Gold, XP, and Role

12 min readAug 19, 2017

In the previous post, we saw the general structure of our League data scraper and how it got us 20k matches. In this post, we’ll analyze these matches to try to answer the following questions:

What is the average GPM over time for every role?
What is the average XPPM over time for every role?
How long are XP gaps and who can profit from them?
How much does a death cost in terms of missing XP and gold?
How much gold do people have at each level?

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 20-minute mark, because this ensured a fairly constant number of games, and there’s enough variance between games that not much coherent stuff can be learned in the 30 mins range anyway.

TL;DR

GPM starts at 320 growing linearly to 440 over 20 mins for all roles except support. Support GPM starts at 206 and grows linearly to 340.
Junglers have a 100–150 gold advantage against other lanes at minute 4, falling down afterward. ADCs equalize this advantage at min 10, Midlaners at min 12, and Toplaners at 17.
A lane death between minute 2 and minute 4 costs about 75–100% of a level, reduced to 30% with TP. After minute 5, 15 seconds of idleness costs 20–15% of a level (depending on time). Additionally, 15s of idleness also costs from 50 to 80 gold for roles other than support in unrealized gains.
Midlaners are the first to hit 6, with a 30s window against top, and a 1:30 window vs junglers, which makes them a lot more dangerous from 7:00 to 8:30. Midlaners have a window of over 3 minutes against botlane.
Rough estimate: Players seem to go back every three and a half minute, with about 1.3–1.4k gold in the bank (support with 0.9-1k).
Read the article for the expected gold for each level of each role.

Why does GPM matter?

If you want to do proper Theorycraft, one of the first things to consider is how much gold you can play with at any point in time. Making wrong assumptions about how much gold a top laner has at level 6 can mean the difference between a viable build and a terrible build, and we’ll see that there are massive gold differences between roles for the same level, which can make guess work, or personal knowledge about one lane, useless. So, how much gold does a player have over time? Well, here’s a graph that shows exactly this:

The “Alright, we’ve learned nothing” chart

Well, we’ve learned… nearly nothing? We already knew that supports earned less than other lanes. And, otherwise, it seems like everyone earns about the same. But is that correct? No. If we squint our eyes and look very closely, we can see that the jungle line is slightly higher at the beginning of the game, and slightly lower at the 20 minute mark. On the other hand, the ADC line starts at the bottom and grows to be the highest at 20. However, it’s pretty hard to see clearly how everything evolves over time. So, how about plotting the evolution of gold per minute over time?

A slightly more interesting graph, at least we can see that GPM increases somewhat linearly

Alright, this is not much better than the previous graph, but at least we can see that the GPM is not exactly constant over time, but seems to be increasing linearly (220–330 for support, 320–440 for the others), which would make the total gold not a linear function of time, but a quadratic function. We also confirm that the jungler has the highest GPM for the first 2 mins, before falling off relatively hard compared to the other roles. But how does that translate into gold advantage? How much gold does the jungler have against the other roles?

Support role was removed from the graph for obvious reasons

Now that’s interesting! Junglers have a peak gold advantage at minute 4, with about 150 extra gold over Top, or 100 over mid and bot. This gold advantage slowly decreases over time, with the fastest decrease being ADCs, who equalizes gold at the 10' mark, shortly followed by midlaners at 12'. However, the gold advantage stays relatively important for much longer against toplaners, equalizing at 17'. This suggests that a Jungler should strive to have a strong impact before min 10, while he holds a possible itemization advantage (we’ll see in a future article if that gold advantage can actually translate into an itemization advantage), as his gold advantage is short lived.

As a bonus, here’s the same chart but the gold difference between the jungler and each role is represented as a percent of the total gold earned by that role over time.

This chart demonstrates even further how short-lived the gold lead is, as a 5% gold lead is not exactly noticeable, which gives the jungler 6 to 7 minutes to illustrate himself bot or mid, and about 10 minutes to have an impact top.

Why does XPPM matter?

