The best players of SC2 (extra dry read!)

Big J

Austria16289 Posts

March 24 2015 13:31 GMT

This is a blog about the results of a little tool I have have developed to rate players overall and their peak performances.
Without going too much into technicalities here - more below - the method is based upon each and every top4 Premier Tournament finish from 2010 up to today (19th of March 2015), all of them carefully handrated in 5 Tiers.

The Best Players of All Time (by overall performance):
+ Show Spoiler +

1. Mvp (7.592 points)
2. Life (7.458 points)
3. MC (7.175 points)
4. MMA (6.692 points)
5. TaeJa (6.083 points)
6. Polt (4.866 points)
7. Maru (4.65 points)
8. DongRaeGu (4.292 points)
9. PartinG (4.25 points)
10. INnoVation (4.075 points)

The Best Players of All Time (by peak performance):
+ Show Spoiler +

1. Mvp (3.625 points)
2. soO (3.25 points)
3. DongRaeGu (2.979 points)
4. Life (2.967 points)
5. Zest (2.875 points)
6. MMA (2.808 points)
7. Soulkey (2.65 points)
8. INnoVation (2.5375 points)
9. Jaedong (2.417 points)
10. Maru (2.4 points)

The Best Players of WoL (by overall performance)
+ Show Spoiler +

1. Mvp (6.675 points)
2. MC (5.008 points)
3. DongRaeGu (4.125 points)
4. NesTea (3.517 points)
5. MMA (3.467 points)
6. MarineKing (2.992 points)
7. Life (2.416points)
8. Leenock (2.1 points)
9. TaeJa (2.083 points)
10. Polt/PartinG (2.025 points)

The Best Players of WoL (by peak performance)
+ Show Spoiler +

1. Mvp (3.625 points)
2. DongRaeGu (2.979 points)
3. MMA (2.808 points)
4. Life (2.417 points)
5. NesTea (2.375 points)
6. MC (2.167 points)
7. TaeJa (1.833 points)
8. Rain (1.8 points)
9. PartinG (1.775 points)
10. MarineKing (1.6 points)

The Best Players of HotS so far (by overall performance)
+ Show Spoiler +

1. Life (5.042 points)
2. Maru (4.65 points)
3. soO & TaeJa (4.00 points)
5. INnoVation (3.575 points)
6. Zest (3.5 points)
7. Jaedong (3.4 points)
8. MMA (3.225 points)
9. Polt (2.842 points)
10. Soulkey (2.817 points)

The Best Players of HotS so far (by peak performance)
+ Show Spoiler +

1. soO (3.25 points)
2. Life (2.967 points)
3. Zest (2.875 points)
4. Soulkey (2.65 points)
5. Jaedong (2.417 points)
6. Maru (2.4 points)
7. INnoVation (2.325 points)
8. TaeJa (2.092 points)
9. sOs (2.025 points)
10. Dear (2.0 points)

The best Foreigners (by overall performance):
+ Show Spoiler +

1. Stephano (2.133 points)
2. NaNiwa (1.817 points)
3. HuK (1.375 points)
4. MaNa (0.917 points)
5. ThorZaIN (0.908 points)

The best Foreigners (by peak performance):
+ Show Spoiler +

1. Stephano (1.217 points)
2. NaNiwa (0.983 points)
3. Jinro (0.8 points)
4. HuK (0.725 points)
5. ThorZaIN (0.575 points)

How it works
+ Show Spoiler +

The Five Tiers of Tournaments:
Tier 1 - Preparation tournaments of the highest level of their time with a certain factor of exclusiveness usually granted through a tough qualification process. E.g. GSL, WCS KR, NSSL 2012-2015, OSL 2013

Tier 2 - Tournaments that for a certain reason (e.g. lack of preparation time or too high frequency of the tournament in a year) do not make Tier 1. E.g. GSL 2011, OSL 2012, WCS Blizzcon Finals

