Tuesday, June 18, 2013

Detailed Analysis of LCS Superweek with Statistics!

In my last article titled “6 Things LCS can Improve on,” two of the biggest things I wanted to see from the LCS is more statistics and analysis of the teams. For the Superweek I parsed a ton of the data from the 20 games and they have lead to some interesting insight into the current meta of the NA scene. This article will break down the different team play styles, snowballing, and some bonus quick stats, EU vs. NA and champion analysis!


Out of all the teams performances in this weeks LCS, three teams, TSM, Dig, and C9, stood out the most in their performance and statistics. Breaking down things like FB (first blood), objective control, and total kills we get exciting insight into how their actions show their priorities as a team and create unique play styles for each team. Analyzing these statistics actually gave a lot of insight not just into the meta, but also into the teams specifically as we see different play styles and priorities coming from each team. I am going to highlight a few of their picks and play styles that really stood out to me.


The first interesting team to look at is the “aggressive” TSM. People always hype TSM’s aggressiveness and their playmaker Reginald, but ever since late last split TSM has been singing a different tune on their aggressive plays, focusing objectives rather than kills. Their matches had the slowest first bloods of any games, coming in at an average of 9:05 (compared to the week’s average of 5:40). They prioritize kills the least and they have even said in interviews that they don’t go after kills unless it will provide objectives. This is clearly seen as they have given up FB 4 out of their 5 games. Despite their games averaging the slowest FBs, they have the fastest objectives in the league. TSM took dragon all 5 games for themselves and averaged at the quickest pace (8 min and 25 seconds, compared to league average of 9:55). They also force slightly faster first towers at the pace of 6:34, compared to the average 6:54. This prioritization of objectives over kills shows in their picks as well as they often choose Shen or jungle Elise. Oddone’s Elise is a priority pick for TSM not just because he hit challenger with it, but because Elise is one of the best early dragon takers due to her spiders acting as the tank which is essential to mitigate the high amount of damage that Dragon does early game. Out of the 4 quickest dragons during the week, TSM got 3 of them - all with Elise jungle. TSM’s objective focused play style can also be seen in their kill/death stats, having the lowest team kills (54, 2nd lowest is VES at 57) in the league but also the lowest team deaths (45, 2nd lowest is VUL at 57)). While people have been saying “TSM is getting back to their roots from s2” I think this is untrue, as they played a tanky-dps team fight oriented style back then and their style now is more close to the Season 2 CLG style of “objectives over teamfights”.


On the other side of the coin, we look at Dignitas. Dignitas actually has the highest amount of kills per minute, at 2.43 (the average being 1.95). They also average the fastest FB time of any team, coming in at 4:23 compared to the league average at 5:40. However, at the same time they have by FAR the slowest tower and Dragon times out of any other team, averaging 11:53 Dragon time (vs. league average 9:55), and 8:53 average tower time (vs. league average 6:54). These slow times come from Dig’s tendency to 2v2 over 2v1. In the 20 games of Superweek there were only three 2v2 matchups in NA, every other game was 2v1 mid/top vs 2v1 bot. All three of these 2v2s were forced by Dignitas. Many may attribute this to “slow adoption of the meta” and think that this contributed to Dig’s poor performance this week yet on the contrary in all three of these 2v2 games Dignitas ‘won’ the early game and came out with a gold lead at 10 minutes. Both of Dig’s wins were in these 3 games and the 3rd was the CLG/Dig game which... we all knew what a mess that was. I believe that Dig’s 2v2 abilities will be very important going forward as I believe the 3.8 patch will bring forth more 2v2 style in the NA LCS.


The last team I am going to look at is the very hyped Cloud 9. Going into this super week I strongly felt that they were going to do well yet their play still surprised me significantly - not just due to the resulting 5-0, but in HOW they got that 5-0. Many people attribute their success to ‘replicating the korean scene’ but I believe that extremely cheapens their accomplishments. What Cloud 9 showed in this super week was not strategic brilliance, or extreme mechanical skill (like we often see from ‘new talent’ teams), but a very poised and flexible team who has great decision making abilities. Cloud 9 does not get the fastest towers or push the most out of any team nor do they get an early advantage every game. They prioritize objectives, but there is nothing exceptional about it. They have average FB timers (5:36), a bit slower than average dragon timer (10:57), a bit faster tower timer (6:20), and they only got out to an early lead 2 out of 5 games. This is what makes Cloud 9 the scariest team in the league. They play like an experienced team despite this being their first week in LCS, they don’t get flustered when behind, they don’t throw games that they have the advantage in - they just play solid with excellent teamfighting skills (as seen by them having the 2nd highest assist per kill average, only beaten by CLG’s inflated stats due to their long games and thus having more teamfights than the average team).


