ambosen, on 23 March 2024 - 11:48 PM, said:
How many times am I going to have to explain basic math to some of you again?
Let me start off by saying that I am genuinely interested in understanding your perspective/point more. I will admit that I do not quite understand what you are trying to say. Let me also say that I have match data for thousands of matches across several years, tiers, and accounts. If I understood what you are saying a bit more, maybe I could look into the data I have and see if I can find any supporting evidence.
Quote
Averaging is a function of sample sizing. That's how PSR works. They've never pretended otherwise. What this means is that functionally your PSR score is controlled by a combination of how many matches you've played, the few remaining variables they track to determine score, and then derived from an averaging of the two.
The last part of the last sentence confuses me a bit. Could you clarify? A pilot's given PSR is a cumulative sum of all of the individual game PSR scorings. Each individual game PSR scoring comes from their match score relative to the pilot average for the match, and the pilot average for their team, with a bit of weighting.
Quote
What this means is a player with a single game where inexplicably the entire enemy team, or even just a significant percentage of both teams doesn't actually do much can quite easily end up with a high or low match score, and in effect PSR rating in comparison to a user with many more games under their belt, who is going to have a much harder time adjusting their rating. This is one of the reasons why you see a lot of the streamers who genuinely care about this game creating new user names, with stock mechs, and trying to show how quickly or or slowly it can take to change tier status; as far as the direct mechanic utilized applies, newer users are at a marked advantage for obtaining seemingly high PSR ratings. Longer term users are at a marked disadvantage.
When you say "adjusting their rating", you lose me. A pilot's PSR is a cumulative sum. How quickly one moves up through the tiers, on the aggregate, largely depends on one's skill (tempered by things like chassis/build) relative to the players in that tier. With our player population there really are only three "effective" tiers.
Once a pilot hits Tier 3, they get exposed to Tier 1. Due to seed-player dynamics and the three-tier-spread max, I believe that many Tier 2 players end up filling Tier 1 games. This would also mean those games (generally) do not see Tier 4 players. Tier 1 and Tier 2 are smaller tiers so my intuition is that Tier 1 and Tier 2 seeded games tend to pull each other in, plus Tier 3. It has been my experience (through several accounts) that the game changes dramatically once my account hits Tier 3.
Quote
Unless there's some sort of magic arbitrary cut off point to when PSR becomes completely fossilized (meaning no data is adjusted by subsequent games) that's set in the first game, whiich many user accounts to this game never get past more then one game played), then as a simple mathematical function, players with more games played literally have a mathematical certaintity of watching their PSR rating decrease after a certain point. Especially when the devs start removing variables that were once considered from the equation, which by the way, they've been pretty open with us have been multiple times.
Again, what do you mean by "PSR rating" here? Do you mean the orange bar in the UI that shows tier progress? This account has 1,311 games and is sitting in the bottom third of Tier 3. Another account I have has 2,304 games and is oscillating between Tier 3 and Tier 2. Another account I have reached Tier 3 in 156 games.
The more I re-read your post the more I wonder if you are getting into a tier above your skill level and then experiencing big losses and dropping down. The granularity of the tiers (especially with the "effective" tiers) can land people in a place where the game changes for them in the new tier and they get pushed back down into the prior tier only to rise back up again. This happens to me at the Tier 3/Tier 2 border.
Quote
"Refreshing" by resetting scores every so often actually only makes this issue worse, since now you're just creating even more players at the shallow end of the sampling pool.
The pilots that operate at higher tier levels will absolutely tear through the lower tiers (unless they are just faffing around or intentionally handicapping themselves or something). In many cases their outsized performance can skew the match score means/distribution such that players operating at that tier level will see bigger losses (or weaker gains) to their individual PSR scorings. That skew can also create really big gains for that higher tier player.
Quote
But I mean sure, PSR is always a reliable metric of a player's skill, even though the mathematics it's based on have been fundamentally modified several times, and shows a clear statistical bias favoring newer players towards the high end of the chart. Just because people who either don't understand or care how statistics are actually mathematically derived say so. And despite the questionable utility of some tracked metrics or sometimes straight up inability of some quite useful metrics to even *be* accurately tracked to begin with.
This paragraph feels like a general criticism of the PSR formula and match score. There are definitely some legitimate (in my opinion) complaints about match score and the PSR formula, however it really just attempts to measure and quantify a pilot's performance relative to everyone else in a match. It has flaws and edge cases; it is the system we have. Over time if that pilot tends to do better than average, they will move up in tiers until they do not, and then they will not.
Edited by dubstep albatross, 24 March 2024 - 08:04 PM.
1
