Common misconception. Larger sample size has diminishing value on predictive ability, especially if additional data is out of date. If I wanted to predict the U.S. election for example, I wouldn’t just average today’s swing state pole with one from January simply because it has a larger sample size. That error would be obvious and it’s what you’re doing here. You’re obsessing over sample size at the cost of quality data I.e. current meta.
1000 games is plenty large and the current trend line shows a balanced game. If the trend changes in the future maybe we should revisit the issue but any claim of imbalance in the current meta is speculative at best and goes against current data at worst.
Your regression used the wrong variables so again is useless. No one claimed date or specific best of fives defined the meta. Meta and balance occur over many weeks and thousands of games. Your regression just shows that most matches aren’t decided by the meta which we already knew and, while I’m glad to see it’s true, wasn’t what you set out to prove.
I just proved that there is no correlation between PvT performance and time and you are saying it’s wrong by fantasizing about a completely unrelated scenario where you assume there to be a change over time.
I don’t argue against things that exist only in your head and which are in complete contradiction to the facts of the situation at hand. If there is no significant trend of PvT win-rates over time, e.g. if PvT has stayed the same the past year, then there is no reason to exclude PvT games from previous months.
No you didn’t. You think you did because stats is hard.
You showed that most matches aren’t meta dependent (because match delta was your dependent variable rather than overall Terran win performance). That’s true over all time periods.
Once you start denying fact you’ve lost the debate kiddo. It’s sorta like a rage-quit in SC2.
The time-variation of PvT matches was 0.000224. That’s a fact. It’s so close to zero it means PvT performance is stationary with regards to time. That means PvT games from up to a year ago are valid for today. PvT has favored Protoss this year. Protoss is overpowered and it’s not a meta/map pool issue.
Do you want to talk about pvt win % or do you want to talk about the average games in a pvt best of 5? You started with win percentage and then went to a different variable to run your regression. That’s not how math works lol.
Not a fact just because you say it is, especially when you’re trying to draw a conclusion about a different variable than what you used in the regression.
Nope, what he said is blatant fact denial and therefore not only wrong but willfully wrong. He said recent TvP win-rates are becoming more balanced. He does this because he wants to put more weight on the more recent win-rates which confirm his conclusion. I proved that PvT performance has no correlation with the progression of time, thus obliterating his claim that win-rates are becoming more balanced. In reality he’s just cherry picking a small data cluster so that he can ignore the bigger set of data (which proves beyond doubt that PvT is Protoss favored).
It is a fact kiddo and not because I said so, but because it is factual. I proved beyond doubt that PvT performance is static with regards to time. Your justification for ignoring PvT winrates in 2020 was that the game is becoming more balanced in recent months. You are factually wrong. I proved it.
You then fantasized about a scenario where there is a trend over time and said this situation, which has none, is wrong because of your imaginary scenario. You are operating in a realm of imagination where the things you imagine are in direct contradiction to the facts of this particular scenario yet are magically not only relevant to this scenario, but supersede the actual facts of this scenario.
I don’t debate against delusion. I debate against facts and only facts.
So back in May I did an analysis of mirror Elo rankings (using the Elo rankings of players’ mirror matchups as a base-line for player skill). I then computed the difference between mirror matchup performance and non mirror matchup performance, assuming the difference must be caused by balance, which allowed me to define a win-rate for balance at every skill level in Aligulac. I predicted a 61/39% win-rate split for Korean TvZ. I forgot about the analysis until just recently I found out that TvZ win-rates for Korean Premier events is exactly 61/39%. The REEEEEing of forum trolls has been marvelous.
No you didn’t, you proved match length was static with time, which has probably been turn for all 10 years of the game. You were careless with your variable selection and sloppy with your application trying to draw conclusions about variables not studied.
My guess is you have an off the rack excel spreadsheet and no godly idea how to use it.
You call us low skill, yet your race is the only one that has never won a GSL in LOTV… Just saying man, the Protoss victim complex has exceeded the forum Terrans.
Where did all these extremism posters come from. Spewing not an ounce of evidence beyond their 1 sentence bs claims. So many names I don’t recognize, guess many of the older trolls have changed their names.
They all ended up being banned. Forum smurfing is worse than ever since they can now create an infinite number of free accounts and if one gets banned they just stop using it and make a new one.
Stats won one back with the super-adepts at the start of LotV. Other than that I agree 100%. Maru was winning Code S from a massive balance disadvantage in PvT but still won. It’s a mystery to Protoss that a 5% win-rate advantage doesn’t magically win every top tournament, and that skill, which can affect a win-rate much more than 5%, is what decides that. If Protoss can’t win tournaments regardless of balance, it’s a skill issue.
The PPP see reality completely inverted from the way that it is. The guy in this thread is a great example. All the facts are wrong except the ones which confirm he is right because reasons. His reasons shown to be wrong because of facts. Those new facts are wrong because reasons. Now trapped so he’s resorted to lying about there being a difference between match length and win-rates (there is not).
They’ve convinced themselves the skill deficit at the absolute top of the pro scene is proof that they lose games in diamond league, never-mind the fact that it’s Zerg that is being wrecked on the ladder right now. Lmao.
I agree. Protoss at this point just need to work harder. I heard awhile ago Stats was more focused on streaming than winning tournaments. All the really good Protoss players are just gone. Parting is arguably still the best with Zoun being a promising newcomer.
Streaming is literally the worst thing a pro player can do to win tournaments. They need to hide their ladder account and make sure nobody knows how they are playing.