The next big topic is one that comes up a lot in interviews with players: the issue of snowballing. Often times a lane getting first blood is attributed to it’s success, or when they gave up that dragon it doomed them - but how strong is the correlation between these things? I broke down the “snowball” effect on 6 possible advantages and looked at their correlation with winning. These six factors were first blood, first tower, first dragon, first baron, gold advantage at 10 minutes, and gold advantage at 20 minutes.
  • First blood was the most neutral statistic with the winning team scoring first blood in 50% of the games, therefore it seems to have no impact on the winner. Also, first blood didn’t have much of a pattern to it, with the average time being 5 min 40 sec and almost equally in all 3 lanes (6 top, 5 mid, 6 bot, 3 drag).
  • The next statistic, first tower, was also fairly neutral with 55% of the teams scoring first tower going on to win the game. The first tower fell on average at 6:54, was way more often top and bot than mid (6 top, 3 mid, 11 bot), and was favored to the blue side (13 purple tower deaths, 7 blue deaths).  
  • The most bizarre statistic goes to first dragon (average time 9:55), which had a 40% win rate. More teams who got the first dragon went on to lose the game than win the game. This statistic really confused me at first but I think this can be attributed to the successful counter-play that has come from dragon fights, as often times teams will trade towers or kills for this dragon, and it often puts the team that initially started dragon into a vulnerable position to be ganked.
  • The last neutral objective statistic - first Baron - leads to the highest win rate as expected, yet still a bit lower than predicted, at 66.6%.  The average time for the first Baron is 28:32. Baron is becoming less of a snowball instrument for teams as pushing has gained increased priority. Baron has become more of a common comeback attempt, as 1/3rd of the successful barons were gotten by teams who at the time had a gold disadvantage. This is often due to the fact that after winning a teamfight as a losing team you don’t have the map pressure to take towers, and so instead they will take baron to give a few minute buffer to try and make a stronger comeback and deny objectives from the other team.
  • The next snowball factor to look at is gold lead at 10 minutes. This was the most definitive early factor with 64.7% of teams with a gold lead at 10 minutes going on to win the game. Of these advantages, the leading team averaged 13.5k gold while the trailing team averaged 12.2k, which is about a 10% lead. The largest lead was by team CST over team VUL on the first day of Superweek, coming in at 14.5k gold vs 11.8k gold with CST winning the game. While many people would think that blue side would tend to lead early game due to the double golem advantages, it turns out purple actually gets out to an early lead more often at a rate of 76.47%!
  • The last factor tracked was the gold advantage at 20 min. This is the biggest predictor of win likeliness, with 70% of teams at a 20 minute gold lead going on to win their games.


