Jungle strength over time

tl;dr You were right about Sejuani—but maybe we should befriend Amumu too.

Welcome back for the final installment of the LeagueMath.com season 5 per-role analysis of champion strength over time. Be sure to check out the earlier posts in this series:

• Marksman strength over time
• Mid-lane strength over time
• Top-lane strength over time

The plots below summarize data from 1,049,917 matches (which is quite a few). This collection of matches is all from NA ranked solo-queue, played on patches 5.2-5.7 and includes games played by players of all skill levels.

Here is the game plan:

1. For each popular jungle pick find all the games where that champion was played as a jungler (there are no Gragas mid, top, or support games in this data for example).
2. Bin those filtered games by game-length in two-minute groups from 20 to 60 minutes.
3. Then, for each bin calculate an 'advantage' value for that champion in that bin. The advantage will be 1 if teams with that jungler won every game in that bin, -1 if they lost every game in that bin, and linearly interpolated between otherwise.

Disclaimer: An analysis like this cannot even begin to tell you exactly which jungler you want on your team at any given point in time. League of Legends is a team game, and producing plots like these is an exercise fraught with myriad difficulties both technical and social. So, take a deep breath, relax, and join me for a fun-filled and hopefully thought provoking (if not sufficiently statistically significant) romp through jungle champion power curves.

I will begin with the early game. These champions win more frequently in games that end earlier rather than later.

These four champions have similar-ish plots that skew toward the early game. They all have good sustain, and movement speed buffs and decent ganks both before and after level 6. None of these is a super-reliable carry in the late game though, so I am not too surprised by the shape of these plots. One might argue that Nidalee can carry hard in the late game, and I agree with this, but I think the mechanical skill required to do so rules her out from consistently doing it.

Skarner's plot is a bit ragged here. I cut the analysis off at champions with more than 20,000 jungle games in the one million plus game dataset. He was the closest to that cutoff, and so has the least data behind his plot. The smoothness of plots like these follows from the strong law of large numbers and the central limit theorem of statistics. Without more data to work with, all we can do is squint, imagine, and then outright lie (ily).

Continuing with the early game...

These four also have similar plots. Before 30 minutes or so, they have more of an advantage than they do after that point in the game. Gragas and Hecarim never have too big an advantage (notice that the y-axes on the plots vary), while Pantheon and Rek'Sai seem to do best when the other team forfeits.

If the enemy team is sporting one of these, then perhaps it is best not to give up too early. Of course, it is always best not to give up too early. So, no earth-shattering insight here.

Rounding out the early game we have a pair of bruisers in the jungle:

Both Udy and Volibear start out strong and carry their advantage a bit further into the game. Note that the noisy spikes toward the right-hand side of these plots are just that (noisy). Unfortunately from a data perspective, few games go super-duper late. For instance, that last bin of Volibear games between 58 and 60 minutes is only based on 88 matches. The bar between 56 and 58 minutes draws from 177 games and so is somewhat more likely to be accurate, but not much. Ha!

Both these picks do have an early advantage by the measure used in this analysis (many bars above 0), and they share the attributes of not being too mechanically intensive.

Here is an interesting plot.

Amumu is the only champion with this type of shape to his plot. It seems that his kit gives him a unique advantage near the 35-40 minute mark. The agrees with my personal experience; he gets tanky, and then uses his ultimate to engage good fights that either win his team the Baron or inhibitors around that point in the game. With basically all of his bars well above zero, he will have one of the highest win rates among junglers for sure.

Here are some tanky junglers:

All three of Malphite, Maokai, and Shyvana have relatively flat plots by this measure, meaning they do not confer a strong advantage at any specific point in the game. It also means that they are never at a strong disadvantage either.

Notably, these three are also approximately equal in popularity; all quite a bit less popular than some other jungle picks.

It might be a mistake to sleep on these picks in solo-queue. They are not likely to win you the game outright, but they are also unlikely to lose it by themselves.

