TURBO MODE

THE DATA BEHIND

SPELL SPAMMING

Clairvoyance Blog

SUP PLAYAZ. It’s your favorite data science playboy JEOPARDY with famed machine learning mogul RIOT NITSUABO here to share some insights about how to be less of a scrub. We gather data on just about everything that happens in live League of Legends matches—and we mean everything, including the exact number of times every player in the world uses their abilities in-game.

When we started looking into the data we collected, we found some pretty weird results that will be quite interesting for all kinds of players. Let's jump into it.



Observation 1: Riven Only

So, before we get into the weird stuff, let's start out with one example that's relatively simple to wrap your mind around. Riven is commonly perceived as a champion who requires #mechanics. With no mana costs and relatively short cooldowns, getting the most out of the combo queen requires careful timing and deliberate sequencing of your spell rotations. The data below doesn’t capture the sequencing stuff, but it does let us look at the correspondence between player skill and frequency of Riven's spell casts. Basically, does MMR increase with number of buttons mashed?

Riven's Q

The number of Broken Wing spell casts increases as MMR increases.

Do y'all like reading charts? Because we're about to read the everloving crap out of this chart. Here's how, in seventeen easy steps:

1. The y-axis shows the number of spell casts per minute, with higher values meaning the player activated an ability more often.

2. The x-axis is player rank. The far left side consists of wood tier scrublords like Riot's very own Rumtumtummers. The far right side is probably Faker.

11. The dark line shows the median number of spell casts per minute for Riven players at that MMR. That means that half of all Riven players cast the ability less than this amount, and half cast it more than this amount.

17. The pale ribbon captures what we would consider a representative range of activations for Riven players at each rank—the lower bound of the ribbon shows the 5th percentile of activations per minute, meaning only five percent of Riven players of this skill level in this region cast the ability fewer times than this. The upper bound of the ribbon shows the 95th percentile, meaning that only five percent of Riven players of this skill level in this region cast the ability more times than this. Put differently, 90 percent of Riven players in a region have spell casts per minute within the values bound by the Riven ribbon.

So, the average Diamond-ranked Riven player uses Broken Wings (Q) more than the average Bronze player. BUT, even some of the lowest-ranked Riven players use Q more often than the typical Diamond-ranked player, as per the ribbon. So, obviously, just spamming Q doesn't make you good. We can tell from the pale ribbon that there are actually plenty of Diamond-ranked Riven players who use their Q less than the typical Bronze-ranked player. Not all great Rivens spam Q.

If we look at the rest of Riven's abilities, we see that the trend also applies to her E, and to a lesser extent for her W or R.


Riven's W, E, and R



Observation 2: Support Utility

So, based on the Riven example, you're probably thinking "okay, higher-ranked players tend to use their abilities more." But—pushes up glasses and sniffs—it actually depends on which champ is being played. Many supports have high-utility spells that tend to be used less frequently at higher MMR’s.

Thresh's Q

The number of Death Sentence casts decreases as MMR increases.

Let’s take Thresh’s hook (Q), for example. It has tons of power whenever you haven't yet used it. Saving it forces your opponents to play as if you could use it at any point. They have to stay positioned safely behind their own minions, even if it causes them to miss some CS in the process. The threat of a hook also discourages enemies from poking you aggressively under your own tower, because who wants to write their own Death Sentence?


Janna's Q

Higher MMR players tend to hold Howling Gale.

Nami's Q

Higher MMR players tend to hold Aqua Prison.

Janna's Howling Gale (Q) and Nami's Aqua Prison (Q) work similarly—they can be used as an aggressive poke or engage tool, but they tend be more reliable as a counter-engage. If an enemy is charging full-speed into a fight, it's easier to accurately predict their movements compared to when they're just pacing back and forth to farm, so you've got a better shot at landing the skillshot.



Observation 3: Carry Utility

Of course, supports aren’t the only group of champions this applies to. Some carry champs have abilities with valuable skirmish utility. For the most part, these spells are very reliant on situational awareness; they often determine whether or not you win a trade against an opponent. These high-risk-high-reward spells aren’t usually used for damage—nobody in their right mind builds AP just so Fiora’s Riposte can poke a little harder, or so that Wukong’s Decoy does a little more damage. These spells also tend to have higher cooldowns (Riposte is 24 seconds at level one), so you have to make every button press count.

Draven's E

Higher MMR players tend to hold Stand Aside.

Draven’s Stand Aside (E) is an example of such a utility spell. At lower MMR’s, players might get flustered in fights and default to mashing out all of their abilities to maximize damage as quickly as possible. However, Draven’s E has the highest mana cost of all his non-ultimate abilities, so it isn’t a tool you should use in every trade. Instead, it should be saved as an engage or disengage against enemy champions—for example, if timed properly, it can interrupt Tristana’s Rocket Jump.

We’ve all seen the famous clip of Daigo’s Ken parrying Justin Wong’s Chun-Li for a game-winning comeback in Evo Moment #37. What if we told you that you could do the same thing with just a fraction of the skill or effort? That’s what Fiora’s Riposte (W) is for.


Fiora's W

The number of Riposte spell casts decreases as MMR increases.

You can use it reactively or to make hard reads. Use it in response to CC to stun your opponent (see that Morgana Dark Binding coming at you?) or shield yourself from big damage (like a Darius execute). If you misuse it, you snare yourself in place for .75 seconds while it’s being cast, giving your opponent plenty of time to reposition around you. If you land it properly, you survive a huge burst of damage and might even turn the tables on your opponent with some hard CC.



Observation 4: Regional Differences

(KR, basically)

Up until this point, we've been showing spell cast data across a few different regions, and it's all been pretty similar. But that isn't always the case, mostly because Koreans party differently than most other regions.

Udyr

Korea

Udyr

Global

All around the world, Udyr players tend to use Phoenix stance (R) less as MMR increases—except in Korea. It's the one region that uses Phoenix stance more at higher MMR’s. Perhaps Phoenix stance is better for applying solo split-push pressure around the map? Its waveclear damage is unmatched by the rest of Udyr’s kit, especially if supplemented by an AP jungle item such as Runic Echoes. We don’t actually know—we don’t play Udyr. Let us know your theories in the comments below!


Trundle

Korea

Trundle

Global

Another example of regional differences is the decline of Trundle Chomp (Q) in—yup, you guessed it, Korea—as MMR increases. While it's the spell of choice for top lane melee matchups and jungling, the spell is actually rarely used when Trundle is played as a support—in that role, his Pillar of Ice (E) is more widely used instead as a momentary CC/engage tool. In most regions, we see that Trundle’s Q increases with MMR, suggesting that players are playing him and using his Chomp more often in higher MMR melee top lane matchups. However, in Korea, we see Trundle being played more as a support in higher MMR, which, of course, gives him less opportunity to walk up and auto attack his lane opponents.




If you've read this far, congrats: you have a Diamond-level MMR in reading and chart appreciation. What do you think your champion pool’s charts look like? Are you spamming some spells too often and other spells too little? Have any questions or comments? Our data people are gonna hang out in the comments below for a while, so ask us anything!