[GSL Code S] Season 2 2017 Semi Finals - Day 1 - Page 19

Post a Reply

Prev 1 17 18 19 All

Tyrhanius

France947 Posts

June 15 2017 18:19 GMT

#361

On June 15 2017 04:00 TerranOwnsAll wrote:
Do you guys think Gumiho has a better chance against SoO or Classic?

SoO.

For me Classic is the favorite right now, but soo can still beat him with good cheese

pvsnp

7676 Posts

June 15 2017 20:10 GMT

#362

On June 15 2017 22:37 lolfail9001 wrote:

Show nested quote +

On June 15 2017 12:17 pvsnp wrote:

On June 15 2017 10:07 lolfail9001 wrote:

On June 15 2017 08:18 pvsnp wrote:

On June 15 2017 05:54 lolfail9001 wrote:

On June 15 2017 04:35 pvsnp wrote:

On June 15 2017 04:16 Elentos wrote:

On June 15 2017 04:08 pvsnp wrote:
TY has the best TvP around.

GuMiho can actually take games/matches against herO and every other top Protoss for that matter, I think the jury is still out on that. His style might not be as pretty and refined as some of the others but it's really strong. It's actually his best match-up.

And soO just kinda struggles against GuMiho in general usually, even though his bio TvZ is weak-ish.

Gumiho can take games, yes, but not matches. In 2017 offline matches, the only series he's won against a top Protoss was Stats in the Super Tournament. And that happened right as Stats was slumping after GSL so I'm not sure how much that counts. Other than that one win, he's lost to herO three times and Stats once.

Eh, if soO beats Classic he will have plenty of prep time and Inno as a practice partner.

Talking about offline matches when these barely happen is kind of funny way of cherry picking.

Perhaps you missed this event, I think it was called "GSL." I hear there are a lot of offline matches going on there. In fact, I hear this guy called 'Gumiho' is going to be playing an offline finals there, amazing right?!

And there was SSL, and IEM, and the Super Tournament. Guniho has played 41 offline TvP games in premier tournaments alone since the start of the year.

But of course it's totally picking cherries to use offline matches to predict an offline final, I mean who the hell would think a representative sample is actually a good idea?

41 offline TvP games in 6 months? 10 hours worth of games over 6 months is such a representative sample, in a game where players regularly peak in form for a week or two (of calendar time). In every match-up, because that's how peaks of form work, just ask Dear.

No, your sample is indeed a classic case of cherrypicking, you went digging few months back just to try and make a point about his match-ups but forgot in the process that knowledge from 6 months back is never relevant in SC2 for judging present form.

I used the games since the new year because the sample whose size you are already bitching about would get even smaller if I shrank the date range. So either we have a non-representative sample because of date, or a non-representative sample because of size. Or a non-representative sample because of online/offline type. You seem to have forgotten that Starcraft II players simply don't play as many games as your obtuse definition of sufficiently large sample requires. You've insisted on using a sweeping definition of "cherry-picking" such that there's no good answer, the games are useless data in all cases, and ergo it is pointless to mention any of this at all. Worthless definition, worthless discussion.

Congratulations on wasting my time.

Yeah, sample is unrepresentative so maybe it's time you'll learn that every time a Korean player went on a tear in offline matches he first went on a tear in online matches/ladder in general. So, online matches freaking matter, and excluding them just because they don't allow you to push your point bringing up half a year old games really does not help you.

Ah, so that was your point the entire time! That online matches "matter?" For a moment I thought you actually had a legitimate statistical objection about sufficiently large samples. Thanks for clarifying.

Now about online matches....perhaps you've heard of this guy called "ByuN?" In all of his HotS online matches, he had a TvP winrate of 73% and a TvZ winrate of 78%. Now according to you, this insanely good form should most certainly be reflected in his offline matches. Except we can look at his offline winrates..... and we see 35% and 41%.

And that's just one example. INnoVation beat the crap out of all his top-tier opponents in GSLTV right before he got stopped by Stats in Season 1 and aLive at Katowice. Scarlett had huge hype after she beat ByuN and Inno during NationWars and somehow never translated that into amazing offline performance. I can go on. The abundantly obvious point is that online results are chronically unreliable at best for predicting offline results.

And don't even get me started on the ladder. Does the term "ladder hero" mean nothing to you? Before the Season 1 finals soO had multiple accounts in top 10 gm. Didn't help him. Before his Ro8 match Rogue was #1 gm. Didn't help him. INnoVation was consistently in the top 5 throughout the first season. Didn't help him either. Endless examples that all prove you wrong.

To be overly pedantic, yes online matches/ladder do matter (for instance if a player had 0 offline matches we could use online since anything is better than nothing). They just don't matter anywhere near as much as offline ones.

Not to mention, in this context of predicting finals (Gumiho vs Classic/soO) your assertion that "every time a Korean player went on a tear in offline matches he first went on a tear in online matches/ladder in general" is completely worthless. If we can only judge retrospectively (after a tear offline we look at the previous online) then how the fuck can you predict offline results before they happen? (Hint: you can't). Depending on whether you meant this to be a biconditional, it's either worthless or wrong. Take your pick.

I'll make this simple: online isn't offline. Seems kinda obvious that online shouldn't be treated as offline for the purposes of predicting a match. Can't believe I have to explain that, but hey, "alternative facts" are still a thing. Once again, congratulations for wasting my time (actually this is getting fun, so maybe that should be thanks for wasting my time instead).

Elentos

55458 Posts

June 15 2017 20:56 GMT

#363

On June 16 2017 05:10 pvsnp wrote:
Now about online matches....perhaps you've heard of this guy called "ByuN?" In all of his HotS online matches, he had a TvP winrate of 73% and a TvZ winrate of 78%. Now according to you, this insanely good form should most certainly be reflected in his offline matches. Except we can look at his offline winrates..... and we see 35% and 41%.

And that's just one example.

A horrendous one. ByuN went on a break after December 2013 and came back in 2015. He played 392 of his 460 HotS games in 2015 and it was the phase where he was doing the best online in HotS. However, he didn't compete offline at all at that point, meaning there actually was no chance for his online results to translate to offline.

At least put in more effort to find a good example. And ByuN is also an example of online tear before offline success for LotV, he was winning every online tournament with ease and eventually it translated to good offline performances.

pvsnp

7676 Posts

June 15 2017 21:26 GMT

#364

On June 16 2017 05:56 Elentos wrote:

Show nested quote +

My bad, didn't notice the dates jumped a year.

But looking at 2017 ByuN, he was doing great (over 70% winrate in all matchups) before he got knocked out in GSL Ro32. He's been doing consistently well online all year, but offline the best tournament result he's had is Ro8. The original Inno + Scarlett online examples are also still valid.

I suppose I could put in more effort and find more examples, but the point of them is showing that online form is very much a nondeterministic and therefore unreliable predictor of offline performance. One counterexample is enough to prove falsity of a universal quantification like "always" in the original assertion. (Obviously, if something is not true in all cases it is not true at all)

Also I am at work and have probably spent too much time on TL as it is

Prev 1 17 18 19 All

Please or register to reply.

[GSL Code S] Season 2 2017 Semi Finals - Day 1 - Page 19

Completed

Ongoing

Upcoming