FerahgoTheGreat
Flag: United States
Registered: July 25, 2021
Last post: November 25, 2024 at 9:47 PM
Posts: 548
1 2 3 4 •• 11

buffoons

posted 3 weeks ago

For various reasons (tech issues etc) the tournament organizer will have to stop a game and restart it from a new lobby (or maybe it is just some restart of the same lobby idk). I don't know how it works, but I believe they can edit the score, economy, guns/util people have, etc to get the game back to the same state they left off. That does however, break our data acquisition to Riot's api.

posted 3 months ago

The rating doesn't start anywhere. It is essentially (not exactly but close enough) calculated on a per round basis and averaged over rounds. If you get average stats, you will get a rating close to 1.

posted 3 months ago

Yeah, when the lobby has a remake, we lose most/all of the round by round data used to construct some stats like the rating. Sometimes the end of match stats can still populate because they can be plucked directly. The 0.68 rating probably only applied to the last handful of rounds.

posted 3 months ago

Nope, that has been a thing the whole time

posted 3 months ago

2021: cNed
2022: yay
2023: Leo
2024: aspas so far

posted 6 months ago

Hopefully the lack of CN stats wont be permanent, but from what I understand, CN data security laws will continue to prevent the kind of api access that we have in other regions.

posted 6 months ago

Pretty good. I got my Masters, and I am starting on a PhD.

posted 7 months ago

Due to rounding cutoffs, I think it would be possible for the rating to show as 0.00 on the website if the true value is <0.005, but I have never seen one that low.

posted 7 months ago

I do know the exact formula since I made the rating, but I can't distribute it. This article explains the general principles behind it.

posted 7 months ago

Thats not mathematically how that works. The rating is a linear combination of several individual statistical components (KAST is not used in the calculation). The distribution of each component can be approximated as a gaussian, and the overall rating also looks like a gaussian. A mathematical bell curve has a tail that goes to negative infinity, and because the rating is arbitrarily scaled to have a mean of 1.0 and a custom standard deviation, there is no mathematical reason the pure rating couldn't go negative. In practice, almost no ratings get below 0.1, and the algorithm is built to force an asymptote at 0.0, so negative ratings are not possible.

posted 7 months ago

The rating is built to asymptote at 0.0. Significantly less than 1% of ratings get below 0.1, and the system is built to keep them from going negative.

posted 7 months ago

Throwback to one of the most legendary forum posts in VLR history.

posted 8 months ago

Shoutout to the person who added this to the wikipedia VLR disambiguation page on Oct 18, 2023:

"Vlr is a website where there is a valorant forum where people can see stats and what happening in valorant"

(It has since been updated with something more professional)

I assume it was one of the forum users here because of the broken english in some of their wikipedia updates and the fact that their latest update is adding "hi I am so cool and giga chad" in the middle of a random sentence for a random basketball player (reverted literally 4 minutes later for vandalism lol). Classic internet forum user behavior right there.

posted 8 months ago

This reply to a reply that will be updated with a full day recap about an article that will be updated with a full day recap will not be updated at all

posted 9 months ago

Thanks man

posted 9 months ago

Maybe 👀

posted 9 months ago

I was shocked by how good the released Sora videos are. The previous state of the art latent diffusion technologies are so far behind Sora in so many ways. I've never been able to make a photorealistic character that is consistent between frames without flicker, and they got that plus camera panning, multiple angles, and halfway realistic movement/physics all at quality that I struggle to reach with still frames. Very impressive stuff.

posted 9 months ago

They didn't exactly reveal the format other than EG get the top seed and everyone else is put in pools and then randomly drawn.
Before the reveal, I predicted that the remaining 10 teams would be in 2 pools of 5 based on the LCQ/Champs performance, and it looks like that was the case for all three regions (idk what China will do). NRG, C9, LOUD, LEV and KRU got the high seeds while FUR, MIBR, SEN, 100T, and G2 got the low seeds. High seeds face low seeds but matchups are random.
This reasoning works in EMEA and the Pacific. The only question mark is Vitality getting a high seed despite not playing at the LCQ.

posted 10 months ago

My tweet was also made in 5 minutes, so if I did a deep dive into the analysis, I'm sure some players would move around the list.

posted 11 months ago

Demon1 didn't play in the first half of the year, and he was good but not great for the second half of the Americas League and Masters Tokyo. He was only elite at Champs (which is a huge deal that boosts him into the top 20), and his importance in the transformation of EG gives him a top 10 spot, but I think recency bias is really making people forget that for the majority of the year, he wasn't a top 20 player. Sayf absolutely dominated the EMEA league, and I give him a lot of credit for carrying a TL roster that I thought was pretty bad overall.
f0rsaken is nowhere near this list for his 2023 performance although he could definitely get back onto it next year. Jinggg was a top honorable mention, but his playstyle didn't translate to good performances against top teams even though PRX were winning.

posted 11 months ago

Massively needed improvements to the Challengers structure and League roster flexibility.

posted about a year ago

Dang, I haven't built a scraper in a while. This looks cool.

posted about a year ago

I like pleets, but the fact that he is a complete liability in terms of firepower may be a holdup.

posted about a year ago

Nosirskiiii

posted about a year ago

Common 🐔joe W

posted about a year ago

Pai

posted about a year ago

I don't know why people are getting confused by the twitter/vlr forum debate. Both are not the professional sites I'm talking about. I'm not saying that he shouldn't be allowed to post anywhere. I would find it incredibly disturbing if he was barred from posting in any public forum/twitter/etc. But, there is a market for trustworthy news, and the places with money pay to protect that trust. I am not saying he should be discouraged from posting, or from trying to climb, but to paraphrase an analogy I made earlier in this thread, people invested in making/selling/consuming complete diets can and will criticize the pure carb diet.

posted about a year ago

The analogy is, its a 100% carb diet that gets quick energy but will struggle to be healthy long term. There is a market for consuming that kind of content, but that doesn't mean people can't speak out against the glucose in favor of the complete diet.