Here are three more brusiers who lie slightly outside the current meta:

Of these three, it seems that Wukong performs the best, though Xin Zhao may be a slightly better pick for early game strength. Nocturne has one of the flattest plots I have seen in this entire analysis (any champion, any role). As the game continues to develop season-to-season, I would not be surprised to see him make a surge when current choice picks get nerf'd and fall out favor, or if he gets a tiny buff at some point.

Here is an interesting group of junglers that were all super-popular in season 4:

In these four I see disadvantage. Kha'Zix and Rengar do appear to have more advantage when the game goes late, but in games less than 40 minutes these two are losing a good fraction. Lee Sin and Jarvan are rank 1 and 3 in terms of popularity, but are getting crushed in terms of advantage at basically every point in the game. The season 5 jungle items do not favor these picks, but people continue to go to them. Combine this with the intense mechanical skill-ceiling of Lee Sin and we have a recipe for elo disaster.

Here is an interesting pair of AP jungle surprise attackers:

With these two it seems best to either win early by intimidation or to have the game go super-late. Other champions have their advantage leaning toward one or the other of early/late, but these two can go either way. Shaco wins late by split-pushing (in my experience) and Fiddlesticks wins by initiating a game-ending fight. This provides some evidence that Riot has done well with their initiative to enhance strategic diversity in jungle picks.

Here are a pair of tanks that ease us into the late game:

Neither of these is super-popular in the current meta (as evidenced by the relatively small number of matches), but both certainly have plots that trend up as games go later. I think Rammus is a bit easier to play than Zac, which may account for his slightly better win rate. The low popularity of these picks may be deserved; there are statistically better late-game junglers to choose from in the season 5 meta.

Two point-and-click ultimates that secure ganks:

Vi and Warwick have similar plots, which is perhaps unsurprising. Both have a disadvantage in the super-early game since their ganks are relatively weak before level 6. Both trend up to a respectable advantage in games that go over 35 minutes. Vi is a lot more popular than Warwick however; her Vault Breaker skill that allows her to go over walls completely changes the topology of the map for her, which gives her many more options at almost every point in the game. The halcyon days of Warwick + Skirmisher + Devourer + Blade of the Ruined King are long past.

Here are the latest of the late game Junglers

Fizz is still trying to find his identity in season 5, and six-item Master Yi is as silly as always has been. But, perhaps it is interesting to see Nautilus with a plot that vaguely resembles these other two. What is it after 45 minutes that would give him a large (and increasing) advantage? I would not have predicted this particular shape for his plot.

The season 5 jungle has been all about Sejuani, so why not end the day on her?

This plot is just silly. Sejuani confers a sizable advantage at almost every point in the game. Her ultimate is hard to hit (well, harder than Vi or Warwick, ha!) and some of the other skill-shot champions have not fared well in this analysis, but Sejuani is such a good fit for Cinderhulk that she has literally run over the competition since its introduction. I wonder how long her reign of terror will last. Anyway, all aboard the freelo train while it rides.

Yowza, 31 junglers all popular enough to support this analysis. That is by far more than the other roles. Riot's initiative to increase strategic diversity in the jungle appears to be quite successful. I appreciate you making it all the way here to the bottom of the page with me. What a journey.

There are a million ways to slice data like this, and I made a great number of choices and assumptions in my analysis. I am sure I screwed up somewhere. Get at me on twitter and share your thoughts.

I think this is it for champion strength over time. I have a few ideas simmering for future posts, so be sure to come back every week for fresh insights. While you wait, check the index and see what you have missed. Then, subscribe to the rss feed so no one reads the next post before you.

That is it for now. LeagueMath.com is ready for a nap—to rest up for all the League Math you can muster.

Only Krugs, Murkwovles, Gromps, Razorbeaks, Sentinels, and Bramblebacks were harmed in the making of this article.

Peace.