Here we’ll take a look at one of the most accessible statistics - champion picks. <Here> is a link to the basic stats of the champions picked in LCS. There are some crazy things emerging here. For example: Kha’Zix’s mega OP patch where he is banned or picked in every game and is undefeated (5-0) when picked (this mostly due to the fact that in patch 3.7 all of his competition like Zed and TF were nerfed yet Kha’Zix remained untouched). However, the biggest topic I am going to be analysing is what I see as the “break-out” picks of this Superweek - champions rising in popularity in their respective roles, and find what to look for in the future week's picks.  
  • Top: Kennen. This was one of the biggest breakouts of this Superweek, being picked/banned 19 times. Previously, in the Summer Promotion series and the Spring Playoffs, he was only picked 5 times in 44 games (with no patch changing him since then). What sparked his popularity was the rise of Kennen in Korean matches. Kennen is often paired with an aggressive initiator like Zac or tanky AP counterpart like Ryze who both focus on boosting his teamfighting ability. Although he has a weak win rate at 44.4%, I expect to see more of him in the coming weeks especially from teams like C9, VES, and VUL.
  • Jungle: Zac and Elise. These two junglers dominated the Superweek. With Elise and Zac being picked 12 times and 10 times, respectively, they were a very common sight (even going head to head 6 games!). Even though both of these champions can play multiple roles, they were chosen for jungle most of the time. Elise showed her excellence at jungle control and ganking while Zac excelled at diving and teamfight initiations. I think Zac really showed his strength this week and will be pick/banned even more next week as more teams adopt him.
  • Middle: Kha’Zix.  While Kha’Zix was always a popular pick, he really showed his dominance this week by being the only champion picked or banned in every game and with 100% win rate with more than 3 games played. However, I think the most interesting thing in this week with mid lane is the variety shown. Mid lane is the thing left most up to the personality of the player, as we’ve seen things like Zed from C9 and CST, Kayle from VES, Ori from TSM and DIG, and even Xerath from VUL. Also, the rise of 2v1 mid lanes has often left “mid” players to the side lanes for 2v1ing.
  • ADC: Draven. Draven has made a big appearance in this week’s LCS, spearheaded by the new blood of ADCs; TSM’s WildTurtle, C9’s Sneaky, and VES’s Maplestreet. Draven leads his team to an early gold lead of at least 500g at the 10 minute mark 75% of the time (6/8 games, 1 tie and 1 fail). With the 3.8 patch and the nerf of Spirit of the Elder Lizard I can see a fall of Ezreal and a rise in the League of Draven!
  • Support: Thresh. More than just EDward aka the “Thresh Prince” has been choosing this champion lately. It seems that TSM’s Xpecial, C9’s Lemonnation, CST’s Daydreamin and VUL’s Bloodwater have had success with him leading to a 95% pick/ban rate, and being tied with the most picked champion at 12 games. His immense playmaking abilities as well as his proficiency at 2v1ing (which happened in 85% of the games) has really brought him front and center in the NA scene. A shoutout to Nami as well, who despite her 0-3 performance this opening week will get more and more play as the weeks progress.

EU vs. NA

While this is more focused on analyzing the NA games in depth, I think one of the most interesting aspects that are easy to highlight in statistics are the differences between the NA and EU scenes.

The most apparent differences are in champion picks. <Here> is the excel data for the EU picks. There are a bunch of differences between these EU picks and NA picks. In NA there are some very popular picks which EU teams don’t value very much. For example Kennen who had a 95% inclusion rate in NA only had 35% pick/ban rate in EU, or Elise who had a 95% rate in NA but only 35% in EU as well. Kha’Zix was a must-pick in NA having 100% pick/ban rate and 100% win rate, while in EU he was only banned or picked 75% of the time. There were a few other popular NA picks like Draven, Ryze, Zac, and Karthus that did not get near the same amount of attention in EU. Similarly there were several champions in EU that were not popular in NA. Nunu was one of the big ones, with 65% pick/ban rate in EU while only having a 10% pick/ban rate in NA. Also Shen, who was selected 18 times in EU but only 5 times in NA (mostly from TSM). The other EU picks that were popular are Nami, TF, Malphite, Lissandra, and Varus.

The next crazy difference in EU and NA play was how they set up their lanes. In NA they have a heavy 2v1 focus, with the lanes ending up in a 2v1 matchup 85% of the time. However in EU it’s a different story, with about 40% of the games being 2v2. While on blue side, NA sent their dual lane bot 17 times, mid once, and top twice. EU moved their dual lane more when on blue side, only going bot 14 times, mid 2 times, and top 4 times. The biggest difference was in how they chose their dual lane position as purple side. NA sent their dual lane top 7 times, mid 10 times, and bot 3 times. EU sent their dual lane top 9 times, mid only 2 times, and bot 9 times. NA really favors the 2v1 mid as they had 2v1s mid in 50% of the games, while EU really doesnt like mid 2v1, only sending them mid in a few games. I think this has to deal with the EU mid laners being the “star players” of most of their teams (xpeke, Alex Ich, Bjergsen, Ocelote,  Froggen, etc.) and they believe they can or should 1v1.

Statistic Quick Fires (Stats are from NA Superweek)

  • Average time for an LCS game is 38:20, for a CLG game it is 52:46.
  • If game time had a normal distribution , the probability of the 71 minute Dig vs. CLG game is 0.2% (1 in 335 games). [mean = 38.345, std = 11.95]
  • Team Velocity has both the lowest kills per minute of any team (1.617), and the highest deaths per minute (2.838).
  • At 10 minutes the gold lead went to the Purple team 76.4% of the time.
  • The Blue team wins 60% of the games.
  • The team with the lead in gold at 10 minutes continues to lead at 20 minutes 76.4% of the time.
  • TSM & CLG gave up FB in 4 out of 5 of their games, while CST got FB in 4 out of 5 of their games.
  • While NA 2v1’s in 85% of games, EU only 2v1’s in 60% of games.
  • 47.7% of all champs were picked or banned in the first week

Thank you for reading. If you want to see the full statistics of all the data I used for these numbers, go HERE.
I also wanted to give a big thanks to @jjordizzle for his help on editing and feedback and some help from @zeroaurora_hf and @zerglinator for help on the EU stats.