In terms of a good starting point... its rough. It mainly comes down to networking which I would not enjoy doing myself. DMing journalists/talent/org staff/players/etc out of the blue will lead to a big strikeout rate, but getting some connections with people who do the kind of work you want to do in order to learn their techniques/collaborate/network is the most straightforward path. I don't have experience with investigative journalism, so I am not the best source for getting started, but thats how I would go. What I do know is: Patience is key.

posted about a year ago

Nope
twitter = public forums like VLR

posted about a year ago

twitter = public forums like VLR
you missed the point

posted about a year ago

Yeah, I am a graduate student currently. My summer internship was in a fenced off location with armed guards (although the kind of work they let interns do is not the kind that requires that stuff).

posted about a year ago

Just adding my two cents.

I quite enjoy the kind of "leaks" put out by people like james_ff and others. However, there is a clear difference between that kind of content and professional content produced by paid writers at sites like Blix, Dexerto, etc. Those sites would not publish that kind of work because the sources cannot be vetted or trusted. The problem a lot of "mainstream" journalists have with this kind of amateur content is the community reaction discrediting the long and difficult process it takes to produce actually trustworthy content and the confusion between the two. Its a similar battle they've had with the kind of "twitter journalism" produced by the likes of the Jake Luckys of the world. Trust takes months if not years to build and a single incident to destroy which is why professional investigative journalists spend days or weeks making sure they have multiple primary sources and working with editors that not only improve the writing but also guide the decision making process of whether something is publishable. The best investigative journalists have folders of unpublished reports (some that happened and some that didn't) that could have gotten them clicks, but were too risky to stake their reputation to. I think the amateur leaks have a place (VLR forums are a pretty good place for that), and I would actually encourage discussions like it (although hoping for civil discussions on public forums might be a tad ambitious of me). However, I would warn people trying to make it beyond that point, that it can and will become a detriment to have an untrustworthy reputation around anyone willing to pay real money for journalism.

Whether or not amateur leaks should be counted as "journalism" is a semantics game that I don't really care that much about. I'm a nuclear engineer, I just write for fun.

Will this change many people's minds? Probably not. Was writing this a waste of my time? Not to me. I got my start talking to anyone about esports on these forums, and I care far more about the ten sane people than the million rabid dogs. Voicing my opinions into the chaotic void is more about the principle than the outcome.

posted about a year ago

click heads. get rating

posted about a year ago

VLR and RIB have so much going on in terms of every fight being weighted by a ton of other factors (player differential, economy, trading, etc), and that is what drives the disparity. All three ratings play with damage differently, but it isn't a massive factor in any of them.

posted about a year ago

Since VLR, RIB, and THESPIKE all have ratings, but they are all really complicated (I know VLR's is because I designed it, and RIB's is open source), I made a "reverse engineered" approximation using only kills, deaths, assists, first kills, and first deaths.

https://www.reddit.com/r/ValorantCompetitive/comments/14v4s7q/approximating_valorant_ratings_vlrribthespike/

Some TLDR notes:
I can get under 2% relative error for THESPIKE while VLR and RIB are too complex to approximate with my methodology.
THESPIKE values kills ~10 times more than deaths
The value of an assist relative to the average kill/death is 0.77 for THESPIKE, 0.43 for VLR and 0.32 for RIB
RIB heavily favors opening duels (particularly deaths), but this might be somewhat affected by the poor approximation for VLR/RIB.

posted about a year ago

I don't know the link to the official rulebook, but I heard similar things from people who did read it. Here is what their public announcement said.

Those four teams will then be randomly placed into the double elimination bracket, where the top seeded teams have been waiting. Teams from the same group will be placed on opposite sides of the bracket to prevent a repeat match early in the winners bracket.
posted about a year ago

I do not believe that is guaranteed. From my interpretation of the stated rules, each group stage qualified team will randomly draw for an opponent out of the four already in the playoffs. Teams from the same group will be on opposite sides of the bracket.
FNC has a lower seed than Liquid, but that only impacts map veto priority I believe. There are four "veto pools". Within each pool, map veto priority is determined by a coin flip, but group 1 gets prio over group 2 etc.
Group 1: LOUD/TL/PRX, Group 2: FNC, Group 3: EG/NRG, Group 4: lower bracket group qualified teams.

posted about a year ago

RIP

posted about a year ago

If you have been following his twitter for the past months, he was absolutely backing more than just Uganda's "nonconformity" (which isn't all that nonconformist for that area of the world). Carlos has openly endorsed old school religious concepts of the specific roles for men and women in society along with anti-LGBT statements. I'm not saying he can't say these things, but you seem to think that he wasn't backing the sentiment behind that law which... he very much was.

posted about a year ago

Gens 4 and 5 were amazing.

posted about a year ago

He writes the scripts that Riot use for Valorant tournaments.

posted about a year ago

I already did one when LOUD won Champs. I also did one with AI generated images of my face when they made the playoffs of LOCK//IN. A twitter advanced search of my account and the word "promised" should get you both of them.

posted about a year ago

Some twitch clips have a specific character in the string identifier that breaks the display on articles. I had to go find a different one since mine broke. Anyways its there now.

posted about a year ago

Me? The person that picked LOUD to win at every single international event last year?

All jokes aside, my clip broke, so I just published, and searched for a different version of the clip afterwards lol.

posted about a year ago

Wow, what a massive roster move. That announcement video is so hype.

posted about a year ago
1 2 3 4 •• 11