The Art of Flanking at Valorant Masters: Berlin

Perfecting this art may even give you the upper-hand in your Ranked matches! 😏

by S4B0T4G3FIRE | October 01, 2021, 8:00 AM EDT

Abstract

With the most recent edition of the Valorant Champions Tour 21 wrapped up, it is time for Riot Games and its gaming community to analyze the event and make some micro-adjustments to gameplay. Which agent flanked the most? Which agent had the highest flank success rate? At which time during each round did the majority of flanks take place? All of these questions will be answered in this very random, yet very detailed, article.

Disclaimers: First, note that the data presented is the accumulation of the Group Stage matches only. Second, some of the sample sizes used will not be sufficient to draw general conclusions from the tournament. For instance, ‘x’ agent may have only been chosen once or twice during the entire event, so there will be very limited data about its success. If you come across data like this, just take it with a grain of salt, and move on to the next statistic.

The Best-Performing Agents

Before analyzing the art of flanking, let us establish some base data about the overall performance of each agent at the tournament. Yes, “best-performing” can be quite a vague term in a game where many possible contributions fall outside the “KDA” statistics. However, KDA and winning do tend to correlate positively, and they are the most measurable and comparable data, so they will be used to determine the best-performing agents.

Table 1: Agent Success (Sorted by Average Kill/Death Ratio)

Immediately, you probably notice that six (6) agents were chosen fewer than ten times during the whole Group Stage. They are here to fill the data, but you may go ahead and ignore them.

• It is safe to say that Jett was the best duelist of the tournament and/or was played by the best players of the tournament. She was chosen most frequently, had a very good ratio of kills to deaths, and recorded very few assists (which is actually a good thing in her case because it means she was incredibly reliable when it came to finishing what she started).
• Though quite a different agent from Jett, Reyna produced very similar statistics over a smaller sample size, validating herself as another strong duelist.
• The best supporting characters of the tournament appeared to be Sage, Astra, Sova, and Viper when played to their full potential.
• Killjoy and Skye, though popular agents, did not rack up numbers as well as they probably could have. Their utilities are strong enough, but they were slightly below average when it came to supporting their duelists. Perhaps their main purpose was to gather intelligence or buy time for their duelists to rotate, which is still very important, but not as measurable.
• Yoru was only chosen once, and it was likely only “for the memes.” When played absolutely perfectly, maybe he can perform well enough to be a viable duelist, but he is much more gimmicky than the other duelists in the game. Perhaps he will be reimagined in the future because it would be nice to see all agents have an opportunity to be utilized at the highest level of Valorant.

On the Flank

Over the beginning half of the Group Stage, gameplay was watched closely to see when flanks were attempted. A total of 216 flanks were analyzed, ensuring that sufficient data was collected for every team and every map. For both the Attacking and Defending side, any agents attempting a flank were recorded. Also, the status of their flank (Successful or Unsuccessful in getting a Kill) and the time (seconds remaining in the round) at which their flank impacted the round were recorded as well.

Disclaimer: Some context is required when determining whether or not a “push” can be considered a flank. For the purpose of consistency, the following parameters were used when recording flanks.

• Not counting simple pushes up the middle of the map (unless the player wraps around toward a site from the opposition’s spawn).
• Not counting post-plant instances where players are running around to find enemies who are attempting to save weapons for the next round.
• Not counting when a player gives up on the objective and solely seeks out “exit frags.”
• Teleports on Bind were not counted either because the nature of a flank (the element of surprise) is compromised by noise cues. If every map had teleporters, maybe then the use of them would be counted under certain circumstances.
• Two+ players flanking together is recorded as two+ individual flanks (not one flank of multiple players).

Who was Flanking?

Naturally, the agents that get chosen the most are going to flank the most, so there is some bias in this part of the study. This is still useful information, though!

