This article was originally published on GameSpot's sister site onGamers.com, which was dedicated to esports coverage.

　　Note: This article does not include this past weekend's data.

　　

An In-Depth Look at Parity in the LCS

One of Riot’s assumed goals with the LCS is to make the leagues the highest level of competition in NA and EU. The structure of the leagues limits the number of teams, and the relegation system makes the bottom league spots up for grabs after every split, which in theory ensures that each region has its best eight teams or close to its best eight teams in the league.

　　Yet, even having a league with the “best” eight teams within a region does not ensure that the league will be competitive overall, as the pool of talent doesn’t always match this arbitrary number. Recently, people have speculated the player talent pools in the LCS have improved from an influx of new players as well as importing talent from other regions. Has this made the league championship series closer and more competitive?

　　

How Has the Competitiveness within the LCS Changed over Time?

The first topic we should tackle is calculating how “close” the LCS has become since its inception. The first way we will calculate the closeness of a split is by looking at the standard deviation of team win totals. This measures the dispersion of wins teams have at the end of a split. Meaning if every team went 14-14, the standard deviation would be 0 (an extremely close league), on the other hand if a few teams had over 20 wins, the standard deviation would be very high, as the league isn’t very close. The following graphic shows the standard deviations for every split for the NA and EU LCS:

　　As you can see from the graph, the leagues started out pretty divided as the LCS pools included a few teams that weren’t cut out for the LCS, such as the DragonBorns or Velocity. As time continued, EU got closer pretty quickly in summer 2013, where MYM was the only team below 13 wins, and Lemondogs the only team over 15 wins. This closeness continued in EU while NA still had division between teams like C9 and TSM compared to teams like Velocity and XDG. EU however has historically been the closer league, while NA only this split has started to become a neck and neck race.

　　The stats for the first three splits include 112 games, whereas the Summer 2014 stats include only 80 games. To account for this difference, we looked at the state of each league after 80 games for the current split and for the last split. These results better indicate the current state of each league relative to its state last split. Thru 80 games, the standard deviation for NA Spring was 4.24 vs. 2.78 for NA Summer, which suggests that the league is more competitive. Likewise, the standard deviation for EU Spring was 1.66 vs. 2.92 for EU Summer, which suggests that the league is less competitive.

　　Most viewers would most likely agree that NA is more competitive this split even without seeing the statistics. The standard deviation for the EU LCS win totals in the first graph seems on par with the three other values, yet the thru-80-games comparison above shows that the EU LCS is perhaps less competitive this split (as the standard deviation is higher). Seeing that the EU LCS appears to have gotten less competitive is not surprising given the departure of several key players from the league from last split, such as Alex Ich and Amazing, among other reasons.

　　

A Look at Parity vs. Inconsistency

While we have proved that the leagues are in total the closest they have ever been, does that mean they are more competitive? Does the closeness of the league come from parity (every team being equal skill allowing any team to beat any team at any time)? Or does it come from inconsistency (teams have become less consistent meaning that better teams can drop games to lower skill teams due to lack of execution)? Boiled down, the big question is: Is the increasing closeness of the league due to parity or inconsistency?

　　In order to investigate this we had to get creative and think about how competitiveness and parity are defined. We came up with a testing method based on seeing if the favored team won more often, based on three gradients. If they were slightly favored, favored, or significantly favored, with these definitions being based on the team’s overall win rate. We would compare, at the time of the match, the team’s win rate to determine who was favored to win this game, and by how much. As an example lets look at the Evil Geniuses vs. CLG this past weekend. CLG was 13-7 (65% win rate) to EG’s 5-15 (25% win rate), making the win rate difference be 40% (65%-25%), leading to a “favored” match for CLG.

　　With this process in mind we mapped out what ideally the distributions for a league with parity, compared to what a league with inconsistency would look like.

