Jungling, Kills and Deaths

11 min readAug 19, 2017

In the previous article, we’ve explored GPM, XPPM and saw how the junglers held a small gold advantage for the first 10 minutes of the game over the lanes, as well as a very short-lived XP advantage against toplane. We also estimated the size of the window during which solo lanes are level 6 while the other roles are not. Finally, we put a price in terms of unrealized XP and Gold gains on idleness. Overall, we’ve realized how constrained the Jungler was in the way it could optimally play to his strengths.

In this article, we’ll try to see what that implies in terms of kills and deaths for the junglers by trying to answer the following questions:

Who and when do Junglers kill?
Who and when kill Junglers?
Do Junglers actually use their very-short-lived XP advantage on solo lanes?
Are Counter-Ganks impactful?

Caveats

Most of what will be talked about in this article is stuff most people already intuitively know. This article just reinforces this intuition by putting numbers on it.

The data we are using is from high ELO players (dia2+), who *may* be more efficient and more knowledgeable about the game than the average player.

This is aggregated data. The findings presented in this article are not to be taken as gospel, but rather as a general rule of thumb.

We’ll only be looking at the evolution of the game before the 10 or 20-minute mark, as this was determined to be the period during which the Junglers held a true advantage over other lanes.

There was a bug during the acquisition of the data, and we could not get the assisting participants during kills, which the data we’ll discuss are only the events were a Jungler was the killer, or was killed, and we won’t have any additional information about who was present at the time of the kill, so if a Toplaner killed a Midlaner, with the assist of its Jungler, it won’t show up in the data we’ll be exploring.

In the event of a Jungler on Jungler kill, we cannot tell which of the two junglers was the first to initiate a fight, and which one was the one to do the counter-gank, as the data Riot provides us is not detailed enough for this analysis.

TL;DR

At the 4' mark, blue side junglers kill a lot more toplaners (15% kill chance) than red side junglers (9%), who have a higher kill rate against the botlane (2% vs 4% for both ADC and Support). This suggests that a larger portion of blue-side junglers start red-buff side, whereas the results should be split a bit more evenly for red-side junglers. This may be due to the attractiveness of the Raptor/Red start. A guesstimate puts blue-side red-buff start at 75% min likelihood, and red-side red-buff start at 50% max likelihood.
Before the 10' mark, the most dangerous player on the map for the jungler is the other jungler, with the midlaner following closely once the game progresses past the 7th minute, which is also around the time the midlaner has a level-6 advantage over the jungler. This suggests that counter-ganks are very impactful.
The highest payoff for junglers is the top-lane at minute 4. The worst lane payoff is the midlane, with a higher chance of death than kill past the 7th minute.
Jungler on Jungler kills peak at the 4th-minute mark and are strongly centered around midlane, which suggests that ganking this lane is a high-risk strategy, as the enemy jungler usually is in the vicinity.
There are clear differences in jungler gank timings depending on the side. See the table in the article for more information.
The choice of confrontation between junglers lies with the red-side jungler.

Jungler Kills

Analyzing junglers’ kills can yield a lot of information about jungling behavior. Indeed, by analyzing who junglers kill and when they do it, it is possible to extract information about their pathing, but also about the impact of the differences between sides. From prior knowledge, we know that the jungling patterns for blue and red side can be asymmetric. Using champion kills, we can learn about the position of the jungler.

The first metric we’ll analyze is the percentage of times that a jungler gets a kill in a minute, for the blue side and red side. This is done by calculating the number of jungler kills in each minute and dividing by the number of games. This metric can be useful to determine whether a side holds an advantage in terms jungling patterns.

There are indeed small differences between the two graphs, but nothing is significant enough to be able to say whether a side possesses an advantage over the other.

The second metric we’ll analyze is the first metric but broken down against each role. This is done by calculating the number of kills against each role in each minute and dividing by the number of games. This gives us the two following tables:

The only obvious observation that can be made from these two tables is that the kill chance of the jungler in the toplane during the 4th minute seems to be much stronger for the blue side with respect to red side (~15% vs ~9%). There are two possible explanations for this: either the blue side jungler has a ganking advantage over the red-side jungler when it comes to the top lane, or there are simply more blue-side junglers that go gank top. Although it is possible that red-side toplaners are more exposed to ganks than their counterparts due to the asymmetry of the lane, it should not account for an increase of 60% in kill percentage, or we’d see a similar discrepancy for the red-side junglers against botlane. It is much more likely that more junglers gank top simply because their jungle paths bring them there. A way to confirm this would be to look at junglers’ deaths against toplaners, if they also increase, then the kills percentage increase should be associated with the popularity of the route rather than efficiency (we’ll see below that it does).

This suggests that most blue-side junglers start red-buff side, whereas the starting points are more split for red-side junglers. We’ll see in a future article about jungle pathing if we can verify this or not with other data points. If we assume equal efficiency for ganking patterns (this is a big if), then the distribution of kills between bot and top at the 4th minute can give us insights into the proportion of red-buff side starts for each side. This gives us a 75% minimum likelihood that blue side starts with red-buff/raptors, whereas the red side would only have a 50% likelihood. Or, if we assume that the jungler’s behavior is the same for both side, except for the starting point, then the strong oscillations in the blue side pattern would suggest that junglers alternate between top and bot lane ganks. Then the lack of oscillation for the red side would indeed suggest that the starting points are 50%-split between blue-buff side and red-buff side starts.

