likely determined by your F2L solution. The “best” solution to F2L is usually difficult to find. In APB, you can just spam algorithms for the entirety of the solve (assuming you use the full version of it). These algorithms are also optimized for ergonomics AND movecount, which makes the method even better. Most of the algs (aside from ZBLL) are also easy to learn, being only a few moves long.

Summary
APB is the best method

note: sorry for any typos you might find
 
For the entirety of speedcubing’s existence, every method out of the dozens of them always have atleast one flaw. CFOP has high TPS potential, but a higher movecount. Roux has a lower TPS potential, but a lower movecount. And ZZ… it’s kinda like CFOP but with better(?) ergonomics. These methods all follow a certain “rule”. If X(TPS) increases, Y(movecount) increases as well. And if X decreases, Y decreases too. But what if i told you that it doesn’t have to be this way? What if X increases and Y increases? What if i told you that the best method already exists among the dozens of methods, and has been presented to the public for 5 years? I’ll be explaining the story and overview of what is, in my opinion, the best 3x3 method created: APB.

Overview
APB is a heavily algorithmic method in it’s full form, and averages around 47-50 moves. The method is divided into 5 steps
1.) 2x2x3
This step is intuitive. The most common way cubers solve this step is by forming a 1x2x3 (as in roux), then solving the DFDB edges intuitively. Other methods of block building may be used, but 1x2x3 -> DFDB is the most popular. This step averages around 12-15 moves. This is a large step of the method and is one of the main areas of improvement for APB users.
2.) BR Pair
This step is usually done intuitively, but there is also an alg set for it if you want to learn it. For most solvers, they will be able to figure it out with their knowledge from the methods they already use (assuming you’re switching from a method). This step takes around 6 moves on average.
Optional Step: EOPair
This used to be the second step of the APB method, however, cubers realized that this alg set did not benefit the solver significantly, saving a shocking amount of 0.3 moves. It was decided that doing BR Pair and EO separately was just as good as EOPair. This alg set is not recommended.
3.) EO
This is done through one of eleven EO algs, averaging around 5 moves (rounded from 4.94 moves). These algs are easy to memorize, and most people can probably learn this alg set in a day.
4.) LXS (Last Extended Slot)
This is regarded as the best step of APB. Since EO is already solved, you can solve the rest of the cube with just RU gen moves. This step averages around 9 moves (including AUF). There is also an algorithm set for this step, which has around 116 algorithms. However, these algs are all easy to learn and are intuitive, making the 116 algorithms less intimidating. This step can still be done using your knowledge from other methods.
5.) ZBLL
Since we oriented the edges at step 3, that means the cross on top will be solved. ZBLL requires you to do 1 of 493 algorithms. On average, this step takes around 15 moves (including AUF) to solve.

In total, the full version of APB uses 620 algorithms. However, this algorithm count can be reduced to 11 algorithms. BR Pair and LXS can be done similarly to F2L in CFOP. Instead of ZBLL, you can use OLL and PLL, which you probably already know. All in all, the only algs you need to end up learning are the EO algs, which are easy to learn. However, do note that using OLL+PLL instead of ZBLL will significantly increase your movecount (in my experience, it seemed to have made my average movecount around 55-60 moves instead of 48-50).

Short Biography of the Inventor of the APB method (James Straughan)
James Straughan, also known as @Athefre, is the inventor/developer of APB. He’s made many meaningful contributions to the cubing community, from methods used by top solvers such as LEG-1 and EOLR, to recognition methods like Straughan Recognition and DFDB recognition for Roux. His method development journey started in 2006, where he developed two methods that you may or may not know about: MI1 and Nautilus (which would later be inspiration for other methods of his). He continued to pursue this passion, creating dozens of steps, recognition methods, and speedsolving methods. And in 2021, he proposed the APB method. Needless to say, James Straughan is arguably one of the most influential cubers in the community.

