I don't usually respond to obvious hate comments since I prefer to focus my energy towards positivity, but I'll make an exception here.
CONTEXT: Arktheos aka Aribo recently embarked on a quest to stop the EVIL "statbros" (:D), leaving numerous not-so-nice comments on my tweets and now making an entire thread bashing my recent video.
When you strip away the anger from his comments, his underlying allegations/criticisms focus on the following two topics:
1) "Statbros" like me just blindly trust data and don't think for themselves, and this is corrupting the League player base in their itemization choices.
2) "Win probability" is a fake stat that I invented that has no practical applications in League of Legends theory-crafting.
He bolsters these opinions by highlighting individual item recommendations by me or my mutuals (such as Dead Man's Plate Akali), which he considers to be objectively bad.
Now let's debunk these points a bit:
1) I constantly emphasize the need for a holistic approach to itemization, consisting of
a) theory (ability scalings, gold efficiency)
b) data (statistics, analytics)
c) practical experience (game knowledge, trial and error)
I explain this approach in detail in the video he linked (28:31). I even did a whole podcast episode about this topic with Coach Curtis and Nathan, explaining in detail why good itemization stems from a combination of these different lenses.
For a long time, people have ignored the data perspective, because the available data has not been trustworthy, which gets me to point number two:
2) Win probability (and especially WPA) isn’t something I made up. It’s a well-established metric in sports analytics, used for decades to evaluate performance and strategy.
For a more in-depth literature review, check the Master's Thesis in my pinned tweet, where I also explain the (simple) mathematical theory behind Win Probability Added.
Calling this an "imaginary number" (probabilities are always in ℝ btw :D) is ignorant.
Yes, some of my build recommendations don't pan out. That’s part of exploring off-meta ideas. But this doesn’t invalidate data-driven analysis.
Data requires interpretation. Most of my builds are extrapolations based on data—not blind trust in numbers. If anything, critique my game knowledge, not the data itself. I will not try to defend every recommendation I have made, but they have served their purpose: they've made people think about alternative builds.
I hold no ill will towards anyone here; my objective is to simply advance esports analytics and have a career in this field.
Thankfully, multiple teams, players, and coaches from the LEC and NLC have found my approach to League stats useful, so I will continue my work despite the backlash.
So I watched this whole xPetu video and it was suggested that upon reaching the lategame you should sell Guinsoo for GA, or some other items, because it's weak late, even on Kog'Maw?
We are entering an era of statbros causing the devolution of LoL... crazy =))
(vid had 0 value)