Tier 3 - Similar to Tier 2 tournaments which are usually below T1 and T2 tournaments in tournament length, price pool and prestige, yet still feature the absolute top players in the world. E.g. Hot6ix Cup, GSL Blizzard Cup, IEM WC 2014 & 2015, GSL World Championship 2011, WCS AM, WCS EU

Tier 4 - Frequent Premier Tournament Series which usually include a lot of traveling, no preparation time for opponents and are usually held at LAN events. E.g. MLG, DH, IEM

Tier 5 - Like Tier 4, but with a weaker player line up, a very small bracket or less of a competetive character. Basically all the other tournaments that are not listed in the other Tiers. E.g. HomeStoryCup, Red Bull Battlegrounds, MLG/DH/IEMs with weak player line ups.

For each placement as 1st, 2nd, 3rd or 4th player in an SC2 Premier Tournament, said player gets points.
Tier 1 - 1 point/0.75points/0.5points/0.5points
Tier 2 - 0.8 points/0.6points/0.4points/0.4points
Tier 3 - 0.5 points/0.375points/0.25points/0.25points
Tier 4 - 0.33 points/0.25points/0.17points/0.17points
Tier 5 - 0.15 points/0.1points/0.075points/0.075points

So, to put it into perspective: you need three Dreamhack championships to match the points of one Code S champion. For every one point a player has, he has achieved as much as Seed. A player that won one Tier 1 GSL, but never ever placed top4 elsewhere.
Example: Mvp's overall achievments are more impressive than winning 7 GSL titles through all of SC2's lifetime. More than the best 5 foreigners combined.

The second statistic I'm showcasing here is the peak performance of a player. The performance is a curve that is a weighted sum of overall points achieved within a thirdteen month time frame.
E.g. Mvp peaking at 3.625points in October 2011 means that the sum of points he has achieved between July'11 and Jan'12 as well as half the points he got from April'11-June'11 and Feb'12-April'12 is worth more than 3 GSL wins.
Only the peak performances (hence the highest point of a players performance curve) are given above. For some complete graphs look below.

Some performance Graphs
DISCLAIMER: Due to the points handed from events to the past it is only natural that the graphs around early 2015 are declining in comparison. (those points in the plot have zero points from future events right now) This makes late peaks like Life's even more impressive, since those peaks could be even higher a few month from now.
+ Show Spoiler +

Mvp vs Life+ Show Spoiler +

Mvp: He has reached the highest peak ever achieved, showcased in every graph by the blue Mvp-line.
Life: The only player (I believe) that would have made it into the top10 peak performances with two
far apart peaks. His current, higher peak is ongoing and does not profit from future events in the same way that Mvp's 2011 peak profits from 2012 wins yet!

MMA, MC, TaeJa+ Show Spoiler +

some of the most steady players overall. Even though they had their ups and downs (in MMA's case a down to zero even), they have been around for a really long time and performed well throughout most of it.

Rain, sOs, Zest, Dear, herO, PartinG+ Show Spoiler +

a lot of Protoss before the Void.

Maru vs INnoVation+ Show Spoiler +

No clear answer to the best Terran of HotS question.

Life, Soulkey, soO+ Show Spoiler +

The Heart of the Swarm!

Seed, jjakji, Sniper, Classic, RorO + Show Spoiler +

Who is the biggest one-hit wonder amongst the GSL winners?

Stephano vs NaNiwa+ Show Spoiler +

Foreigners fighting ~~back~~ each other.

Flash vs Jaedong+ Show Spoiler +

A clear advantage in the eternal duel so far.

herO vs HerO+ Show Spoiler +

Which hero do we deserve?

DRG vs NesTea + Show Spoiler +

Behold Artosis, DRG #1 in WoL!

soO vs Zest + Show Spoiler +

2014's best?!

There may be a lot of smaller mistakes in there. (e.g. discovered that I typed INnoVation with just one 'n' in GSL 2013 S1 HotS... missed out on quite some points before I corrected that one) And since I was too lazy to program proper tools to analyse all the handcopied Premier results everything was manually observered lol... If I'm somehow missing some godlike player that you really think should be on some list, please tell me and I'll doublecheck.

