PROOF - You CAN rank down & lose SR with Positive Win Ratio

Please quote the myst cause I don’t see what you see, unless it’s something you wrote which I take a grain of salt.

Emphasis mine:

You probably shouldn’t quote things unless you’ve actually read and understand what’s being communicated.

3 Likes

I just read this second analogy. I had a snarky response…but you dont deserve snark. I genuinely enjoyed our back and forth, so I respect you.

With that stocks dont work the same as group project. Stick to that analogy, because 10 percenter would still get a (ex) 50 percent gain same as 45 percenter. Its not the same profit but it is the same profit margin. They win or lose the same amount of profit they put in.

And no competition I know works like that.

I think the stock example still works. Let’s put some numbers on it:

$100 total

You: $45
Friend 1: $45
Friend 2: $10

50% return = +$50

You: +$22.50
Friend 1: +$22.50
Friend 2: +$5

If this is SR, you and Friend 1 will rapidly gain more SR compared to Friend 2–meaning you two climb 4.5x faster.

2 Likes

Still now how do you PBSR in that example?

This is what it adds too. Your other analogy works better because you canPBSR. Doesnt mean I agree with you. All of them still win im the stock.

PBSR in this context: the average investment per person in other groups is 33% of the pot. They can expect around $16.50 on the return. However, you invest 12% more than them on average and so you get 12% more on the return.

Friend 2 gets 23% less than average back because they only invested 10%.

And they don’t necessarily all win. Experiencing losses due to market volatility might completely negate Friend 2’s +$5 but still leave you with a net gain of +$17.5.

I think your other argument is better. (Still and will disagree with it) but if you think this one is a keeper do it.

Ive sidetracked this long enough trying to keep a good debater debating.

Borderline insulting my intelligence is the last resort. When someone says most importantly… you want to emphasize what they said before?

Yes schedule and pbsr are factors but most importantly it should be the wins and losses that should affect skill rating.

Again 52% and is down 300… that’s a notch in the Mmr deviation with a winning record.

For shame!

I’ve repeated the same argument multiple times on this thread using numerous analogies and comparisons to other Elo systems. I have responded multiple times to the exact claim you keep making using the exact quote over and over.

I’m more than happy to keep repeating the same thing over and over, but don’t expect my charity when you’ve ignored the other responses.

Yes, gaining SR is contingent upon winning. You will never lose SR for winning a game. You will never gain SR for losing a game. Gaining more SR than you lose is contingent upon MMR, PBSR, and the confidence interval of your skill.

It does not say, “Win rate is the most important factor.” Whether or not you win given these factors is what determines your SR adjustment.

Additionally, “most important” does not mean “overpowers all other variables” nor does it mean “positive win rate is all you need.” The coefficient for “winning” might only account for 30% of the SR adjustment while the other 3 factors collectively account for 70%. However, 30% from a single factor is still the “largest” factor.

And, we know that every single Elo system out there requires winning against better opponents to climb.

This is at least the 4th time I’m responding this exact claim: no one in their right mind expects to stay at their career SR high. No GM peaking at 4.7k expects to be at or near 4.7k every single season. 4.7k is not their average SR. It’s a BS fantasy that you should disabuse yourself of.

OP has an average SR across 7 seasons of 1754. That means he was down 49 SR from his average. I’m not going to waste my time explaining this again.

2 Likes

Good because you didn’t convince me nor the community as the community is dying and the match maker and the rank system was one of the many issues.

“Guess my rank” and “Unranked to gm” are flags that something is not right. This story and many others should also clue you in.

But no… the same arguments must be repeated. I’ll see you in the next thread!

And I think this is where your argument fails the most, the expectation to climb has been lost, it was the expectation to maintain at a 52%. The argument has always been over exaggerating- “to climb”, GM in bronze, pro players vs hs teams… I think I agree with the op, the expectation was to just maintain!

You are losing more than you win on support. That’s all.

