How can AI help me improve at gaming?
How could an AI Coach work for League of Legends (LoL)? What data would it need? What models could it use? How could it help a player get better?
If you speak to any solo queue coach worth their salt, they’ll say roughly the same thing. If you want to climb, limit your Champion Pool. The question is, how do you build a Champion Pool? Is there such thing as an optimal Champion Pool? How many Champion’s should be in it? Today, we’ll talk about the factors you need to consider, as well as introduce the new iTero AI powered Champion Pool Builder.
It is such a consistent answer amongst coaches, but why? There’s a few key reasons. First, muscle memory. By limiting your pool you’ll start associating finger movements to abilities far faster than if you were playing a large range of Champions.
Similar to muscle memory, you’ll also build an instinctual understanding of the Champion and it’s capabilities. Patches change items and Champions often, and only by playing the same few characters repetitively will you be able best estimate what you’re capable of in the heat of battle.
Each Champion also has a different playstyle. Even between similar classes there is a notable variance to their playstyle. Diana is an AP Assassin, and so is Evelynn - and yet one is aiming to hit as many targets as possible, whilst the other is looking for as few. This impacts how they approach the game, and so by sticking to a small Champion pool you not only learn their moves, by how to optimize their playstyle.
There isn’t a known answer to this question. It can’t be proven statistically easily, either. However, from speaking to coaches we know the right answer lies somewhere between 1 and 6.
Although many disagree with as many as 6, remember that these should be played with decreasing frequency. Each of your top 3 picks should be played considerably more than your 5th and 6th.
The benefit of a smaller pool is simple. You become exceptionally good at the Champion, and the fewer you play the more exceptional you can become. Even by going from 1 to 2, you would get half as much experience on the first.
The benefit of a larger pool is a little more complicated. To fully understand it, it’s best to read my full explanation on winning the draft. In essence, these are what we have shown to be the most important factors to winning a draft:
The first two factors suggest limiting your pool to a strong meta Champion. However, it’s the 3rd and 4th point that encourage a larger pool.
For example, a Champion that regularly performs well in the Top Lane is Olaf. So, why not just become an Olaf One-Trick Player (OTP)?
Vayne, Kled, Trundle, Camille, Fiora. All these Champion’s make an Olaf’s life hell. Even outside of those, there’s plenty more that are not ideal. In fact, half of Olaf’s lane match-ups are considered worse than the average (obvious, if you think about it).
And so, what you need is more Champion’s to cover that weakness. If they pick Vayne, what do you do? The Olaf OTP grits their teeth and locks it in anyway, but at this point the Mastery & Win Rate is being heavily outweighed by the Counter match-up. What you need, is the equivalent of a Malphite in your pool, who is to a Vayne what Vayne is to an Olaf.
That leads us onto the question of building the optimal pool. Yes, we advise using our AI-powered pool builder, but we’re biased like that. So, let’s assume you want to do it manually.
The first thing to do is to choose a solid core pick. This is someone that has these qualities:
Then, it’s worth picking another. This time, you’ll want a different playstyle to the first. For example, if you picked an Assassin, look for a Mage, Bruiser or Tank.
Once you’ve done this, check what the bad match-ups are for your current pick. You may find your Top Laner gets countered by a lot of ranged lanes, or maybe your Assassin doesn’t enjoy tanks. This again, is hard to do manually, but you can get a rough approximation (if you’re anti-AI or have some other moral aversion to iTero’s tools).
You can then repeat this 1-2 more times, giving you a Champion Pool of 4-6 that are well varied and can handle most situations. It’s also worth doing two extra checks, which I find useful:
Luckily, that’s about as far as you need to go. There’s probably some ultra-optimization that can be done, however for 99% of player’s you will see a tremendous improvement in just creating a pool of 1-6 and sticking with it consistently.
If you haven’t already, I strongly advise checking out the iTero AI Champion Pool Builder, which automates this process for you, saving you doing the back-of-the-paper maths.
Here are some other articles that may be of interest.
How could an AI Coach work for League of Legends (LoL)? What data would it need? What models could it use? How could it help a player get better?
Does AI have the ability to shape the future of esports? Have its potential performance benefits been overhyped? Where does it differ from traditional sports?
What is the latest with iTero Gaming? Are we still fundraising? What is the “AI Coach” and what difference do we think it will make.