Sunday, June 9, 2013

Support Pick & Counter Guide s3 Update w/ Unorthodox supports

Alright, I typically do not do an update to this chart so soon after the last one, but I think I made a few mistakes in the last one that needed to be corrected. The biggest thing was that because of the increased number of viable supports and the addition of champions like Thresh and Nami, the format started to break a little bit as there became less and less overlap due to the higher increased number of possible counters. The whole point of this chart is for overlap so this was going against the goals I had in mind in creation. To fix these goals, and to fix the "where is zilean on the chart!?!" requests, I added a new section titled Unorthodox Supports. These have the typical counterpicks that I've asked people who specialize in this champion as well as what I've seen in my experience. This should keep overlap in the actual chart without leaving out some supports which people feel so endeared with. 

As always, let me know what you think and you can follow me (or let me know) via facebook or twitter. And you can also check out an android phone app I made with some cool people: "Spellsy's Support Cards".

For a blog/facebook-special I have also released an alpha version of the description of each countermatchup in the chart from above, it is pre-release so it is not the best format, but here it is: https://docs.google.com/spreadsheet/pub?key=0AllLJAxUt7qcdEhWbnVfTEtZM0otdDd2R2pkTUJ1bGc&single=true&gid=0&output=html

Also, here is the new desktop version <coming soon>.

Saturday, June 1, 2013

6 Things LCS can improve on.

I think the Esports staff has created a great hit with the LCS last split, it brought one thing to the western esports scene that hasn't really ever been achieved before, structure and production quality. It has been a great time watching with friends, hyping the matches, and all that fun stuff. However, while there were gross advancements in structure and production, there are still some things I think can be improved especially in the analytical aspects and some fan-features.

1. Statistics

The first thing is something that is going to be the biggest point for me I guess, statistics. I think that nerds everywhere whether they are traditional sports nerds, gamers, or anything always love analytics and statistics. Ask anyone why baseball is so popular in the US! And in LoL there are very many possibilities for very interesting statistics, whether in banner form during matches, talked about by (fed to) the casters, or even in pre-match stuff and ESPECIALLY pick phase. Due to the very structured advancements in LCS this I think can amplify the power and accuracy of the statistics. The riot esports crew is already pushing some statistics but I think that there is a lot of room to grow. Here are some examples of possible interesting statistics:

  • Basic Gold-Winning relationships. 
    • People who have a 50% gold lead have 95% win rate
    • People who get the first dragon win 75% of the time 
    • When teams take dragon before the 5 min (baron before 18 min) they have 85% win rate
  • Team related strategies
    • When TSM first picks Nasus they go Renekton top lane 80% of the time
    • When CLG 2v1s they have a 70% chance to get out to an early gold lead (10 min).
    • Vulcun has a 80% win rate with the protect-the-kog strat (60% win rate with AoE comp!)
  • Item Builds
    • Vlads 87% of the time get will of the ancients first i tem.
    • Wildturtle averages the fastest bt rush in the league, averaging at 9 min on draven. 
  •  Graphs
    • Analyzing things like hot picks hot bans like that one week rammus was popular or the downward trend of someone like diana (idk of diana truly has downward trend)
    • Gold advantage map over time, how did this team gain their lead and when did it happen, did they continue to build the lead or has it remained stagnant?
I think statistics are a powerful way to analyze the game quantifiably in a measure other than KDA or gold, which I think should be pushed out a little bit cause as we know KDA =/= skill/execution (crs cop as example, most kda but not many votes from pros for allstars)

2. Voice over replays ("Listen Ins")

This is another personal favorite of mine, I think if there are replays of teamfights and such we should hear either team's perspective (sometimes losing team can be more fun/interesting). I know there is the issue of cursing and such but I believe that can be worked out as it has been implemented in the past effectively (I know we did it at ipl all-stars, albeit a less stressful environment). 

One of my favorite tournaments of all time has been the MLG Summer Arena, this is not because the production value or any standout matches, but at that tournament MGL offered us multiple streams to the point where we could actually choose to listen to one team for the entire match. I listened to TSM and learned a lot about them and about the game, while I'm not asking the LCS to do this as it would be too invasive strategically, I think it would be fun to hear listen-in moments like we have seen in past tournaments more often in the LCS. 

