Warding and Vision Gaps

15 min readSep 13, 2017

Previously in our series of articles, we covered which lane junglers tend to focus at a given point in time, when junglers were ahead or behind of each lane, and when each lane generally goes back to base. This general knowledge is already quite useful, both as a jungler and as a laner, as it can give you clear hints about when you are vulnerable or not. We can venture that laners tend to ward accordingly, so that they can safely ride out their periods of weakness. However, this is merely a conjecture. In this article, we’ll try to see when laners actually ward, and see if we can find exploitable gaps in vision.

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.

Riot does not provide the location of wards placed. This means that this analysis will only be about when wards are placed rather than where.

TL;DR

The duration and cooldown of the yellow trinket greatly influence when people ward or not. This create clear vision gaps before the 6' mark and before the 8:50 mark across all lanes.
The vision gaps are most marked in the botlane, with an absolute difference between best coverage and worst coverage of around 33%, compared to a difference of about 10% for the other lanes.
Despite the vision gaps, botlane still has the best ward coverage of all 3 lanes, with a worst coverage of about 50%, and a best coverage of 91%.
Both Mid and Top worst coverage are at the 6' mark, with about 30% coverage.
Junglers ward most around the 4' mark. This timing coincides with the moment most junglers have finished clearing their jungle.
Junglers purchase Tracker’s Knife or its upgrades in about 30% of the games.
Tracking Knife has a very strong impact on vision, representing an increase in total vision of over 20% past the 8th minute.
Most early stealth wards can be attributed to Tracking Knife Junglers, with Supports evening out Junglers only after the 12th minute.
Supports purchase about 4 control wards per game. Junglers purchase around 2.3. Mids purchase about 1.7. Tops purchase about 1.4. And ADCs purchase the least control wards, with about 1 control ward per game.
37% of all the control wards that have been placed during a match are not cleared by the end of the match.
On average, there’s about 1 control ward on the map at 7:00, 2 control wards at 13:30, and about 3 around 25:00, for each team.
ADCs are the least concerned with purchasing a control ward, with 50% never buying ward, and a half-life of 1 control ward. This means that for every control ward an ADC buys that gets cleared out, there’s a 50% chance he won’t ever buy another.
Top and Mids are also not very concerned with control wards, with 26% and 31% of players never bothering to buy one. Both roles also have half-lives around 1, with Mid being a little bit more persistent (1.27).
20% of Junglers never buy a pink ward. The other 80% on the other hand are a lot more persistent, as their half-life is of ~2 (1.74). This means that you need to clear 2 control wards from the junglers for 50% of them to stop buying pink wards.
Over 88% of Supports buy at least one pink, and their half-life is the longest, being slightly over 3, meaning that clearing 3 control wards from a Support will only discourage 50% of the 88% from continuing to purchase control wards.

Not All Wards Are Born Equal

Before we start looking into how players ward, I’ll do a quick presentation of the differences between each types of wards. First, there three types of wards: Farsight wards — aka blue wards, control wards — aka pink wards, and stealth wards. However, the last type of wards should actually be split in two: regular stealth wards, and warding totem stealth wards — aka trinket wards. Indeed, while regular stealth wards always last 150 seconds (2:30), a warding totem stealth ward duration ranges from 60 seconds to 120 seconds based on level. Another reason to distinguish between the two is availability of the ward. Trinket wards are strongly constrained by their charge cooldown, which ranges from 180 seconds to 90, depending on level, whereas regular stealth wards are only obtainable through items but are basically recharged for free on every back. These two crucial differences have a noticeable impact on how these wards are used, as we’ll see.

Here’s a reminder of when players usually get a new ward from their yellow trinket:

0:05
3:05
6:00
8:45
11:20

What Wards Do Players Use?

In this section we’ll see how each role shapes what warding items are purchased. Most of this section will be beyond obvious for non-beginners. You can skip ahead to the next section without missing much if you think you already know this stuff.

Yellow Trinkets

Obviously nearly everyone buys it. (Nothing very interesting to say about it)

Farsight Trinkets

Support and Jungle almost never buy the Farsight Alteration, while ADC make the most use of this upgrade.

Control Wards

Control wards are a much more interesting story. As we can see, Supports and Junglers represent the majority of the control wards placed during a game, with over 60% of all control wards being placed by these two roles. ADCs place on average only one single pink ward. A. Single. Pink. In an entire game.

Stealth Wards: Sight Stone & Tracker’s Knife

Other roles not included because their purchase rates are ~0%

Unsurprisingly the sole two providers of stealth wards are the Support and Jungler, and while the Support almost always buys a Sightstone, Junglers only buy Tracker’s Knife (or its upgrades) in about 30% of games.

When Do Players Use their Yellow Trinkets?

Now let’s get to the more interesting parts. Yellow Trinkets are actually pretty weak in the early game, with a duration of 60 seconds for a cooldown of 180s at level 1. This large gap between the duration and cooldown will create gaps in vision during which the laner will be blind.

We’ll start by looking at when players place their wards depending on their roles. To do so, we grouped ward placement in slices of 6 seconds and compared them to the total number of games.