We’ve seen that Junglers had an early gold advantage, which can translate into a potential itemization advantage. However, in the early game, each level is major step-up in terms of damage, tanking, and utility. Gaining 80HP when you have 500HP definitely feels not the same as when you already have 1.5k HP. This is why XP per minute matters. It is obvious that the botlane has an XP deficit compared to other lanes due to sharing the XP of a single lane, but what about Junglers? How much ahead/behind are they against the other lanes? Understanding these dynamics can be the key to correct paths in the jungle, as well as understanding when one is weak and when one is strong. A final point to take into account is that while XP is the thing we earn when killing things, we are not really interested in the XP gains per se, but rather level gains. So we’ll switch between the two metrics depending on which makes the most sense to analyze.

Note: If you want to know how much XP you need to get from one level to another, you can use this simple formula:

XP = 80 + 100 * NextLevel

So let’s start by graphing the evolution of levels over time by role:

Meh. Like the Total Gold over Time, not much can be learned here.

Alright, like with the total gold graph, not much can be learned from this. We can see that the Mid and Top level up at basically the same rate, which is to be expected, since there are no difference in their XP source (solo XP from minions). The duolane is behind the solo lanes by quite a margin, although the support sacrifices himself even more to let the ADC gain more XP. By drawing a line for the level 6, we can see that the solo laners reach level 6 roughly 1:30–2:00 before the jungler and about 2:30–3:00 before botlane, which can be a nice window solo laners can use to make plays.

Here is the same data, but as a table, to make reading easier:

Let’s do the same as we did with Gold and look at the evolution XP per minute over time:

The oscillating XPPM for the Jungler may be due to Blue and Red buff XP.

This is quite interesting! We can see that the Jungler starts off higher than anyone else at the second minute, and gets decent XP gains on minute 3, but then drastically drops off on minute 4 and 5 before picking up again, whereas the curves for the other roles are much more stable. There are also strong oscillations in the XPPM of the jungler, which suggests the existence of “void” periods where there isn’t much XP to gain on the map for the Jungler. Additionally, it seems that the peaks of these oscillations are about 6 minutes apart, which would suggest that this bonus XP comes from the Blue and Red buffs, and that they have a rather strong impact on the XP a jungler can get, although this is only speculation. As with the gold, let’s see how much of an advantage the Jungler has in XP against each role, as a percentage of this role’s total XP.

If the gold advantage the jungler over the other roles lasted for 7 to 10 minutes, it seems that the XP advantage closes much more rapidly for the solo lanes, while existing nearly indefinitely for the duo lane. Indeed, the XP gap with mid closes nearly immediately (ahead for 1 minute), while the gap top lasts just a tad longer (2 mins). This means that except for a cheesy gank at minute 3, the midlane will always hold an XP advantage against the jungler. On the other hand, there is a 2 minute window during which the toplaner is behind in XP on the jungler, which coincidentally closes about 30s after the jungler has finished his first jungle clear (for meta jungling champions), and makes the toplane a great target for a first gank. On the other hand, there is no XP pressure on the jungler to gank bot, which means that the jungler can use more opportunistic strategies for the botlane compared to the toplane.

Total Gold per Level

We’ve seen what the gold and XP progression were. We can now combine them to see how much gold should one expect when reaching a certain level.

Average Total Gold when reaching level X (level 10–11 were not included for support and ADC as they happen after 20')

This table is particularly useful for Theorycraft, as it gives you a reasonable piggy bank size for your builds. And, as anyone can see, there are massive differences in terms of how much gold you can expect from each role, with the poorest role for each level being the solo lanes. This is naturally due to the fact that they are faster at gaining levels than the other roles and have therefore less time to accrue gold.

