No the odds are that it is not useless, it will be very inaccurate from time to time but way more often than not it will be somewhat accurate at least 85% of the time
you seem very good at maths so teach me how to derive a win rate based on 6 games out of 252, which were all almost certainly uploaded by someone above average competitive
It is gonna take a while cause im not very fast but i do know how to do it so give me a bit
what are you going to calculate? there is no data to go from. are you missing my point or are you just bored
why you sound so scared, you asked me a question now let me complete it it is gonna take a while so just sit ight
Parser thread:
5% posts about Parsers
95% posts trash talking other posters
5 Likes
to be fair there isnt that much to talk about regarding parsers
1 Like
odds of 6 straight wins - %1.5625
odds of 5 or greater wins= %10.938
odds of 4 or greater wins = %34.375
odds of 3 or more wins = %65.63
odds of 2 or more wins = %89.06
odds of 1 or more wins = %98.44
My original estimate was that i wouldnāt even shed any weight on a 6 battle sample size unless it has either 5 or 6 wins or 5 or 6 losses with least an 85% degree of accuracy for the 5/6 number(it is actually 91% after taking the time to do the math and the 6/6 is 98.5% accurate). Now lets suspend disbelief for a moment and lets compare getting a win or a loss to simply flipping a coin where heads equals a win and tails equals a loss then the odds of getting 6 straight wins is %1.5625. If i seen 6 straight wins then the 6 sample size actually got a decent amount of weight that way soo more often than not you suck with 5 straight losses and the odds drastically shoot up that you suck with 6 straight losses.
So basically when i saw that you had 5 losses out of 6 battles meant that there was a 90% probability that you suck which is not that bad odds at being right(i originally estimated 85% so not to far off). Now ill be back later cause the 252 one is gonna be next to impossible has the math will be way to complex but ill try but you get the idea.
- flipping a coin on the same side 6 times in a row has a 1.5% chance. so with your gigantic sample size of 6, there is a 1.5% chance you predict that flipping a coin has a 100% probability to fall on a certain side. sounds like a great way to calculate winrates.
- the actual data is nothing like flipping a coin, because only competitive players will upload, which means the less games i play the less chance there is that i play with a player to uploads their log and my team wins.
your calculation is useless
I told you to suspend disbelief for a monent cause yes it is more complicated than that due to unseen variables such as persons skill, and all kinds of stuff that you will never be able to calculate but the luck element of the game still remains which do add weight. so that 1.5% still adds weight to the equation but how much weight i donāt know but it do most definitely increase the odds that you suck when you either got 5 or 6 losses out of a 6 game sample because a luck element do in fact exist just like flipping a coin.
.
one variable which is seen and entirely dictates what the data says is the players who upload unverified combat logs. i could literally just inflate my damage numbers the data gets processed normally. not that most players do this.
on a russian stat tracking website that shows top players, im confident in saying that 90% of everyone who uploads are sweats. which means this data says nothing about 99% of all games i actually play
yea and thats why you got to add a little weighting like i said earlier, add or subtract 2% problem solved. The other 10% is not sweats and are average skill. Also even good players lose alot. Take userlol for example he is at the up most extreme of the outliers so if all your data came solely from user lol who got a 71% win percentage then you would add even more to your percentages and im talking greater than 10% at least but most of the players uploading donāt have win rates that high they are more slightly above 50%but win rates above 60% are pretty rare even among the sweats so you donāt have to weight it all that much actually just a couple% maximum.
i dont know what that means
will the lowest loss rate you can possibly have is 38% in pvp and maybe 30% for 4 max battles so the other end of the extreme is highly skilled players with 70-80% win rate but those are few and far between and im talking like 1 in 500 players if not more few and far between. Majority of good players will have win rate of around 55%ish that like to upload their logs so tack on another 2% for the extreme outliers such as user lol then you can add 2.5% to your percentages and that is most likely your actual skill level.
If you got a 44% winrate then your actual win rate is prob around 46 to 47% across the entire board
Imagine a player like userlol who is in the top .1% of players losing to randoms and having 6 straight losses. The odds of that are next to none unless he wasnāt trying so then i got to determine the odds that he wasnāt trying which i canāt do but i can safely assume that the odds are very low so if i seen 6 straight losses from him then i would chalk it up to him not trying which still doesnāt make him a bad player. If i seen 2 straight losses from him i would chalk it up to bad luck because that is most likely the highest gap for his losing streaks but you are trying which is the difference and having 6 straight losses are more in line the average player so i can definitely rule out that you are most definitely not in the top 1% of players. So basically having 5 or 6 straight losses can narrow you out of the top 1% of players with super extremely high probability which makes you average or a little above average at best.
So to simplify it me seeing 5 losses out of 6 ruled you out from the elites but donāt mean that you suck however but it do mean that you most likely no better than above average just from that microscopic puny sample alone
Edit: So me being able to extract that much info just from 6 games alone then 100 games is a slam dunk in making a judgment call on just how good a player is.
Edit 2: I updated this reply with a diagram showing the class of players that you most likely definitely donāt belong in from those 6 games just simply due to the mega losing streaks that top 2% to 3% of players rarely experience when playing reg pvp(clan wars is different however because you got the best fighting the best)
the difference between an average mission player and somebody who uploads cw logs in winrates is minimum 10%. most cw players dont upload their logs to a russian stat tracker.
that wont happen because he is more skilled, probably plays with a real team, and has fused relics.
well you couldve just asked or checked my exhibition, its pretty clear im not anywhere near the āeliteā in cw. the average player who uploads logs will definitely be closer, hence only losses recorded.
the other 100 times i couldve played against an average / ānon-eliteā team and won easily will never be recorded.
if you had info on 100 games i wouldnt question the data. 6 is nothing though
I know one personally who go a 52% win rate not true at all, theres more than just slightly above average people uploading then you think and i could tell by the drastic increase in uploads from 2024 compared to 2023. 2023 might of been from all elites but more just slightly above average players are uploading these days. They are still on the good side donāt get me wrong but the top 0.1% elites only normally play solely cw and they donāt have much time for reg pvp it donāt interest them. Just go watch a xberlinx video it is 95% clan wars with some reg pvp mixed in but not very often.
6 is enough to rule you out of the top 1% at the top end of the bell curve with a very high degree of probability but only with either 5 or 6 wins or 5 or 6 losses is when you can add any sort of weight to a sample size that small.
(I realize the topic has changed but I found that gif too funny to not post)
1 Like
Good idea.
Iāll try a Torrero build with a spread minimizing codriver and Finwhale, and see if it feels significantly different.
Iāll also try an oppressor/buggy wheel build with Jay, to see if projectile speed is the issue.
I want to figure this out, even if the answer turns out to be that I suck at aiming.