Checking Riot's lane swap assumptions: Part 1

by theScore Staff Jul 22 2016
Thumbnail image courtesy of Riot esports Flickr

Riot recently released a statement on impending changes to the game, targeting the frequency of lane swaps. As a prolific League of Legends writer, this game and its appeal to both viewers and the teams who play it competitively is important to me. I want to assess claims made in Riot’s original statement as well as community response, not to decide whether lane swaps are valuable, but to understand why Riot have decided this change and the timing of the change is necessary and gauge whether the changes they propose will provide the solution they want.

In Riot’s original statement, they made the following opening assumption:

“Laneswapping, while difficult to do successfully, is starting to feel pretty formulaic with few strategic tradeoffs. As it’s become more prevalent and teams do it more efficiently, it’s led to passive turret trading and less direct early conflict. When laneswapping becomes a default opener, it creates a non-interactive early game with [sic]. We’ll be making some changes in the upcoming patch to address this."

The key assumptions here are:

1) Lane swapping is formulaic with few strategic tradeoffs
2) Lane swapping is the “default opener” of games
3) There are fewer early direct conflicts in lane swaps

Default start

While the first point is more nuanced and will require a deeper discussion to address, Points 2 and 3 are somewhat easy to support or undermine statistically. Eike “Timbolt” Heimpel and Florian “Bridgeburner” Dorner run a website called League of Analytics looking to deepen the metrics available to people who want to analyze League of Legends. They’ve developed a metric for lane swaps that they believe captures 95 percent of all lane swaps in professional play. One can look at this metric to get a very comfortable sense for how frequently lane swaps currently occur in professional League of Legends in four of the five major regions.

According to their metric, lane swap games only account for 48.7 percent of total games played this split so far (up to the second day of the EU LCS Week 8), and only 17 pro teams in the NA LCS, EU LCS, LMS and LCK combined lane swap in 50 percent of more of their games. I counted the lane swaps in the fifth major region, the LPL, by hand. In this case, I count any instance where the number of players on one team differs from the number of players on the other team in top and bottom lanes at three minutes. My definition of lane swaps differs from theirs, so the data is kept separate. It is assumed that using either definition will give relatively comparable results Only one team — LGD Gaming — lane swaps in more than 50 percent of their games as of the end of LPL Week 7 with a total of 61 lane swap games in 156 games played (39 percent).

This suggests that the majority of professional games actually do not have lane swaps as the “default start.” Perhaps, instead, Riot Games intended to say that lane swaps are the “ideal start,” meaning that, at the highest level of play, teams will always choose to lane swap. The less than 50 percent lane swap rates only exist because not all teams are at a high enough level to know that they should be using swaps as their default start.

Team ROCCAT love lane swaps more than any other team in the five major regions

Looking at the list of teams that lane swap in more than 50 percent of their games, however, only three are ranked in the top three of their regions: Fnatic, G2 Esports and J Team. In fact, teams near the bottom of their regional ranking feature prominently in this list. Team ROCCAT has the highest lane swap rate in any of the five major leagues of 73.3 percent, and of the 18 teams in the five major regions that lane swap more than 50 percent of the time, 10 are actually in the bottom four of their respective regions.

While it does seem like, based on this information, lane swapping is not considered the “default” or “ideal start” by most teams in professional league of legends, it does seem that teams that lane swap with a high frequency are much more common in the West. Seven of the 18 teams are European, and five are North American, — 12 of all teams in the five major leagues that lane swap more than 50 percent of the time are western teams with only three Korean, two LMS and one LPL team lane swapping more than half the time.

Looking only at Western teams, the EU LCS has a lane swap in 54.7 percent of games, and NA LCS has a lane swap in 51.9 percent of games. While these numbers are greater than 50 percent, it’s far from an overwhelming majority and not enough to refer to as a “default start.” The only region in which it could be argued that lane swapping is considered ideal is Europe, with more than 50 percent of teams lane swapping more than half the time, but even then G2 and Fnatic, two of the three top teams, only lane swap in 56.7 percent of their games. Still, given the likelihood Riot, as a western gaming company, look more at western regions, this could partly explain their assumption that lane swapping is the “default start” in competitive, though it's a slight stretch.