OtherWorld

France17333 Posts

March 24 2015 13:46 GMT

I like these rankings, gj

Deleted User 137586

7859 Posts

March 24 2015 13:54 GMT

Great read. But it's not dry enough! Can we see your calculations? It would be good to see how the points actually form (staring at the data, the most scientific of methodologies.)

Intuitively, having the weighting of winning a tournament being only double that of a 4th place seems off to me. Winning the last two matches, especially the generally bo7 at the end, should carry more weight in the system.

A typo: second foreigner spoiler should be peak performance.

Edit: Here's a wrench into the gears that I don't see a quick solution for: accounting for team league performance.

Big J

Austria16289 Posts

March 24 2015 14:42 GMT

Intuitively, having the weighting of winning a tournament being only double that of a 4th place seems off to me. Winning the last two matches, especially the generally bo7 at the end, should carry more weight in the system.

Usually such sorts of ratings take quite some experience to be finetuned. The reason I chose this rating system of 100%-66%-33% or 100%-50%-25% was plainly because I liked the results.
For example a benchmark I was (briefly) looking at was Zest vs soO, making sure that soO wouldn't drop off too hard in comparison. Some of the Kongs we have had really were some of the very best players in the world placing high much more constant than 1time champions.
It's far from a perfect system of course, question whether there is such stays open anyways. Every system really just does what his creator intends to do with it. And graphs just look much better if there is real movement in there and not just one peak where a GSL has been won.

I think the way you can look at it is that:
- multitime champions will place high regardless
- players who only have few high (non-champion) placements still don't make it anywhere.
--> you're either a champion or a top-Kong to make those lists. And that feels about right to me.

Obviously one could also argue that regardless of whether you get 2nd or 3rd, you only lost one series, possibly even only against the later champion. But I feel like that goes against the prestige that comes with making a finals.
At the end, "the best" is who we remember as such. And Curious, Cure or even TaeJa making GSL semi's will never be as much on our minds as Inca with his infamous second place.

Edit: Here's a wrench into the gears that I don't see a quick solution for: accounting for team league performance.

Yeah, I agree. You could say this is sort of a first version, if the interest is high I could include those too. At this point it was mainly a question of work efficiancy to not include them. It means a lot of staring at stats and deciding how much a GSTL is worth and how much a Proleague. In the end it came down to "well, do I really try to find a somewhat objective system just to give Flash and sOs a few extra points. Since Teamleagues besides Proleague simply haven't been that prestigeous/meaningful anyways."
But I'll keep it on my mind.

Boucot

France15997 Posts

March 24 2015 15:56 GMT

That seems to be pretty accurate. I like it.

Sakat

Croatia1599 Posts

March 24 2015 15:59 GMT

Wow, I didn't realize how close Life and Mvp are...

linwqdwqdwqd77

Albania3 Posts

March 24 2015 16:06 GMT

--- Nuked ---

Ej_

47656 Posts

March 24 2015 16:36 GMT

So it means that a WCS AM or EU title has the same weight as a GSL one?

Big J

Austria16289 Posts

March 24 2015 16:41 GMT

On March 25 2015 01:36 Ej_ wrote:
So it means that a WCS AM or EU title has the same weight as a GSL one?

Oh sorry, forgot to mention them. Only WCS KR is a Tier 1 event. AM and EU are T3.
Gonna change that when I'm back at a Computer.

OtherWorld

France17333 Posts

March 24 2015 16:44 GMT

#10

I'd say that WCS EU and AM are definitely on the level of 2011's GSL. (except the 2012 ones ofc)

BisuDagger

Bisutopia19230 Posts

March 24 2015 16:45 GMT

#11

Amazing Jaedong makes both of the HoTS the list.

Shellshock

United States97276 Posts

March 24 2015 16:51 GMT

#12

On March 25 2015 01:45 BisuDagger wrote:
Amazing Jaedong makes both of the HoTS the list.