Alright, this is a mess, but we can already learn a few things if we squint our eyes a bit. If we look closely at the Mid, Top and ADC lines, we can see that there are very strong fluctuations. We can also notice that the Top and ADC lines are in sync, dipping and rising at the same time. Finally, we can observe that the periodicity of the fluctuation of this oscillation is 30 seconds. What we see there is the “fixing” effect of waves on laners: When a wave is present, laners are less likely to move to ward.

Here’s the same graph, but with only Top and ADC on the graph, so that we can better see this effect:

On average, ADCs are 1.5x more likely to stay in lane if there is a wave, while Toplaners are 1.8x more likely, and Midlaners are 2.2x more likely. I’m not sure if there is any way to use that knowledge to any kind of advantage.

Yellow Ward Presence

So we’ve seen that laners tendency to ward is strongly determined by the presence of a wave or not, and we’ve got our average number of wards per slice, which is nice, but not exactly that helpful. Indeed, from the point of view of a Jungler, knowing when players ward is not exactly what is interesting, unless you want to come over and catch them when they go to ward. Junglers are much more interested in knowing whether a ward is present or not.

Apart from a few technical details, determining the presence likelihood of a yellow trinket ward is actually rather simple. All you need to do is to sum the number of wards placed in the 60s prior to that point in time, as this represents the duration of the ward. As you may remember, the duration of a yellow trinket is not actually 60s, but rather a function of level that starts at 60s. So, in order to be more accurate, I also took into account the average level of each role at each point in time to have a slightly better estimate of the duration of the wards.

Now that we’ve seen how to obtain the ward presence likelihood, let’s plot it.

There are 3 very clear dips in yellow trinket warding presence. These dips happen for every role at the same time and correspond to the 30–60 seconds right before the yellow trinket ward renewal (6:00, 8:45, 11:20), and are a lot more pronounced for the bottom lane than for other lanes. Indeed, the botlane oscillates between 91% ward presence at their peaks and around 50% at their worst. This massive difference can be the key between a successful gank and a failed one, and therefore provides Junglers with clear windows of ganks on the botlane. We’ll see how those gap align with first objectives taken in a future article.

For Mid and Top, the cooldown effect is a lot more nuanced. However, both roles still have their worst ward presence around the 6' mark with around 30% ward presence, and their overall ward presence is much lower than botlane. This means that in general, solo laners have less vision, and the low oscillation suggests that their warding patterns are a lot more random compared to the ones from the botlane.

Finally, the Jungler have a massive warding peak around the 4th minute. This is probably explained by the fact that Junglers are done with farming their jungle around that time.

Stealth Ward Presence: Tracker’s Knife and Sightstone

In the previous section, we’ve defined how to calculate the ward presence likelihood for yellow trinket wards from a table containing the times at which players placed their wards. We can naturally do the same with stealth wards by adjusting the duration of our window to 150s intead of 60s.

We’ve already established that only Junglers and Supports ever had the means to place stealth wards. However, Junglers and Supports do not have those means right away, as they need to purchase their corresponding items: Tracker’s Knife and Sightstone. This means that the placement of stealth wards is strongly constrained by the time of purchase of these items. However, we won’t go into the evolution of the purchase of these items over time in this article, but instead will only focus on the average warding presence in the case where a player buys these items in a game. While this gives us less insights into how players actually use these items, it will allow us to extract knowledge that can be used before the game begins, which will help us create game plans for the jungler in the next article.

So, let’s plot the average stealth ward presence by source over time:

Interestingly, a Tracking Knife Jungler plays a much more prominent role in gaining vision early in the game compared to Supports, which is probably explained by the fact that the Sightstone purchase usually happens later compared to a Tracking Knife purchase. To better illustrate this difference, we’ve also plotted how many of the stealth wards can be attributed to each role, over time.

Tracking Knife plays a pretty big role in early game vision

Remember that Junglers only purchase Tracking Knife in about 30% of the games. So how much more vision do Tracking Knife Jungler’s teams gain compared to teams with Junglers that take another jungling item?

The difference in vision is actually quite massive, with Tracking Knife Junglers providing about 20% more vision to their team past the 8th minute mark.

Control Ward Presence

Control Ward Presence is going to be tricky to evaluate, mostly because control wards last indefinitely, which makes their lifetime dependent on when they are killed and Riot does not provide us with anyway of identifying which Control Ward was killed. So we can only do very a very basic analysis of this type of ward.

We’ll start by plotting the number of control ward placed and killed over time.

Cool. We see the same 30s periodicity we had on the Yellow Trinket, which is logical. However what is much more interesting is the permanent gap between control wards placed and killed. At any point in time, more control wards are being placed than are killed. A look at the asymptotic behavior of this difference suggests that about 37% of all control wards placed are not cleared by the end of the game, which is a rather surprising result, and show how valuable control wards can actually be.

Another way to look at control wards would be to measure the evolution of the number control wards each team has placed on the map over time, how many of these have been killed, and also how that translate into average control wards present on the map at any point in time.

You can see how our model fairs against the actual data

The Control Ward Presence over Time has a fairly regular shape and actually looks a lot like an exponential of proportional growth, a term you may remember from your chemistry classes. This type of model has a few interesting properties.

