283 replies to this topic

[Nightbird]

OneTeamPlayer, on 06 July 2020 - 02:22 PM, said:

Aside, but with them admitting we struggle to make 1K weekdays and 1.5K weekends the only way those charts are correct is if they also place people who have barely played, as in everyone who has played 10 matches in 6 months or something silly like that.

The idea that we have 47 thousand-ish regular players is ridiculous on its face.

I'd be impressed if it were actually a quarter of that number.

Agreed that the numbers seem inflated, but relatively speaking T1 should be the smallest tier even if you cut all the numbers y 75%.

[Brauer]

Nightbird, on 06 July 2020 - 05:18 PM, said:

Agreed that the numbers seem inflated, but relatively speaking T1 should be the smallest tier even if you cut all the numbers y 75%.

I don't think that necessarily follows. Tier 4 and 5 are both very likely to include players who do not play much due to the XP bar nature of the old PSR system, AND they are the tiers most likely to include new players who tried the game briefly but didn't stick with it. Course it's hard to know how big of a impact that has.

[Nightbird]

edit: fixed graph thanks to Anamalocaris for finding an issue

Edited by Nightbird, 07 July 2020 - 07:24 AM.

[Anomalocaris]

Thanks so much for posting this data Nightbird. It's a very helpful visualization.

Question. In the 7/5 graphs I see the bars add up to about 99% of the players. In the 7/6 graphs I see 89.1%. Are we missing some players that fall outside the bounds of 0.3 to 2.0 WLR?

[Brauer]

Anomalocaris, on 07 July 2020 - 07:18 AM, said:

Thanks so much for posting this data Nightbird. It's a very helpful visualization.

Question. In the 7/5 graphs I see the bars add up to about 99% of the players. In the 7/6 graphs I see 89.1%. Are we missing some players that fall outside the bounds of 0.3 to 2.0 WLR?

There are over 500 players with a WLR currently over 2.0, which is actually far more than I expected. I don't know how that translates to a percentage of the population though, and some of those players may have under 20 games played.

[Nightbird]

Anomalocaris, on 07 July 2020 - 07:18 AM, said:

Thanks so much for posting this data Nightbird. It's a very helpful visualization.

Question. In the 7/5 graphs I see the bars add up to about 99% of the players. In the 7/6 graphs I see 89.1%. Are we missing some players that fall outside the bounds of 0.3 to 2.0 WLR?

Thanks for pointing this out, never hurts to have someone QC. I was tweaking code between the runs and make an error. It's fixed now. 0.3 and 2 includes players <0.3 and >2. I will add those symbols in eventually...

[Hellbringer]

i came to the realization that there will be few if any t4/t5 players with this current reset. All the players who belong in t4/t5 will likely quit way before reaching that tier. you will have the majority of the player base in tier 3 and the rest distributed to tier 2 and 1 respectively.

with the current psr values, losses dont drop people enough, and wins are given too much psr. Its easy to reach tier 1, but trying to get down to t4/t5 requires you to actively focus on not doing any damage.

[OneTeamPlayer]

Hellbringer, on 07 July 2020 - 08:08 AM, said:

i came to the realization that there will be few if any t4/t5 players with this current reset. All the players who belong in t4/t5 will likely quit way before reaching that tier. you will have the majority of the player base in tier 3 and the rest distributed to tier 2 and 1 respectively.

with the current psr values, losses dont drop people enough, and wins are given too much psr. Its easy to reach tier 1, but trying to get down to t4/t5 requires you to actively focus on not doing any damage.

We have no idea how easy it is to reach T1 or how hard it is to reach T5 because T5 people aren't showing their tier, and T1 people are more likely to self-select into talking about their tier.

Keep in mind that we had the view that most people eventually made T1 when the chart came out that there were a ton of tier 5 players.

Though that chart was suspect (and if you said it didn't count i'd be inclined to agree) there are at least a reasonable number of players who populate T4 and T5 yet they almost literally never proclaim that fact in any discussion.

Then again the culture of ridicule and cruelty towards other players is such that anyone who is Tier 4 or tier 5 would be wasting their time to even mention anything about any sort of system, enjoyment, or preference as when they do they would inevitable be set upon by "elites" whose sole job seems to be to convince people to never engage in this community or play this game again.

I will note this, however, the people who are willing to put up with the current sorting in matchmaker are the sort of people who are the most dedicated to playing this game.

There is next to no reason for someone who casually plays to suffer current gameplay for the dozens of matches it would take them to reach their proper rank, and for many if their proper rank is actually Tier 3 they have playing against Tier 1s for the forseeable future to look forward to, not an enticing proposition to some.

