How to Improve at 3x3: The Three Pillars of Speed

How do you get faster at 3x3? The question is as old as cubing itself. The answer has changed again and again as the sport developed. It also depends on many factors: your method, your age, the time you have and the goal you are working towards.

In this article we describe two general concepts that apply to every cuber at any level, as long as you can already solve the cube with some method. The first splits your speed into three pillars you can measure. The second breaks each pillar into sub-skills you can train on their own.

What Getting Faster Means

On its face, the question is easy to answer. Getting faster means solving the cube in less time. Your time is your movecount divided by your TPS (turns per second). So "just turn faster" is a valid answer, but there is a lot more nuance to turning faster.

Imagine your fingers suddenly turned the cube twice as fast. Your times would improve, but you would not be twice as fast. Your fingers only do half of the work. The other half happens in your head, and your brain has not caught up with the new finger speed yet. You would pause longer during your solves and most likely use more moves as well.

This is why it pays to split your cubing skill into several sub-skills that you can analyse and target one by one. Whether you have 20 minutes or 6 hours of practice a day, getting the most out of that time will make a large difference to your development.

The Three Pillars

Every solve comes down to three pillars, and each of them has a metric you can measure:

An example solve

Let us look at a 16.53 second CFOP solve by one of our users. Below is the replay, next to every step as acubemy reconstructed it from the smart cube. You can also open the full solve with its step-by-step analysis.

The solve took 63 turns at 3.81 TPS. Its fluency was 46 %, so the cuber spent about 8.9 seconds pausing and about 7.6 seconds turning. While turning, they reached roughly 8.3 TPS.

With these three numbers you can calculate how much time each pillar could save. For this solve:

• If the cuber needed 57 moves, the average CFOP movecount, with the same turning speed and pauses, they would save about 0.7 seconds.
• If they turned at 14 TPS, the speed of world record solves, whenever they turn, with the same movecount and pauses, they would save about 3.1 seconds.
• If their movecount and turning speed stayed the same but lookahead removed all of their pauses, they would save about 8.9 seconds.

For this cuber, lookahead is the clear weakness. Removing the pauses is worth almost three times as much as turning at world record speed, and the six moves above average cost less than a second.

It is not always obvious which of the three pillars is easiest to improve at the moment. Doing this calculation for your own solves gives you a solid first estimate of your strengths and weaknesses, and shows where you have room to improve at all.

The nuance goes deeper, though. Each pillar consists of several sub-skills, and each of them can be improved separately.

Lookahead: Removing Your Pauses

The first way to remove pauses is classic lookahead. While you solve one step, you track the pieces you need for the next one, so you can start it without stopping. The second way is faster case recognition, although recognition time never drops to zero. The third way is prediction: either planning further ahead in inspection, or predicting on the fly, as in OLL prediction or AUF prediction.

Prediction is becoming more and more important in modern competitive speedcubing. That does not make the fundamentals of classic lookahead less important. For intermediate cubers, lookahead is often the most important weakness to work on.

Metronome training

If you are learning lookahead for the first time, metronome training helps you get into the groove. Set a metronome to 60 BPM, start a solve and do exactly one turn per beat. This will feel very slow. Use the time between turns to plan ahead, at every single turn. Once you can finish a full solve without missing a beat, raise the tempo by 20 BPM and repeat. You may give yourself one beat of grace on recognition steps.

Building prediction

At the same time, start working on prediction. When you practise, do not limit yourself to 15 seconds of inspection. Instead, always plan one step further than you usually do. If you usually plan two cross edges, plan three. If you plan the entire cross, plan the cross and the first pair. Keep at it until you can do this comfortably within 15 seconds, then add the next step. This is how Xuanyi Geng and Yiheng Wang built their prediction skills.

Efficiency: Lowering Your Movecount

You lower your movecount mainly by studying the cube. The classic example is learning more and shorter algorithms, cutting a step from 4-look to 2-look to 1-look. There is much more to it, though. Planning your first step, choosing the right pairs or blocks, and colour neutrality can all cut the number of moves you need by a large margin.

In our experience, it is a good idea to lower your movecount before you grind thousands of solves. Too many cubers reach sub-10 times with a movecount of 70. When they then start learning new techniques, the payoff is delayed. Their times get worse before they get better while they adjust to the new techniques and algorithms, and that can be very demotivating.

This does not mean learning 1000 algorithms as soon as you can solve the cube. It means always having something new to study on the side while you improve. It also means making it a priority to finish a workable version of your method. For CFOP, that is crosses in 8 moves or fewer, full OLL and PLL, and both basic and advanced F2L.

acubemy can help with both parts. You can learn new algorithms in training, and the solve analysis lets you review your cross and F2L solutions after every solve.

Turning Speed

Turning speed is the most deceptively simple of the three pillars. At first glance, all you need to do is move your fingers faster. There is a lot of nuance here as well. Drilling your fingers to move faster is central, but other things set the limits of what you can reach: a well set-up, quality speedcube, good fingertricks, avoiding regrips and an ergonomic choice of algorithms. Each of these plays an important part and can be tackled separately.

For example, it does not hurt to drill your algorithms with the smart training in acubemy. If it turns out that your fingertricks are not optimal, though, drilling them might strengthen bad habits. To fix your fingertricks, watch how pro cubers execute your algorithms of choice on YouTube, or talk to them at a WCA competition. For OLL, we have a series of YouTube Shorts that shows the optimal fingertricks for the most popular algorithms.

Putting It Into Practice

Knowing all of this only gets you halfway to training efficiently. You still need time, discipline and a training structure. What this approach gives you is an educated guess at your strengths and weaknesses. From there, you can split your practice time into small, targeted modules, make each skill measurable and track your improvement. That answers the question of training structure.

This is also where acubemy comes in. With a smart cube, you get an overview of all three pillars quickly: movecount, fluency and TPS are measured for every solve in the app. The insights tab shows your strengths and weaknesses across the whole solve at a glance.

More targeted training tools are coming to acubemy, such as cross training and metronome training. We will also publish more articles on this blog that go deeper into the individual sub-skills.