Averaging 25 seconds on 3x3 with CFOP, know 2 look OLL and PLL, as well as a few extra PLLs, and most F2L solutions. How to improve from here?

[VortexCuber]

I have been averaging the same about the past few months, I don't know how to improve.

 

[Chicken-Cuber]

I have been averaging the same about the past few months, I don't know how to improve

Learn full pll it helps a bit and work on more efficient cross and f2l and some look ahead

 

[efattah]

The main trick to get better (with any method...) is to stop doing random solves and instead drill each phase of the solve until you hit a target time for each phase. For CFOP you would be drilling just the cross until you can average X.XX for the cross, then drill all 42 F2L cases until you can solve a random case in say 2 seconds average, then drill OLL until you hit a certain time, drill PLL til you hit a certain time. Use the cstimer scramble generator or other utility to generate scrambles for each of these drills. It is less fun than just random solves but you will get better 100x faster.

If you commit to devoting a FULL MONTH of just drills, then when you go back to random solving you probably be sub-18.

 

[Puzzle Maniac]

The main trick to get better (with any method...)

I read this as, "The main trick is to get better (with any method...)"

 

[kubesolver]

I think the issue is most likely that you keep solving and don't actively learn.
Even if you learned only one new alg per week then in few months you should have already learned full pll.

And there is a limit to how good you can get by just solving. You might have reached your limit of improvement if you keep doing the same things you do now. And potentially doing more of that can be counter productive if you just keep cementing your bad habits.

And solving PLL and OLL in two looks is bad also because you don't improve your OLL and PLL recognition because you just look for simplified patterns. Try to learn some easy cases of OLL(e.g. all 8 movers) and PLL ( both J, T and As) and work on recognizing those from other cases. This will potentially enable a new sense that will help in looking at the cube at all stages.

 

[Caleb Valenzuela]

i suggest learning Full PLL, then after Full OLL. Use an algorithm trainer. For cross I would say to try and get it under 1 second. And for F2L, learn the 42 cases, and learn some more for the back slot. Make sure you learn all this deliberately, which means to really focus your practice around it.

 

[Exmestify]

I have been averaging the same about the past few months, I don't know how to improve.

you can get upto like sub-15 without full oll and full pll, but its not recommended
(below ao5 is only 4 look ll)
Generated By csTimer on 2025-03-10
avg of 5: 13.431

Time List:
10718. 13.048 B2 R2 D L2 R2 F2 D' F2 U2 R2 D' F2 L' B L2 R2 D B' R U2 F'
10719. 12.460 F2 R2 D' F2 L2 U R2 D' F2 D' L' D' R2 D2 U' F R' U R2 D2
10720. (11.017) U L R2 F2 R2 D F2 R2 U' R2 F2 U2 L B' F2 D L2 D2 F' L2
10721. 14.785 B2 D' B' U2 R B2 L' F2 L F2 L2 B2 F2 L' B D L2 U' R' D L'
10722. (18.103) D' R2 F D2 L2 U2 F D2 B D2 R2 L' B2 D2 R' D' L2 B L

 

[Jorian Meeuse]

I just want to say one thing about only doing solves, which is that you can definitely improve a lot with it. I have barely ever done targeted practice other than alg learning, and most algs I've learnt, I've also forgotten. Still, I average close to sub-7 (according to when I last solved a lot, at least). It's just that there are two types of solving. The first is to solve mindlessly, which isn't going to help you improve. The second is solving without caring about being fast, but only about not being stupid. For example, don't solve the first F2L pair you see in the first way you can think of, but maybe solve a different F2L pair which will give you a better case after. Or maybe try that risky double x-cross that you aren't sure will work, instead of the trivial cross solution which puts all corners in the D layer. The more you do this, the faster and more confidently you can do smarter solutions, which means you improved. Of course, the degree to which you can do this highly depends on your level, but I think the concept applies to anyone.

Edit: btw, that's not to say you shouldn't do targeted practice. I just wanted to address something which I believe is quite a common misconception

 

[LukeKing67]

The main trick to get better (with any method...) is to stop doing random solves and instead drill each phase of the solve until you hit a target time for each phase. For CFOP you would be drilling just the cross until you can average X.XX for the cross, then drill all 42 F2L cases until you can solve a random case in say 2 seconds average, then drill OLL until you hit a certain time, drill PLL til you hit a certain time. Use the cstimer scramble generator or other utility to generate scrambles for each of these drills. It is less fun than just random solves but you will get better 100x faster.

If you commit to devoting a FULL MONTH of just drills, then when you go back to random solving you probably be sub-18.

That sounds like it would help but ALL 42 FTL CASES?!