First, it has a maximum value, which, in this case, represents how many vision wards you can hope to have on the field at best. This number is naturally below 5, and is actually around 3.7. This means that it is very unlikely to have 4 to 5 control wards on the map over long period of time.

Second, this type of curve has a constant half-life. This means that it takes the same amount of time to go from 0 control wards to (3.7–0)/2 = 1.85 control wards as it takes to get from 1.85 pinks to (3.7–1.85)/2 = 2.78 pinks. This half life is around 8.9 minutes.

This model therefore gives us a predictable level of vision for any point in time. For instance, there’s about 1 control ward at the 7th minute, 2 control wards around 13:30 and 3 control ward after 25:00.

Farsight Presence

Due to the length of the article, I won’t develop this much. Just know that about 28.5% of blue wards survive to the end of the game. And, on average, teams have 1 blue ward on the map after 24:00 and about 1.37 blue wards on the map past the 30-minute mark.

Control Wards: Willpower and Abnegation

To understand the impact of control wards on vision, we’ve looked into when players place them on the map and when they are killed, and deduced their average presence over time.

Another way to look into this is to look at control ward purchases and its evolution over time.

There are clear differences between each role when it comes to purchasing control wards. Supports are the most dedicated with over 15% likelihood of purchasing a pink ward every minute, which translates into a 50% chance of purchasing a pink ward on every back. On the other hand, solo-laners only seem to care for the first ward, after which their purchase rate drastically diminish. ADCs are the worst of the bunch, barely attempting to purchase a pink at all.

However this type of analysis is still quite unclear, as it is hard to represent what a 5% likelihood of purchasing a pink per window of 60s actually means in terms of control wards over time, and shifts in those percentage is even harder to comprehend. So we need another approach.

We’ve seen that these percentage were decreasing over time, this probably means that a portion of players tend to stop purchasing pinks after they’ve already bought a certain number. Just by looking at these numbers, it’s impossible to tell whether that’s true or not. However, getting data to answer this question is not very difficult. Below, you’ll find the distribution of control purchases per role:

Alright, this is pretty good! We can already see that the proportion of players willing to shell out to keep purchasing pink wards dwindles quite rapidly, and that this is most notable for ADCs, where 50% aren’t even willing to pay for a single control ward. I sure hope that their builds are hyper optimized and that there’s a good reason 50% of ADCs can’t even afford to spend 75 extra gold in an entire game. And here’s a friendly reminder that, at the 24 minute mark, a cannon minion + a single melee minion are worth ~75 gold.

So this graph is already pretty telling, and almost damning for ADCs. But we can do better. What if instead of plotting the percentage of players that stopped at a given number of control ward purchased, we plotted the percentage of players that are still willing to purchase another control ward. This would give us the evolution of willpower of each after a given number of purchases.

This is probably one of the coolest graphs I’ve made in this whole series of article. Not only is the meaning of it relatively simple to understand, it’s also simple to read. For instance, with just a single glance at the graph, you can tell that only 48% of Midlaners will ever buy more than 1 control ward, or that no more than 20% of Toplaners will buy more than 2. But the best part about this graph is that all the curves can be modeled using a decreasing exponential, which gives us a constant half-life for each role.

This half-life can be understood as “How many more control wards do I need to clear before the half of the other players just give up on purchasing them?”. This gives us a clear and powerful model about the behavior of each role when it comes to purchasing control wards. So, here’s the half-life table for each role and also a table that gives us the relative percentage of players that give up whenever you clear a control ward, which is an other representation of the half-life.

Half-life value for each role — Higher is better

Relative percentage of players that give up after each ward cleared

Combining these tables with our previous graph really puts ADCs on the spot. On top of not purchasing even one pink in 50% of games, ADCs are the fastest to give up on purchasing new control wards when their previous one has been cleared, with over half of ADCs giving up on each step of the way. The same goes for Toplaners and Midlaners who also tend to give up very rapidly, though they compensate it with a slightly higher initial purchase rate. On the other hand, Supports, who already have the highest purchase rate, are the most resilient, with more than half of them persisting after 3 control ward cleared.

Conclusion

In this article, we’ve discovered how the cooldown duration of the yellow trinket shaped the amount of vision each lane, and how this translated into 3 clear vision gaps around the 6:00, 8:40, and 11:20 mark.

We’ve also looked into how players compensated that through purchases of either stealth ward items, or control wards. We already knew that supports and junglers did a lot of the warding and vision control, but this article really brought out how little laners actually care about vision. Any laner that purchases 3 control wards in his game is basically a unicorn, whereas this is actually fairly standard for supports and junglers, with 60% and 40% respectively buying 3 or more wards. We also developed a few models to explain the behavior of each roles when it comes to purchasing control wards. These models can maybe be used as in-game knowledge about.

Finally, we analyzed how the purchase of Tracker’s Knife influenced the early stages of the game in terms of vision, and discovered how massive this impact was.

Now that we’ve gained a fair bit of understanding about the warding behavior of each lane, we could try and combine it with what we learned in previous articles and attempt to model some intelligent jungle paths that would make use of this newly acquired knowledge.