He did at least make it to all those finals which is more than a lot of players can say even if it was in foreign tournaments

OtherWorld

France17333 Posts

March 24 2015 16:54 GMT

#13

On March 25 2015 01:45 BisuDagger wrote:
Amazing Jaedong makes both of the HoTS the list.

Well he had an amazing 2013 and was probably among the top3 Zergs in 2013. His 2014 was underwhelming and disappointing but not that bad.

Charoisaur

Germany15920 Posts

March 24 2015 18:56 GMT

#14

surprised soO has the highest peak performance in HOTS despite not winning a single tournament.

edit: i don't agree with the decision to rate preparation based tournaments higher than weekend tournaments with the same level of competition.

who says that a preparation tournament is harder to win? some players excel at preparation tournaments but disappoint at weekend tournaments and other players who dominate weekend tournaments suck in preparation tournaments.
why should weekend tournaments count less even when the competition is equally strong?
you could also argue that weekend tournaments should be rated higher because it's harder to win a tounament where you have to adapt to different opponents on the fly with little to no time to analyse their games and prepare build orders.

As it is it seems very biased to rate preparation tournaments higher.

CosmicSpiral

United States15275 Posts

March 24 2015 19:24 GMT

#15

This blog was so dry, I didn't need a towel after my shower.

To make your calculations more accurate, I would suggest creating a system that rates individual tournaments by the general level of competitors involved. This would be extremely tedious and time-consuming but it's crucial to making meaningful sense of tournament finishes. Level of performance positively correlates with time since the medium player level increases over time in any competitive sport. It's no surprise that most derived statistical highs are recorded early in a sports' lifespan while most flat highs come near the end.

BisuDagger

Bisutopia19230 Posts

March 24 2015 19:28 GMT

#16

On March 25 2015 04:24 CosmicSpiral wrote:
This blog was so dry, I didn't need a towel after my shower.

To make your calculations more accurate, I would suggest creating a system that rates individual tournaments by the general level of competitors involved. This would be extremely tedious and time-consuming but it's crucial to making meaningful sense of tournament finishes. Level of performance positively correlates with time since the medium player level increases over time in any competitive sport. It's no surprise that most derived statistical highs are recorded early in a sports' lifespan while most flat highs come near the end.

What water proof device were you reading on. I want to imagine a projection monitor in the shower and water proof mouse. TL on my shower wall? HELL YES!

Ej_

47656 Posts

March 24 2015 20:05 GMT

#17

On March 25 2015 03:56 Charoisaur wrote:
surprised soO has the highest peak performance in HOTS despite not winning a single tournament.

edit: i don't agree with the decision to rate preparation based tournaments higher than weekend tournaments with the same level of competition.

who says that a preparation tournament is harder to win? some players excel at preparation tournaments but disappoint at weekend tournaments and other players who dominate weekend tournaments suck in preparation tournaments.
why should weekend tournaments count less even when the competition is equally strong?
you could also argue that weekend tournaments should be rated higher because it's harder to win a tounament where you have to adapt to different opponents on the fly with little to no time to analyse their games and prepare build orders.

As it is it seems very biased to rate preparation tournaments higher.

It's simply because you see higher level of play in preperation tournaments. It's harder to win them for people like Taeja, because people who he'd crush in every weekend event prepare and prepare and bring their absolute A-game in GSL. They are considered harder because people play better than in weekend tournaments.

And then there's Life who just wings it and casually 3-0's Taeja in GSL semis.

boxerfred

Germany8360 Posts

March 25 2015 14:36 GMT

#18

nerdy 5 star math

vult

United States9400 Posts

March 25 2015 15:42 GMT

#19

Seems to be quite realistic, to be frank. Strange to see that, even with their seat at the top of the foreigners for quite some time, Snute and Scarlett are not included in the list, while HuK and Thorzain are :o Time goes fast, it's been so long since I have seen them as truly one of the best.

Ctone23

United States1839 Posts

March 25 2015 20:13 GMT

#20

only furthers my respect for JD..