What makes APB the best method
Being an algorithm-based method, it ensures that you get the best (or close to the best) solution for a solve. Algorithms are usually faster than solving intuitively, unlike other methods. In CFOP, the speed of the solve is most likely determined by your F2L solution. The “best” solution to F2L is usually difficult to find. In APB, you can just spam algorithms for the entirety of the solve (assuming you use the full version of it). These algorithms are also optimized for ergonomics AND movecount, which makes the method even better. Most of the algs (aside from ZBLL) are also easy to learn, being only a few moves long.

Summary
APB is the best method

note: sorry for any typos you might find
We alr know u like APB
 
I feel like this introduction is highly biased, presenting ABP as the best method with no disadvantages... Yet it doesn’t even mention the awkward F and S moves in the middle of the solve. Also, how can you speed-recognize or predict ZBLL?
How many of the top solvers use ABP? If it has existed for 5 years, it should be widely used by now if it’s really that awesome.

Sorry if I sound skeptical... It’s probably because I am.

On the other hand, it's always good to explore and experiment with new things, we shouldn't stuck with a few methods unless we exhausted all other (infinite...? :) ) possibilities.

EDIT:
I just tried the method, studied it, played around, and to answer the question, "is this the best method for me"? The answer is no.
 
I feel like this introduction is highly biased, presenting ABP as the best method with no disadvantages... Yet it doesn’t even mention the awkward F and S moves in the middle of the solve. Also, how can you speed-recognize or predict ZBLL?
How many of the top solvers use ABP? If it has existed for 5 years, it should be widely used by now if it’s really that awesome.

Sorry if I sound skeptical... It’s probably because I am.

On the other hand, it's always good to explore and experiment with new things, we shouldn't stuck with a few methods unless we exhausted all other (infinite...? :) ) possibilities.

EDIT:
I just tried the method, studied it, played around, and to answer the question, "is this the best method for me"? The answer is no.
I did not think S and F moves were a disadvantage, since I personally did find them to be awkward, but I will admit that I was too ignorant about the possible flaws of APB and too subjective about my review/introduction. You can recognize/predict ZBLL while doing LXS. The main reason why I made this thread was to help increase the recognition of APB, since objectively and subjectively, it should atleast be able to compete with the big 3 (CFOP, Roux, ZZ). The APB discord server also only has 1k+ members, most of which probably don’t use APB as their main method. The fastest solver according to the 2H APB Leaderboard (from the APB Example Solve thread) has an average of 7.02 seconds by @Phalentin . I believe that APB is not widely recognized because there are few videos that cover the potential of APB.

I do realize that I should have been more critical instead of biased, I will note that for my future overviews
 
I did not think S and F moves were a disadvantage, since I personally did find them to be awkward, but I will admit that I was too ignorant about the possible flaws of APB and too subjective about my review/introduction. You can recognize/predict ZBLL while doing LXS. The main reason why I made this thread was to help increase the recognition of APB, since objectively and subjectively, it should atleast be able to compete with the big 3 (CFOP, Roux, ZZ). The APB discord server also only has 1k+ members, most of which probably don’t use APB as their main method. The fastest solver according to the 2H APB Leaderboard (from the APB Example Solve thread) has an average of 7.02 seconds by @Phalentin . I believe that APB is not widely recognized because there are few videos that cover the potential of APB.

I do realize that I should have been more critical instead of biased, I will note that for my future overviews

Thank you for the clarifications.

I do believe that the biggest disadvantage here is the lack of proper guides and tutorials, which can be easily fixed in the future; and spreading awareness of the method can help that. I guess that is the main reason for this thread. :)
 
Do we know where the 12-15 move estimate for 223 came from? I’ve seen it repeated a lot. Is it for a fixed orientation or something? I’m certainly not a high level solver, and I would consider a 15 move 223 kinda bad. I’d expect optimal solutions to be something closer to 9-12 moves on average (not that many people could consistently find near optimal solutions in inspection for now).
 
Last edited:
Do we know where the 12-15 move estimate for 223 came from? I’ve seen it repeated a lot. Is it for a fixed orientation or something? I’m certainly not a high level solver, and I would consider a 15 move 223 kinda bad. I’d expect optimal solutions to be something closer to 9-12 moves on average (not that many people could consistently find near optimal solutions in inspection for now).
I took it from this website:
Now that i’m on the APB discord server though (i was not in there at the time of writing the article), the general consensus for a “good” 223 is now 9-12 moves as you stated. I’m going to work on a revision for this thread when my schedule isn’t full.