Table 2: Agent Flank Success Rate (Sorted by Highest Overall Success Rate)

• Of the 216 flanks analyzed during some of the Group Stage matches, neither Breach nor Yoru attempted any of them.
• The only agent never to succeed on the flank was Brimstone.
• The most successful agent was Omen, but he only registered 4 flank attempts during the observed matches.
• The agents who registered the greatest number of flank attempts were, of course, the agents who were picked the most — Skye, Astra, Sage, Reyna, Jett, Sova, Viper, and Killjoy.
• The agents who were most successful on the flank (with a satisfactory sample size of flanks) just so happened to be Skye, Astra, Sage, Reyna, and Jett (in descending order of overall success). Funnily enough, this is the reverse order of how these agents’ ranked in terms of Kill/Death Ratios. Therefore, a correlation seems to be present: the lower an agent’s Kill/Death Ratio, the higher that agent’s success rate on flanks. Interesting!
• Overall, flanks attempted from the attacking side were 10% more successful than flanks attempted from the defending side.

When did Flanks Impact the Round?

At the immediate start of a round, it is very unlikely for a flank to occur, so maybe you can let down your guard a little bit. At the very end of a round, a flank probably will not occur because carrying a weapon into the next round is likely more important than a flank, supposing that the spike was never planted. The rest of the round is fair game, however, so you might want to look over your shoulder once in a while.

Figure 1: Frequency of Flanks Occurring Within Each 10-Second Interval of a Round

Based on 216 recorded flanks, players should be most paranoid between 1:20 and 1:11 in any particular round because that is the 10-second window when flanks occurred most frequently. In fact, a flank was 2–3 times more likely to occur from 1:20 to 1:11 than during any interval ranging from 1:10 to 0:21. By the looks of it, players in the tournament were most confident in flanking 20 to 29 seconds into a round because, by that time, they have already gathered some information on the whereabouts of the enemy team.

Also, you probably noticed that the frequencies shown in Figure 1 do not add up to 216 flanks. That is because 120 out of those 216 flanks (56%) impacted the round after the spike had already been planted. So, once the spike is planted, it might be smart to invest one player in watching the flank. If you are the one flanking, however, it might be smart to check every corner unless you already know the whereabouts of everyone on the enemy team.

Flank Times Depend on the Map

To be even more confident in attempting and defending flanks, it is important to know that every map is vulnerable to flanks at different times. After all, the “1:20 to 1:11 interval” from earlier is just the mode of the data set. How about the mean? Discovering the mean for each map is still helpful because it tells you whether or not a map like Ascent is most vulnerable to flanks earlier or later than Icebox might be.

Table 3: Average Time Remaining when Flank Impacted the Round, Per Map

You can make your own observations here, but, on average, Breeze experienced the earliest flanks (1:06 remaining on the clock), and Haven experienced the latest flanks during the tournament (0:49 remaining on the clock). This actually meets expectations in the sense that Breeze is such a large, open map. A Defender could walk a short ways through “A Cave” and gain a side view of the Attackers’ spawn. Likewise, a Defender could push through “Mid Wood Doors” to “Mid Bottom” and quickly flank anyone in “A Lobby” or “B Cannon.” This is comparable to Icebox in the sense that spawnpoints are essentially small cubbies behind very long, high-visibility spaces that stretch across almost the entire width of the map with little to no corners to cut visibility.

Haven, on the other hand, holds a much higher risk to flanking. Being the only map in the game with three (3) sites, pretty much every corner and corridor will be defended. Also, because Haven’s “Mid” is actually just “B Site,” the middle of the map will likely be rigged with some sort of utilities, whether it be a Cypher tripwire, some sort of Killjoy utility, or maybe even a Sage wall. Thus, agents like Cypher and Killjoy will almost always know when an enemy is attempting to flank behind Defenders. The only real route for Attackers to take when flanking is “C Short”→“C Window,” more commonly known as “Garage,” but Cypher and Killjoy can very easily hold that and either “C Site” or “B Site” simultaneously, forcing Attackers down high-visibility routes like “C Long” and “A Long” instead. Since players are aware of this, you will often see Attackers rotating from site to site constantly, in an effort to force Defenders to leave the obvious routes undefended. Then, only near the end of the round or after the spike is planted will you begin to see flanks being attempted.