There are fewer early direct conflicts in lane swaps

If it turns out that lane swaps are not actually the “default start,” then it’s important to understand how they do impact strategic choices. Most have interpreted Riot's wording of "direct conflict" to mean skirmishes or kills. Some of those unhappy with the change in the community have expressed that they don’t believe that standard lanes are necessarily related to higher kill early games, as constantly moving around the map creates more opportunities for skirmishes and players getting caught out. Head Coach of top Brazilian team, INTZ, Alexander “Abaxial” Halibel, made a public statement on the League of Legends subreddit to this effect.

“Swaps create long lanes and greater risk for teams to handle. It's easier to chase people down or set up buff invades when outer turrets are down. Winning skirmishes in these situations is higher reward than standard lanes and* it's easier to avoid risks in standard lanes (as there is less pressure to make macro decisions).”

To try to better quantify the argument, I contacted the owners of the League of Analytics website to ask them for additional data on lane swap scenarios. They provided me with the timing of the first blood in and out of lane swaps for each region as well as the Combined Kills Per Minute (the total of all kills in a region divided by total minutes played in a region, inspired by in lane swap games versus non-lane swap games for each region.

Region Lane swap (Y/N) Time of first blood (minutes) CKPM (5-15 minutes)
EU LCS N 7.2 .48
EU LCS Y 7.4 .39
NA LCS N 5.9 .55
NA LCS Y 8.3 .37
LCK N 6.1 .40
LCK Y 8.7 .30
LMS N 6.7 .37
LMS Y 9.5 .29
LPL N 5.0 N/A
LPL Y 7.1 N/A

Note: CKPM is not available for LPL games.

In all regions except Europe, there's at least a two minute difference in timing of first blood in lane swap. vs non-lane swap games with first blood occurring earlier in non-lane swap games. This implies that the action tends to start earlier in non-lane swap games.

Yet, this doesn't actually imply that early games have more conflict without lane swaps, just that first blood will happen earlier. This is why CKPM are inlcuded. Even Europe, which doesn't have an appreciably different first blood timing, has roughly .09 higher CKPM between 5 and 15 minutes of the game in standard lanes games. All four regions have about .10 higher CKPM in standard lane games, which does support the idea that there are more kills or conflicts in standard lane games.

This .10 only accounts for about one extra kill (for one team) in the span of ten minutes. So the answer is, yes, lane swaps do seem to be correlated with higher kill early games and earlier first bloods. As a result, this seems to be the more likely explanation for Riot deciding to take action to reduce the frequency of lane swaps than to prevent it from being the "default start" to games. One must now ask whether one extra kill in the span of ten minutes warrants radical changes to the game this close to playoffs and World Championship qualifiers.

The final assumption regarding strategic tradeoffs requires a more detailed examination and is much less clear-cut, so I'll cover it in a second article. Based on the data available, however, we can at least conclude that decision-making that goes into whether to lane swap or not isn't as straightforward as is implied by the initial statement. If no strategic tradeoffs are made, then teams would much more uniformly choose to lane swap or not.

Either way, so far we've learned that lane swaps are not the default or even the ideal start, and they are only the majority start in western regions. Even then, games where lane swaps occur or don't are close to 50 percent. Riot stated, "Our goal is not to eliminate laneswapping but to make it a strategic choice with actual tradeoffs." Considering how split the frequency of lane swapping is, it seems like that already exists.

But if Riot's goal is increasing the action in games, the data supports that there is a positive correlation between standard lane instances and combined kills per minute. The correlation only amounts to about one additional kill from five to fifteen minutes on average per game. Additional study would have to be done to determine causation.

Thank you to League of Analytics' Timbolt for help with Part 1. To be continued in Part 2.

Kelsey Moser is a staff writer for theScore esports. You can follow her on Twitter.