Caleb do you main APB for 2H?
i don’t main APB for 2H, but i main it for OH
 
Im not Caleb Valenzuela but

1001523529.webp

Yhe thing is I tried opening this one already but its not working at all and I cant find any other than it I would appreciate if you could screenshot the sheet
 
View attachment 49709
Yhe thing is I tried opening this one already but its not working at all and I cant find any other than it I would appreciate if you could screenshot the sheet
IMG_7938.webp

IMG_7939.webp

IMG_7940.webp

I think the highlighted edges are the ones that are misoriented. Also there’s text above these algs:

[td]
UPDATED STATUS OF EO-PAIR:
The current standard (circa February 2024) for EO-pair solves the pair into the back slot and then does pair-solved EO.
The technique of doing pair-creation followed by eo-pair is an older style. Depending on the particular algs you use and your lookahead capabilities, this may or may not be slightly faster at the cost of learning an additional 128 algs. (One estimate says pair-creation→EO-pair saves ≈0.3 moves over the current standard on average)
It is also possible to solve the pair in the front right, although this is not recommended for most solves as it would require learning an additional 116 LXS algs and there are more blind spots during EO and LXS.​
[/td]​
 
View attachment 49710
View attachment 49711
View attachment 49712
I think the highlighted edges are the ones that are misoriented. Also there’s text above these algs:



[td]
UPDATED STATUS OF EO-PAIR:
The current standard (circa February 2024) for EO-pair solves the pair into the back slot and then does pair-solved EO.
The technique of doing pair-creation followed by eo-pair is an older style. Depending on the particular algs you use and your lookahead capabilities, this may or may not be slightly faster at the cost of learning an additional 128 algs. (One estimate says pair-creation→EO-pair saves ≈0.3 moves over the current standard on average)
It is also possible to solve the pair in the front right, although this is not recommended for most solves as it would require learning an additional 116 LXS algs and there are more blind spots during EO and LXS.

[/td]​
Thank you very much my brtoher
 
i have a friend who is sub 15 with APB, and it looks intaresting. i am currently sub 12 using full cfop and am wondering if there would be benefits to switching. i’m probably not going to because i tend to not like memorizing many algs (oll took a lot of effort) but tbh i don’t really care it’s just time

and i’ve already started memorizing 2GLL in terms of zbll
 
Looks cool. I think claiming it's the "best" is a bit peculiar though, especially when one of the main arguing points for it being better than CFOP is intuitive solving being slower than algorithmic solving on average. Modern high level CFOP involves planning 2-3 pairs (sometimes even 4 in rare cases) during inspection. This is essentially converting most of (and sometimes all of) the intuitive part of CFOP into being algorithmic, and thus executable at the highest TPS the cuber is capable of achieving, completely subconscious. It's also worth noting that modern high level ZB (especially with psuedo F2L techniques) usually gets low 40s movecounts.

Looks like a method with a lot of potential, but I don't think I can agree that it's the "best". I do see the appeal though, as simply learning algs and practicing recall and recog for them is a lot more feasible for the average cuber than developing such an insane level of CFOP inspection depth and efficiency, especially with comp nerves being a factor.
 
4.) LXS (Last Extended Slot)
This is regarded as the best step of APB. Since EO is already solved, you can solve the rest of the cube with just RU gen moves. This step averages around 9 moves (including AUF). There is also an algorithm set for this step, which has around 116 algorithms. However, these algs are all easy to learn and are intuitive, making the 116 algorithms less intimidating. This step can still be done using your knowledge from other methods.
This is not accurate, CP has not been solved so the cube has not been reduced to RU gen.
 
Literally just petrus except instead of 222->223 you do 123->223 and also the eo+sb is restricted to being done a certain way instead of freeform. and then you solve an awkward triplet of pieces in fr for seemingly no reason other than to look cool
 
Last edited:
Back
Top