　　Parity means that even teams that are favored would lose to the teams that were close to them in the standings, as these teams would be relatively even in skill and games could go either way. Meanwhile, when a team has a significant advantage such as 40% or > 50%, they should win less and less often as in a competitive league the better teams rise to the top while the bottom teams are simply less skilled. However, in the inconsistency example shown in the table above, the unpredictable nature of teams would mean that even when teams are significantly favored and have better records overall, any team could win, leading to a high upset rate and making the league “unpredictable.”

　　

The Results of the 2014 Season

　　Now that we know what idealized league parity would look like, let’s see how the leagues actually size up (As a note, none of these results include week 1 performances because there had not been enough games yet to establish who was truly favored).

　　

North America

The first thing to jump out when looking at the North American splits is how many more upsets there have been this split compared to last split, but this doesn't solve our question of parity vs. inconsistency, instead we have to look at how these upsets were distributed. In the summer, teams who have slightly worse records actually win more games, as the upset rate of games between 0 to 25% win rate advantage is 62%.

　　Is it parity or is it inconsistency?

　　The anti-climatic answer is that it appears to be both. As we saw in our idealized parity and inconsistency systems, the upset rate should go down as the win rate differential goes up, which it does somewhat but there are more major upsets than there should be, and overall a lot of upsets in general leading to a feeling of inconsistency, but still that the skill level has improved to a level of parity between top teams.

　　

Europe

Now looking at the European trends we see that last split there were way more upsets than there were in NA, which goes with the standard deviation graph we showed earlier. Honestly last split for EU looks almost identical to the idealized inconsistency graph, if just one of the major upsets had gone the other way each section would be at almost 50% across all differential groups.

　　On the other hand the summer split seems to bring a bit more consistency to the table, as upset totals are much lower (from 31 upsets in spring to 22 upsets in summer) and an upset rate that trends down as win rate differential goes up.

　　Is it parity or inconsistency?

　　EU seems to have gained competitiveness although there are still some inconsistent teams such as Millennium, who lose more games than they win when favored.

　　

Team-by-Team Performance

The final thing we found when looking into this topic was analyzing how consistent teams were, which we found some interesting data. In order to break down how teams performed, we separated their games into three categories based on who they were facing. These three categories were if they were favored (had a better record than the opposing team), if they were equal (had the same record as the opposing team), and if they were underdogs (had worse record than opposing teams). It is interesting to see as some teams were best when underdogs and faltered when favored (CLG), while other teams were the opposite (TSM). Here is the NA breakdown:

　　Interesting Takeaways:

　　Curse has never won a game when favoredCLG was 5-0 when underdog or tied until this weekend (when CLG lost to LMQ)C9 has won 5 of 6 games when an underdog, but not that many when favored and none when facing an equal record opponent.Dignitas is the most inconsistent team, having a 50% win rate in all 3 categories, meaning they beat the good teams but lose to the bad teams equally.DIG and CLG have entered the most games with a better record than their opponentsTSM has won 6 of 7 games when favored, the best win rate in the league in that categoryThe following graphic shows the EU results:

　　Interesting Takeaways:

　　Roccat is undefeated (2 for 2) when favoredMillenium is the most inconsistent team losing a majority of their favored games, a close to 50% win rate in each category.ALL has only the 5th highest win rate when entering a game with a higher win rate than the opponentSK is 0-4 when underdog or when tied, they only ever win games when they are favored over their opponent.Overall EU is more balanced than NA, less upsets means more teams win when they are supposed to with only few exceptions like FNC and MIL who win more games when they are underdogs.

Conclusion

Overall the LCS has been growing closer gradually, and this can be in part attributed to the growing talent pool and increased competitiveness, the LCS has also grown a bit more inconsistent especially in NA where we see teams like C9 and CLG lose to the bottom 3 teams while consistently beat the top 5 teams. This will make each region's playoffs extra exciting given the huge potential for upsets in NA and the pressure on the top teams in EU to win when favored.