3. Post Game Content (on lolesports.com) 

I think Riot has a powerful team of smart individuals and talented artists and could create amazing post game content, whether it be infographics or just tabled information I think there are quite a few things they could do to not just analyze the game in an interesting way, but also TEACH players and help players learn - one important aspect of esports.
  • Runes, Masteries, Skill order information on each game played. This can be done similar to probuilds.net (or even partner with them if you wanted and use their parser). This can even be built on to see most common runes/masteries for specific champions or from specific players etc.. 
  • More analytics in the post game summary like heat maps on when there were kills, towers, and overall objectives. Also these can be analyzed over tons of games to see when the first kill is most of the time and all other fun stuff like that. Also related to this is IN GAME MAP heatmaps, where people die first where people move around and other possible things. I know a DotA replay parser did this and it was very very interesting. 
  • Fan-favorite post game stuff. They have all this fan voting and who will win and such they could release those statistics on their website and be like tsm gets 50% more votes than anyone else! Or TSM games are watched by 20% more people. (This one is more optional, I understand there are business things with not wanting to release those kinda things but this is something that is interesting to me as a fan of esports and wanting to track its growth)
  • Replays (IN GAME REPLAYS - even lolreplay if necessary) downloads of the games. I know they store the games somehow in some replay system for backup uses, if its not an in house replay system on the tourney client then I think they should release the lolreplay file to download if the people want to watch, there are many times where I just want to watch SONA and see how she plays in the normal trade lane and thats hard when the camera is jumping around trying to catch the action on all ends. 
  • With these replays you could release champion sorted replays and other champion sorted information for learning. Someone could be like "I want to learn Diana mid!" so they type in diana see a bunch of replays, builds, runes, masteries and have all these learning tools from the highest competitive level at their fingertips. As someone who spends a lot of their freetime thinking about how to best teach league of legends, I think this would be an amazing resource. 
  • More detailed team pages. When you go to a team page now on lolesports you just kinda see their names, some surface statistics like KDA and gold and win rate, and a little blurb about them. I think this can be greatly expanded to favorite champs, champs most banned against them, favorite strategies (AoE, Poke, Push, etc), leadership status, other team members like managers, just overall more detailed and intricate. 

4. Game oriented player content

One of the things I think that has stood out from the repeat viewers of the LCS is the nice in between game content that has been made, the screen interviews and human stories like Dyrus' dad, and many others. And with the pop up of Machinima's VS series this aspects is being fleshed out a lot and is very interesting. However, the aspect I want to see more and I believe Riot has the power to do this is more game-oriented content. Here are some examples of things they could do:
  • Pro talks their favorite champ, why they like him, what they build what masteries / runes they run what is their favorite champ to pick it against (they do this a bit in their "Pro Picks" series, but more technical).
  • Building in depth for a role, talk to the many conditional situations and a lot of the thought that goes into item builds, this is something I see often undervalued by the non-pro scene, they don't respect how much experience it takes to finally build right as a mid laner or something. 
  • Cross section analyses of a champion, you can pull off 4 top laner pros and ask them about 1 champion that they all commonly pick and see which opinions they share, which contrast, little nuances in how each person plays the champ differently, all these interesting little details. 

5. Fantasy League

With more details and statistics and such a fantasy league would also be a great addition. This is an idea that has been around esports for a long time and there have been people who have tried to create LoL related fantasy leagues on reddit and such for several months but a riot-sponsored one on their website would be a huge hit I think. Starcraft has a great fantasy league (or at least my friends who are starcraft fans enjoy it) on TeamLiquid.net, I can see a League one being a hit as well. 

6. League Esports App

Esports has been growing and not just as a thing to enjoy alone on your computer. Many people have friends who they talk about it with and get in arguments over who has the highest kda at a lunch table or at a watching party, here is when the app would be great to boot it up and prove who is right :D. Also there would be great features for people like schedules, setting up reminders, and would tie in GREAT with fantasy league if that idea goes through as well. And of course the streams too for those who are afk. League is becoming a very social game and with that I think accessibility and mobility would be features very important. 


I think LCS has been a great step forward for esports and the League scene, especially in the western esports scenes. And as some nerd esports lover I've been very impressed with the smoothness and just great implementation of something this size. These are just things that I have experienced in other situations (whether other esports or reg. sports) that I believe would work very well in LoL and would foster even more crazy growth.