Thanks for this! No complaints here..

Shellshock

United States97276 Posts

March 25 2015 20:16 GMT

#21

On March 26 2015 00:42 vult wrote:
Seems to be quite realistic, to be frank. Strange to see that, even with their seat at the top of the foreigners for quite some time, Snute and Scarlett are not included in the list, while HuK and Thorzain are :o Time goes fast, it's been so long since I have seen them as truly one of the best.

Idc I understand right there is a lot of emphasis on winning tournaments or at least going deep which was a lot more doable at huk's and thorzain's peaks

Cricketer12

United States13974 Posts

March 25 2015 20:34 GMT

#22

SO MUCH DONGRAEGU :D:D:D

A3th3r

United States319 Posts

March 26 2015 17:25 GMT

#23

MVP being first doesn't mean he's best.. just first

Kazahk

United States385 Posts

March 26 2015 18:45 GMT

#24

In other words Mvp is god

Big J

Austria16289 Posts

March 26 2015 23:45 GMT

#25

Some replies to some posts. Thanks for all the nice comments!

To be honest, it is a question of time. Since I'm not getting paid for doing math on these things and I have no working model that I could easily adapt to do just that... yeah, the workload/learning effort is simply too high. Sadly.

If I had the knowledge and time I would love to use aligulacs databank to apply and finetune a spatial econometrics model to create player clusters, rate "regions" and try to create a dynamic world ranking of players. Would be a very interesting work, but way too time consuming if you don't do it professionally I guess. More reason to admire aligulac's creators for the work they are putting into their database and rating system, even if people don't really use their top lists that much.

On March 27 2015 02:25 A3th3r wrote:
MVP being first doesn't mean he's best.. just first

This is very much true. I believe that 2011 Mvp would be in trouble against the top 5% of the ladder these days, if not even more. The skill - in particular strategic skill - has grown so much since then. 2015 Mvp would wipe the floor with 2011 Mvp... yet 2015 Mvp has sadly become irrelevant.

But that isn't the purpose of these statistics. It's rather about making achievments comparable in one way or another, possibly setting standards for when we consider someone a Bonjwa or something like that.

On March 26 2015 05:34 Cricketer12 wrote:
SO MUCH DONGRAEGU :D:D:D

I was surprised myself how high DRG peaked. Even more, I found some small mistakes in the rating of soO and Maru, which might have a small impact on the ratings making DRG #2 in peak. (gotta rerun some calculation)
I'm not quite sure why, but the top players of late 2011-early 2012 seem to be rated a little bit higher than the others. There might be some mistake, but also when copying top4's from that period I remember that I had to type the names Mvp, DongRaeGu and MMA very often. They were consistently very strong back then, much more than todays top players I think.

The thing is that back in those days HuK and ThorZaIN actually won and placed high in many tournaments that in general had quite strong foreigner:korean ratios. Even rating them as T5 tournaments (which is very low) gives them points, which top8 finishes of Scarlett and Snute do not. The competition was different back then. I'd say amongst other indicators, this showcases how incredibly thin the air become since Kespa players have switched and caught up in 2012-13. The Korean scene still seems to be oversaturated with talent (more talents than can actually feed of progaming) while the foreign scene has had a hard stand without region lock in WCS and inferior/nonexistant infrastructure, making it very hard for new players to reach the top. Before the Kespa switch and after the SC2 launch hype with WC3, SC1 and new players of all nations flooding the scene the conditions were just much easier/better for a foreigner to reach the top.

Serimek

France2274 Posts

June 04 2018 16:23 GMT

#26

I found this thread today and read it with great interest. Do you think you could do an up-to-date list counting results from the last three years ?

Thank you for the read anyway

Chairman Ray

United States11903 Posts

June 04 2018 19:30 GMT

#27

soO is the second best player of all time by peak performance

Sounds about right

Normal

Please or register to reply.

The best players of SC2 (extra dry read!)

Completed

Ongoing

Upcoming