Disclaimer: The following article is a compilation of self generated statistics. It's just an estimate and not a flawless one. Furthermore, it does include Scan as foreigner. If you feel the urge to place criticism, please do so after fully reading the text. By the way, this post is not related to Teamliquid, Broodwar.de, ICCup, Liquipedia or any platform, but my own doing. This post might or might not contain porn.
In the past weeks we have discussed whether or not Scan should be considered a foreigner or not. Personally, I found the discussion to be absurd - he's most definitely Korean and most definitely a member of the foreign community. Consequently, the subsequent discussions missed the point, and the underlying problems weren't understood by most, as semantics overtowered the core issue: Can Scan be beaten by the ordinary player? Probably not. However, how do you deliver "proof"? You can't, nor can this article. Yet, a ranking based on the Elo number might help to quantify his skill.
It doesn't end here. In the discussions players weren't tired to stress out that they're not in for the money. I believe that. What's in there then? Maybe competitive drive. Maybe the money isn't what really helps you to get your ass up and train some more. A somewhat regularly updated ranking might give you something in return for your effort, and even if it's just a small token of respect - it's at least something.
the basic idea: high Elo numbers demand celebration and increase fandom
Furthermore, numbers, even if not entirely objective, help writers and casters. We can use them to tell the audience how good someone is, even if the person hasn't won anything notable yet. It helps to justify posts advertisings streams of up and coming players. It helps showing trends... I think you get the idea.
Before there are numbers, you should first try to understand what Elo is, or why I chose it over any other rating. There's the simple reason that it's easier to calculate, interprete and archive than Glicko2...
... but that doesn't really matter. Elo rankings can be understood as - this is a very vague explanation - an estimator of skill. Skill can't be observed in nature, hence you have to calculate it artificially. This is done by simply assigning a start value to a player (1000) and giving or taking points depending on games won and lost. The higher the number, the better the player. How many points are assigned per won/lost match depends on the rankings of both players. It's rather difficult and the formula has more letters and symbols than the ordinary person likes. Anyhow:
magic goes in and a number comes out
Edit: Due to questions in the thread: Go and read more about that on Wikipedia if you're interested in the maths. The K Factors are taken from B.net)
The most important part is that the ranking only works if any player has a decent amount of games on his record. With less than 20 games the ranking is anything but accurate, with more it gets better, but only works once there are at least 30 games. My calculations showed that for our small scene the fluctuation of the rankings is rather high - if players have less than fifty games.
The biggest advantage is that you can indirectly read something out of the ranking. E.g. a player with 1700 points matched against one with 1500 will defeat a player with 3 out of 4 times. To be exact, the chances of winning with having 200 points more is 76%.
However, there are downsides to the ranking. For one, there's a thing called point inflation. This means that with more players entering, more and higher points will be distributed. As a result very good players in the beginning will score lower or equal than "above average" players a few thousand games later. This problem increases as we experienced periods in which good players left the scene and thus their rankings stagnated.
As the title suggests, I calculated the Elo ranks of all foreigners in between January 2012 and today. Events included are:
- Defiler Tournaments, including the Super Tour
- TLS 1 - 3 , TLC 1 - 4
- Defiler Mini Tours, Defiler Gaz Tours
- Gambit Cup, Nation War, SBWI Teamleague, some ICCup CL
- small events (eon-, draw-, alc-, heroes of the storm-, hearthstone beta-, ... ,-tour)
The criteria were something like: It was open to everyone, hence any national event (Russian LANs, netwars cups, German KOTH) weren't included. Also, very badly archived tournaments are only partially analyzed - e.g. several Canadian-driven events, several of the bw.de tours and several ICCup grids and MCA tours. Sadly, I couldn't restore everything.
Last but not least, especially Defiler tournaments gave me a headache. There were plenty of smurfs, some of which I couldn't identify. However, since we have a really, really large sample size (about 20k games) it shouldn't skew the estimator too much. That leaves one thing: Defiler featured players like Terror[fOu] (pro), Neagle (hacker), Sunday (hacker), iiN.Jaedong (hacker), FengZi (Chinese pro) or IOPS'Sense (pro). Thanks to the power of hindsight, I added weigths to them, e.g. Sense started with a 1500 point lead, rather than the ordinary 1000 points.
Ok, now for the really last bit of information: Notorious cheaters and hackers have been removed from the finalized rankings, regardless of when the last incident happened; however, their ranking is "invisible" in my data base. This includes map hackers and other people without interest (e.g. aforementioned professionals).
For now, you can read the numbers. However, I recommend you read the article to the end, or you'll bound to critize some statements of the statistics, which will make you sound like a moron. Because they'll be explained...
As you can see in the two pictures above, the rankings changed over time. I already mentioned it, the years 2014 and 2015 didn't feature nearly as many high class events than the previous years did. As a consequence a lot of players should have their status "active" removed. Others didn't participate in the toughest environments - but more on that soon.