Cost of Idleness

This section aims at answering the question of the cost of a gank. This is slightly more iffy than the rest, because it relies a bit more on derived data. We’ll suppose that a gank intervention lasts 15 seconds — including time to walk to the lane, which is on the lower end of the spectrum in my opinion. If a gank succeeds, the rewards are obvious, +300g for the killer, plus some gold for the assisting participants, and some XP. If a gank fails, with one of the gankers getting killed instead, the cost is also obvious. However, if a gank simply fizzles out, where the jungler comes, waits a few seconds, and then leaves, how much was lost?

Note that if you believe the average time taken by a gank is closer to 30s, you can simply double the calculated cost that we’ll expand on below.

XP Cost of Idleness

The first metric we’ll look at is the XP cost of idleness. To represent that, we could just graph 0.25 * XPPM / TotalXP for each role, which would give us how much XP we lost (or more accurately didn’t gain) during the 15 seconds. However, we, as players, are not used to manipulating raw XP values. So we converted the TotalXP into the corresponding level for each minute, and then calculated how much of the XP needed to reach the next level did 0.25 * XPPM represent.

Quite interestingly, the XP cost of being idle is relatively the same for each role, with the cost being slightly lower for Supports. The monstrous peak at the 2–3 minute marks tells us how terrifying an early death (aka best idleness) can be, as you’d need to take into account walking time, for a total time of around 35-40s depending on the lane, which can mean a complete level advantage for the enemy. However, past the 5' mark, the XP cost becomes a nice decreasing slope going down from 20% to about 10% at the 20' mark.

Gold Cost of Idleness

Measuring the unrealized gold gains are trickier than it looks, as one needs to take into account ambient gold gains which only start at minute 2, and are never not earned. However, once this is taken into account, the formula is relatively straightforward, and gives us the following graph.

As we can see, for roles other than support, the unrealized gains for 15s of idleness increase from 50 to 80 over the course of the game. This means that a single control ward making the Jungler waste 15s of time is enough for it to be worth it in unrealized gains. So maybe buy control wards?

Current Gold, a sneak peak into the future

Alright, we’ve answered about every questions we asked ourselves at the beginning of this article, we’ll now do a cool analysis — that’s in no way accurate — of the current gold each role has, and see how that translates into backing timers.

We know the average GPM for each role, and we also have access to the current Gold that a player has on him every minute, also known as Unspent Gold. Now, if we assume that the GPM is constant over periods of a few minutes, we can suppose that the gold a player has on him grows linearly. And the average of a linearly growing function is it’s middle point over the range, so we can deduce that the maximum unspent gold is twice the average unspent gold. From there, since we have a given gold per minute, if we divide the maximum unspent gold by the GPM, we get the duration of the stay, or the time between each backs, which seems to be about 3 minutes and 30 seconds.

This is a rough estimate that uses quite a few assumptions we know are not totally correct, but we’ll see in a future article how we can do a better analysis of the 20 thousand matches to get a clearer picture on the backs, and we’ll be able to compare it to our current guesstimate.

Conclusion

There are notable differences in both Gold and XP for the Jungler with respect to the other lanes in the early game. The fact that the Jungler is ahead of the toplaner only for a couple of minutes, most of which he has to spend farming, makes the time around the 4' mark a very clear window of opportunity that the jungler can exploit. The next window of opportunities top are not dependent on an XP advantage, but should rather depend on an itemization advantage. This itemization advantage is something that can only developed further once we have a clear understanding of when each role backs, as this is when the unspent gold is converted into actual items, and will therefore only be discussed once we have clear timers for each back. We’ve also seen how the Jungler has a dangerous XP disadvantage against solo laners from the minute 7 to 9, as this corresponds to a period where the Jungler is below level 6, whereas the solo laners are 6+. This means that the risk of death is higher during that period, and this in turn suggests that this period of time would be a good period of time to focus on the botlane.

What can be gleaned anyway from these results is that the Jungler is a critical role, that is subject to strong pathing constraints due to the XP and Gold evolution over time, which can make it predictable (and analyzable). In the next article, we’ll therefore focus on the evolution of the chance of kill and death of the Jungler against each role over time, and see if we can learn anything about the successful/unsuccessful pathing of Junglers.