The “total” is irrelevant as that clearly includes games on other roles

3 Likes

Yes I know it’s still a direct measure of winning, but it’s still an unfair both players will be getting the SR for the win but only one player will be getting the extra Sr because his value was recorded. An amazing like a miniscule amount but that’s all the pbsr is a minuscule amount and it’s managed deduct about 1000SR from me so far. Furthermore I began recording my Sr gains and losses a little bit after I began this season so this begins at 1684 Sr instead of 1705

Games: 26
Wins: 15
Loss: 10
Draws: 1
SR Difference +114 1684-1797
Win Rate : 66%
SR Gained: 352
SR Lost: 238
Average SR Gain : 23.4
Average SR Loss 23.8

1684 +26
1710 +23
1733 +23
1756 +24
1780 +27
1807 +33
1840 +33
1873 -22
1851 -21
1830 Draw
1830 -23 leaver
1807 -24
1783 +21
1802 -29
1773 -24
1749 +23
1772 +19
1791 -21
1771 +21
1792 +20
1812 - 24
1788 -25
1763 +21
1784 +14 ??? wtf
1798 +24
1822 -25
1797

These statistics don’t make any sense to me. If PBSR is calculated by comparing me to the rest of the players within my rank, that means that according to this I am a below average player. But a bad player after a thousand games is not supposed to have a win ratio higher than the 50% average that goes against logic. Going by the logic of the win ratio I have always been a good player .

You could say this is due to playing higher or lower ranked teams but let’s suppose I did 6 games with 2 higher/same/lower opponents.
{2H 2S 2L) losing all my lower ranked games plus one same rank with a 50% win ratio means that my remainder of games would be winds against two higher rank and one same rank. this would counter balance the extra Sr I lost to the low ranked games an because I in turn one the higher ranked games

You could argue that due to my skill I did poorly in the higher ranked games I won and PBSR took the extra amount of SR that I would have gained. However the same should be true in that case for the games I lost below my skill level in that case should have prevented me from losing those points in a lower skill game because my performance should equally be better in a lower skill game as I am equally worse im a higher skill game.

So what exactly happened here? Any advantage I might have had or disadvantage I might have had will apply for the enemy team as well over time. Even the logic says that a good player will have higher than 50% win ratio and if PBSR is accurate, a good player will always have higher than average pbsr (greater than 50% of the players) a bad player will have lower than 50% win ratio and lower than average pbsr. So the question is how have I maintained lower than average pbsr and higher than average win ratios? For such a long period of time Every advantage and disadvantage I have would be available and true to everybody. It is counter to the logic for me to have a higher than average win ratio and lower than average pbsr and no other advantage can be said that I have than any other does not. this is why it must be pvsr that’s wrong. Accidental wins losses they will average, but continued miscalculation of pbsr because it’s unable to, will never average out average out.

I’m going to leave it there to get back on points I don’t know much about Chess or other topics and another am I trying to find but as far as I know here I presented every arguments and why it would become its redundant due to averages over time. Win ratio and pbsr cannot be saying the opposite things in the long run if both are accurate systems, win ratio only gets more accurate pbsr could potentially be forever flawed.

You are so patient. This topic has really highlighted to me how some people have brains that dont really work for them.

It is extremely glaring that you will NEVER be able to explain simple concepts to some individuals. Their brain will not allow them to learn no matter how many different directions or ways you explain the same thing. Ive never seen this play out so blatantly as this topic.

5 Likes

This is a take I support.

I dont need to be diamond and above. I have things to do outside of my hobby. The drop from gold was ridiculous…and the games getting harder with doorknobs of teammates after goodwins.

It isnt more difficult opponents that is the issue,. It is the worse teammates on top of the more difficult opponents.

1 Like

I mean, why play Brig tho, she’s not a carry hero. If you believe you deserve to climb you also need to have impact. And by that you just can’t play Brig and only be doing peeling. I suggest practicing Ana and Bap and learn to frag and duel dps heroes by f.ex learing match ups. 55% is not a lot if you want to climb, especially in such a low rank like 2k. You can easily climb to at least Plat by learning the absolut basics like Ult Economy, Hero Matchups, Peel, Staggering, Poking, Brawling, Diving, Rushing etc. No offense, but I bet you are just mindlessly queueing every day without actively thinking about what your actual role is, especially when it’s about the win condition.
I highly suggest watching vod’s from pro’s or vod reviews from other pro’s since they give you the absolute best quality content.