Jungling Presence

We’ve seen that there were differences between the two sides in terms of jungle paths. We’ll now try to represent how strongly these differences are. In order to do this, we’ve combined the two previous tables into a single one where each cell is the ratio of the largest of the two corresponding cells in their respective tables (minus one with the value being set as negative if the red side is bigger — for visualization purposes). This gives us the following table:

Blue-side junglers are 60% more likely to gank top than red-side’s at 4', while red-side’s are ~80% more likely to gank bot.

This table shows us moments where a jungler is likely to show up in a given lane while the other jungler is in another lane, which are windows of danger for laners, as they should not expect support from their jungler during the gank. The midlane seem to receive equal focus from junglers at all time.

Jungler Deaths

Until now, we’ve seen where the jungler could be successful. However, ganks are also a risk. We’ll now look into the deaths of junglers and see if we can learn anything.

Death Percentage for Junglers of each side — Junglers are the highest threat to Junglers.

Look at that Jungler on Jungler death line! It seems that a jungler is most likely to die at the hands of another jungler with a notable spike around the 4th minute. Additionally, the first lane to catch up with the jungler on jungler death is the midlane, with a major increase around the 7th minute, which corresponds to the beginning of the level-6 advantage window of the midlaners vs junglers.

A Look into the 4th Minute

At the 4th minute, a jungler is twice as likely to die from a jungler as they are from a top or mid. This can mean multiple things: Either, after finishing their jungle, junglers tend to go invade their opponent’s jungle, resulting in a strong jungler vs jungler interaction. Or, there is a strong interaction in the lanes, which means both junglers arriving at the same lane and duking it out. Or a mix of the two.

In order to better understand which of these possibilities is the closest to reality, we’ve noted the location of each jungler on jungler kill that we recorded, binned them into a 50x50 grid and counted how many were present in each cell of the grid. This gave us the following beauty:

This “heatmap” makes it abundantly clear that the jungler on jungler interactions mostly happen in lanes, with invade kills being the minority. Additionally, we can see that most of the interactions actually happen mid, with a bit less top, and nearly nothing bot. This can probably be explained by the fact that red side starts are split evenly between blue-buff and red-buff side, and therefore can only have equal presence with blue side in the mid. This also reinforces our observation that mid lane ganks strongly expose the jungler to counter-ganks. Overall, this suggests that counter-ganks are extremely effective.

Gank Payoffs by Lane

We’ve seen how likely a jungler was to get a kill in a given lane, and we’ve also seen how likely it was that a jungler was to get killed for that lane. If we do the ratio between the two, we can see how much more likely junglers are to kill than they are to die during a gank.

A 4' gank top is 5x more likely to go well than bad, a 10' gank mid is 15% more likely to go bad than well.

Red side payoffs are more evenly distributed between the top and bot lanes.

The first notable element is that the highest payoff for both graphs is a top-side gank during the third minute, dropping to a 2x–2.5x payoff in the 4th minute. A third-minute gank, even in the 3:45 zone is pretty early, and is certainly why it is so successful, as the toplaner probably did not ward yet.

The second thing that is worth noting is that the midlane is never the highest payoff possible. Note that this doesn’t mean that one should shun the midlane; it only means that ganks there are harder to pull off than on other lanes, and by quite a margin. This analysis does not go into the impact a successful gank mid can have on the dynamic of the lane, as the payoffs we’ve analyzed are only centered on the jungler.

A cool thing about these two graphs is that they allow the creation of ganking strategies that aim at maximizing the overall payoffs, which we can see below:

Highest payoff per minute for each side — Lane focus

It seems that the highest payoff ganking pattern for blue side is much simpler than the one for red side, with toplane being focused for the first 7 minutes of the game, and botlane afterward. On the other hand, the optimal payoff ganking pattern for seems to be an alternating pattern of 1:30 minute between top and bot.

Highest payoff per minute for each side — Conflict

Due to the differences in play style between the two patterns, if a jungler is aiming at a non-confrontational play style, it can aim for ganks in the blue slots. However, the choice of the confrontation seems to lie on the red side, as the red side is the side with the high starting variance.

Conclusion

By analyzing the way junglers kill and get killed on the map over time, we’ve been able to expose differences between the blue side and red side in terms of starters in the jungle. We’ve discovered the midlane is never the safest gank and has a negative payoff once the midlaner reaches 6. We’ve learned that the 4th minute is the riskiest minute of all, with the most jungler on jungler interaction of the early game, most of it centered around the midlane. On the other hand, the highest payoffs a jungler can get all game are topside during the 3rd and 4th minutes. This suggests that early ganks top are often successful, and late ganks are more likely to be punished, either by the toplaner or the jungler.

There seems to be only one “optimal” start on blue side, which is the red-buff-side start, whereas there are two equally chosen starts for red side. These two starts make red-side junglers move towards opposite direction of the map, which means that the red side can choose whether to play a confrontational style or a cross-map style, with the blue side having little say over it. It is also likely that the plurality of the starts on red side makes it harder for players to accurately track where the red side jungler is without extensive jungling knowledge.

In the next article, we’ll try to see what objectives players tend to prioritize, and how this it can shape the flow of the mid-game.