If players actually reached T5 i think they'd be likely to stay around, but then again from chatting around players who are reaching T1 aren't staying around so it's going to be an interesting trek towards August and September, for sure.

[Nightbird]

We're hitting 60 average games today so I can also share the % of people in each tier according to the simulation.

[Nightbird]

We've hit 60 games average today so I put together a brief comparison of the performance of new PSR today against the old PSR and also my prediction.

We're currently at 7k pilots today with >=20 games instead of the 10k in my simulation. As mentioned a few posts back, I'll use average games played to line up the predictions against the actual results (25k=60 games average, 50k=120, 100k=240, 200k=480). Here's the WLR graph today:



If we compare bar by bar, every result will be off by quite a lot, so I'm going to use the 0.8-1.2 zone to stabilize the numbers a bit. Let's call this the sweet zone. Above, we see 48.2% of the population in the sweet zone.


How does this compare to Season 47, the dialed in old PSR? There's the WLR graph:



In the sweet zone, we have 60.7% of the population in the old PSR system, so at least today we're at around 79% the quality of matches.




What about my prediction? I predicted 50.7% of the population in the sweetzone. Compared with 48.2% actual, I'm only off by 5% which IMO is pretty good. What do you think?


Of course, if the WLR system was implemented, we'd be looking at 67.2% of the population predicted to be in the sweet zone. Even if that prediction is off by 5%, that's still a pretty large margin of difference.

Let's revisit at average games = 120.

[Nightbird]

Nightbird, on 07 July 2020 - 08:32 AM, said:

We're hitting 60 average games today so I can also share the % of people in each tier according to the simulation.

Here's my tier distribution at 25k games simulated:


I don't think it's very accurate mostly because I think high tier players wanted to grind to T1 asap. (I know I did)

Here's my prediction at 200k games which I think is more reasonable:


Two words: business suicide

[MadcatX]

You're still at it eh?

[Anomalocaris]

Any chance you could pull results for season 45 for comparison? I have a feeling that the high WLR spike on the right side of the curve is really due to the merge more than anything else. Would be curious to see if that's the case.

[Nightbird]

Anomalocaris, on 07 July 2020 - 06:12 PM, said:

Any chance you could pull results for season 45 for comparison? I have a feeling that the high WLR spike on the right side of the curve is really due to the merge more than anything else. Would be curious to see if that's the case.

[Anomalocaris]

Thanks, looks like my theory was incorrect.

Although, TBF, one might need to put the games played bar a little higher as perusing the data page by page, it looks like there are a lot of guys who play until they get a couple losses and then quit to preserve very high WLRs.

Again, thanks for running the data.

[Sjorpha]

Anomalocaris, on 07 July 2020 - 07:19 PM, said:

Thanks, looks like my theory was incorrect.

I think one plausible change you don't see on the graph is that some small number of players who has moved towards playing mostly in strong groups after the merge now have absurdly high wlr compared to what they had when it was pure solo. It doesn't change the overall picture much because they are so few, and most of them probably had high wlr (north of 2.0) before the merge as well, but I think that if you looked at how many players now has a super high wlr (like north of 5.0 and so on) you should see a relatively large increase in the seasons after the merge.

[Nightbird]

Anomalocaris, on 07 July 2020 - 07:19 PM, said:

Thanks, looks like my theory was incorrect.

Although, TBF, one might need to put the games played bar a little higher as perusing the data page by page, it looks like there are a lot of guys who play until they get a couple losses and then quit to preserve very high WLRs.

Again, thanks for running the data.

We're still early in the process, 12.5% into the predictions so most players will play more games.

Sjorpha, on 08 July 2020 - 01:10 AM, said:

I think one plausible change you don't see on the graph is that some small number of players who has moved towards playing mostly in strong groups after the merge now have absurdly high wlr compared to what they had when it was pure solo. It doesn't change the overall picture much because they are so few, and most of them probably had high wlr (north of 2.0) before the merge as well, but I think that if you looked at how many players now has a super high wlr (like north of 5.0 and so on) you should see a relatively large increase in the seasons after the merge.

No doubt the merge did have an impact, it's just the elite groups don't represent a large portion of the population and might knock people out of the >2 WLR bar. If there was a stomps bar it would have shown a jump like Paul mentioned.

[Nightbird]

[martian]

Nightbird, on 09 July 2020 - 07:56 PM, said:

I scraped the leader-board data today and graphed it below ...

Thank you for posting.

[Anomalocaris]

Hmm, the sweet zone doesn't seem to be expanding.....