In that regard, good luck in your games - If I could have done it, you can do it too

1 Like

Yeah the wins and losses are the most weighted factors that affect skill rating. The PBSR is just a small adjustment of approximately -7 SR - +7 SR. What determines most of the SR you gain / lose is the predicted win probability.

The system will predict how likely each player in the match is going to win the game. (As far as I know, this is based on the SR)

Let’s say you queue with your friend. I picked numbers above 3000 to ignore PBSR:

  • You: 3230 SR (56.8% win probability)
  • Friend: 3030 SR (49.4% win probability)

You play against a team:

  • avg. 3054 SR

If you now win the game, you will obviously gain SR, but you don’t gain the same SR, since you both have different skill ratings:

  • You: +19 SR
  • Friend: +24 SR

The same if you would lose the game:

  • You: -28 SR
  • Friend: -23 SR

The SR and win probability mentioned above are fictional and are only used as an example. I didn’t calculate them. I just wanted to demonstrate how it happens, that you gain different SR and that it’s not only because of PBSR since in this example there was no PBSR included.

4 Likes

The standard deviation for your wins is 5 SR. For simplicity, let’s say 1 standard deviation above the mean is “above average.” Let’s +/- 10% for MMR differences like 60:40 odds.

Max MMR range for a win/loss : 23 +/- 10% = 20.7 to 25.3

So, if you beat a team that you were predicted to lose at 40:60, you should get ~25.3 SR. If you also performed 1 standard deviation above average, you would get +5 SR for a total of ~30 SR.

Anytime you see an SR gain greater than 25 or an SR loss of less than 21, it’s likely that you’re getting rewarded by PBSR for above average performance.

Anytime you see an SR loss greater than 25 or an SR gain of less than 21, you are likely getting taxed by PBSR for below average performance.

Total games that you performed below/average/above:
4-17-4

Wins that you performed below/average/above:
3-8-4

Losses that you performed below/average/above:
3-7-0

PBSR gained from above average: 19
PBSR lost from below average: 27

Net loss = 8 SR

So what’s happening? You perform below average in 30% of your losses and never perform above average. But when you win, you only perform above average in 26% of your wins and below average 20% of the time.

For your total games, you perform below average 24% of the time and above average 16% of the time. Roughly, you need a 58% (50%+24%-16%) win rate against equal opponents to climb.

How do you fix it?
Perform average or above average when you lose. Perform average or above average in games where you’re favored (roughly, if your SR > enemy team SR, you’re favored).

Edit: For a rough check of this math, if these performance trends represent your career average, then we expect you to lose 8 SR every 25 games. Interestingly, if we take your 869 comp games over 7 seasons and calculate how much you should have lost: 869 / 25 → 34.76 * -8 = -278.08 SR.

That’s a pretty close approximation to your initial complaint of being down 300 SR.

2 Likes

They have maintained an average 1754 SR for 7 seasons. Expectation met.

2 Likes

Well my goodness. It took 100 posts to get that point across. I dont think anyone was expecting to be in GM from gold at a 52% win percentage.

Nor was anyone expecting a GM to be stuck in Bronze from PBSR, or a Pro team to lose their title to a High school team. I truely believe the white knights see any blemish in the system and go from 0 to 100 full on “no rigged, no hardstuck” mode that its really the finite details most of everybody is talking about.

I’ll look at Overbuff too just to see whats going on.

2 Likes

It’s been the same message since the third post on this thread 13 days ago:

People here are claiming that the system makes them “hard stuck,” that a positive win rate should result in climbing, and that a player should expect to maintain their career high SR.

None of that is true and hopefully that has been illustrated by considering patently obvious examples.

3 Likes