The Throws of Passion – Tackling misconceptions about throw rates in competitive LoL

The League of Legends world championship is rapidly approaching, and with it will arrive some much needed international competition. A common topic of discussion is how Western teams will perform against Asian teams. While no one in their right mind is willing to suggest that any Western team would be the favourite in a matchup with any top three team from LPL or LCK, the absolutism with which the superiority of Asian teams is expressed sometimes oversteps the mark. In particular an oft repeated maxim is the following,

“Western teams need to win early game against Asian teams, since there is zero chance of a comeback against an Asian team with a gold lead”

To look into the validity of this claim data obtained from www.leagueofgraphs.com will be presented. In particular the the win rate of teams in the league against the gold differential at certain timing breakpoints.

LCK

EU LCS

NA LCS

 The first thing to consider is how to define what a throw actually is. For this article, it is defined as when a team loses a game after having an overwhelming (+5000) gold lead. Unfortunately the statistic of peak gold differential is not easily available. Instead the gold differential at the end, and at ten and twenty minutes are available. The ending gold differential will naturally be lower than the peak differential, since the losing team has to kill champions, take towers or both in order to win the game. Therefore the accuracy of the game ending gold differential is not ideal for determining throws. However in this case LCK actually boasts the largest negative gold differentials for winning teams. Whether this is a testament to the incredible Korean ability to play from behind, or their inability to close out an overwhelming lead is debateable.

The gold differential at ten minutes also has its problems, since generally a lot of extra gold is earned between the ten to twenty minute mark, through turret taking or a large team fight, it means that a lead at this point does not necessarily translate to an overwhelming advantage. That said, a huge gold differential at ten minutes (+2000) should definitely be counted as a throw if it leads to an eventual defeat.

Of the three available statistics, gold differential at twenty minutes is probably the most useful. Whereas the gold at ten minutes typically revolves around a team’s laning strategy, maybe with some skirmishing thrown in for good measure, the gold at twenty minutes is more reflective of the team composition’s overall strengths and weaknesses. Typically by 20 minutes at least one serious teamfight has occurred, allowing teamfight comps to show their strength. Teams that make pushing down towers a priority have probably had a fair crack at most if not all outer turrets and maybe some inner ones by this point. In short, we should have a better idea how the game is going to play out at twenty minutes than at ten, purely by virtue of more information being available, that makes it all the more surprising when large leads at this point get overturned. Finally I would like to mention that the gold differential at thirty minutes would probably be the perfect statistic for this analysis, since typically at this point, in the games which we refer to as throws, the eventual loser should have a huge lead at this point. However, if the world gives you lemons, you talk to the internet about win rates as a function of gold differential at twenty minutes.

Bearing this in mind, the differences in statistics between the two LCS regions and Korea seem quite minimal. The data from the LCK looks smoother, but this is probably due to the larger number of games played there (432 compared to 182 in each LCS region). Just looking at GD@20, all three graphs show a linear trend in win rate starting at 0% for a -5000 GD@20 passing 50% for 0 GD@20 and ending with 100% for +5000 GD@20. This means that on average, the team that was losing in gold at twenty minutes went on to win the game 25% of the time. That statement is as equally valid in LCS as it is in LCK.

This data suggests that Korean teams are quite capable of throwing games, in fact they are equally as capable as LCS teams. Therefore I propose a modification for the quote written above,

“Western teams need to win early game against Asian teams, since there is only a 25% chance of a comeback against an Asian team with a gold lead at twenty minutes”

Disclaimer:

• LPL data not available.
• This is the average across all teams, it could be argued that the top LCK teams throw less often than top LCS teams, it would be interesting to see the data on that.
• This data does not suggest that it is likely for a team to throw, A team with a gold lead at twenty minutes still wins in 75% of cases.
• This data does not suggest that LCS teams are even in strength with LCK teams, everyone expects LCK teams to perform better at the WC than LCS teams in general.