It's hard to tell which tournament was the last "serious" one before the TLCs started again. In between the third Legacy Starleague finals, the ICCup Ladder Main Event finale and the Defiler Super Tournament a great amount of players vanished - while Scan re-appeared and started to dominate the scene. At first he dropped a couple of games while off-racing, others probably due to sickness (Defiler Super Tour), but then there was no more chance. Right now he leads the ranking of the "active" players by about 300 points. In other words, if the numbers are taken super serious, he will win 81 out 100 against trutaCz. This includes the assumption that he might switch races. However, again, if you take the number super seriously. In reality he might drop more if he off-races, and win even more if he sticks to Terran.
In before you post with a smug comment: Shortly after the Defiler Super Tournament Sziky's still top of the rankings by a few points and the rest of the foreigners would have had similar chances against the leading duo. It wasn't so easy to see if you didn't have enough experience going up against Scan yourself.
Other than that, the obvious fact remains that Sziky is basically untouchable. However, in between January 2012 and December 2014 several people made a good run for his throne, most notably eOnzErG, trutaCz and TechnicS. The least stable of all was the Spanish, as he entered a vast number of minor and fun events. TechnicS was the most consistent, but his period of leadership lasted the shortest. The Polish' dominace lasts until today, as can be extrapolated from the data on first glimpse.
Overall this paints a rather interesting picture: The Top 10 usually consists of one Protoss (Tama, Alfio, dRaW, Arcneon or dsaqwe, depending on what event was the most current [Defiler vs. TL events]), five or six Zerg (Sziky, eOn, tacz, Technics + Whistler/Julia/ZZZero/whoever) and two or three Terrans (usually Gargoyle + Pro7ect/Marwin/Heme + Terror/skzlime).
Event wise another interesting fact shows: The least "important" events are those hosted by Teamliquid. Even though attracting the best players, the TSLs usually only validated the rankings generated by Defiler and rarely helped to estimate "new" players in the data base. The only notable exception had been Julia. However, several players magically did better in TL events than in Defiler - for instance Andrey (dredredre), Sero (marginally though) or FremAN. This isn't surprising as Defiler offered more opportunities to play. As a consequence, the Defiler Series seemed to be the best portal to track the trends - be it the war between Sziky and Scan, or rising talents. Also, a number of "very good" players mostly showed in the Russian event series, but nowhere else: Ramms (until mid-2013), Gargoyle, Sea[SR], Zolotoi, Werdum (until late 2014), flisk (until early 2015).
As for "activating older and newer players" the second most important tournaments were those hosted by the Swedish Brood War Intiative. In those especially the Peruvians showed their capabilities - Dienmax, CastrO and several other somewhat notable players emerged.
The probably most surprising player to appear as "#1 Foreigner after Sziky" is gargoyle. In my wordpress blog I counted the Terran as a top 10 player in between 2010 and early 2013. However, not that much more.
His high placement in the Elo event rankings can be explained in parts. To begin with, he exclusively played in Defiler events - most serious ones (regular Defiler Tours) and some more less important ones (eOn Tour, Gaz Tours). In those, he has defeated a number of the best foreigners. The table above shows his games from the past year against players with more than 1150 points at the time. You can see how he's rather successful against top tier Zerg players, most of all people like TechnicS and eOnzErG. At the same time, Gargoyle usually had to leave early and thus (randomly) minimized the probability of being defeated in a Best of One against a relatively unskilled player.
The question remains where gargoyle should really be placed. My guess is that he would be able to maintain a 1300 rank in the current scene without much of an effort. Unless more really talented Protoss players emerge, his special match ups against Terran and Zerg will carry him on.
Before we finish, some interesting graphs. Everyone likes them.
This graph shows how much Scan improved in his absence during 2013 and most of 2014. He came back and dropped some games with his off race, then suddenly overtook anyone else - from 1500 up to 1700. Sziky however remained where he was, the rising amount of his points can be explained by the "point inflation" appearing automatically over time. Most of the foreigners, even the best ones, still struggle to reach 1400 points and maintain that level. With the exception of trutaCz nobody came close crossing 1500 points.
This graph shows the leading three Zerg players TechnicS, eOnzErG and trutaCz. Notice how trutaCz limps from tree to tree and is now unreachable. It also contains the graph for Marwin, because he's my favourite Terran after Largo, though he doesn't fear the duck.
The last graphic shows how much Elo rankings might help to identify upcoming stars. I spammed these forums a lot, but had no evidence supporting my reasoning to just feature flisk's stream. Starting with the Defiler Super Tournament his points steadily went uphill - so much that he now belongs to the elite. If you don't know him, keep an eye on this player. He really is the player to improve the most!
That'd be all for now. I can't promise I will update this project regularly - but I will try. Also:
Happy Birthday Zaraki
everyone who updated Smurf/list (<3 Malkiyah)
vOddy (proof reading some of this post)