Which Rounds Produced the Greatest Number of Flanks?

Are flanks more likely to occur during the “Pistol Rounds” of a match? Maybe players attempt flanks more frequently during “Eco Rounds” or “Bonus Rounds?” It would certainly make a lot of sense for flanks to occur when there is a mismatch of credits between both teams. A Defender on a “Light Buy” may wait a while and then flank to gain an advantage on an Attacker who is on a “Full Buy.” Likewise, an Attacker on a “Light Buy” might flank and lurk for a while to catch a Defender who is on a “Full Buy.” Doing so would earn the flanking player a better weapon to use for the remainder of the round.

Figure 2: Number of Impactful Flanks that Occurred in Any Particular Round

The data in Figure 2 seems to match the theory about economical advantages and disadvantages, but timing also plays a major factor, which you will soon see. The following table is an effort to explain why certain rounds would produce more flanks than others.

Table 4: Sample Half to Explain Figure 2 Data

To explain the chronology supplied in Table 4, let us assume that the Defenders won the “Pistol” Round, “Force Buy” Round, and even managed to win their “Bonus” Round (Round 3). Following Round 3, the Defenders would have upgraded their Spectres to the Attackers’ Vandals and Phantoms, and the Attackers would be back down to only pistols. So, Round 4 would have the largest mismatch of credits of the entire game, with Vandals/Phantoms versus Pistols (which might be why Figure 2 shows such a high frequency of flank attempts in Round 4s).

Then, having lost 4 rounds in a row, the Attackers would start earning more credits in Round 5 because they are still on a losing streak. The game’s attempt to help the losing team by providing them with bonus credits each round could be why Rounds 5–8 have fewer flank attempts. Or, it is entirely possible that flank attempts are paused because a lot were already attempted in Rounds 1–4, so they may not work because enemies are wise to them.

By Round 9 or 10, you just have to assume that the Defenders are either still dominating, or the Attackers have finally caught up to maintain a couple of late-half “Full Buy”s. Then, for the likes of Rounds 11 and 12, players might begin attempting flanks again because these are the final rounds of the half, a team might be force-buying, and/or flanks have a better chance of working again since they were not attempted as much since Round 4 (so the enemy team may have let down their guard a little bit).

Unfortunately, the data cannot be mirrored for the second half of a match because there is no guarantee that rounds beyond Round 13 will even take place. In other words, the likelihood of a “Round 25” occurring in a match is very small, so there is very little data recorded for high rounds like this one.

Takeaways/Conclusions

This information is a lot to take in, but what it all boils down to is that, if you want to perform well in Valorant, all you need to do is select Jett, Reyna, Raze, and Sage, and watch your flank between 1:20 and 1:11 during Rounds 1, 2, 3, 4, 9, 11, and 12. It is that simple!

Well, not exactly. 🙃 Professional players might not be reflective of common Ranked players, but that does not mean they do not share some of the same instincts. The best Ranked players are good enough to identify the right time to flank the enemy team, so some of the data shown can at least be considered in Ranked matches. With that said, there are many different “correct” ways to play and win a round, some of which involve flanking, and others which do not involve flanking.

When it comes to Agent Selection, however, there are very few correct choices that will win rounds. Hopefully, before the next major tournament, Valorant sees some more agent rebalancing/remodeling to vary the types of viable head-to-heads in matches. When only about 50% of agents are “good enough” to be played at the professional level, matchups can become pretty repetitive. Certainly, room can be made at the highest level of Valorant to accommodate for Brimstone, Breach, KAY/0, and Yoru… What do you say, Riot Games? 😉

Like this article? If so, I encourage you to check out some of my other articles here! There are quite a few different types of studies on there that you may like!