16 replies to this topic

[hyphen]

I was hovering at a #3 on TF2PR for Engineer, and at 4 Stars, but since today I've been pushed down to 2 Stars on Engineer. If a general chart of Star Ratings and what rating you have to be on TF2PR to get that amount of stars would be provided, that'd help everyone involved in Advanced Lobbies.

 

TF2PR: http://beta.tf2playe...561198097422673

 

TF2C Profile: http://tf2center.com...561198097422673

[Dagothur]

I was 4 stars on Soldier when ranked 250th on TF2PR with 0 stars in everything else. Then the 24th update came out (http://tf2center.com/changelog) and I was put down to 3 stars on Soldier and given 1 star in each of the other 8 classes. This is pretty weird because I've only played 2 lobbies that weren't on Soldier yet somehow now I'm qualified to play the other 8 classes despite having no experience or rankings.

 

It would be cool to have the algorithm calculating star rankings transparent. 

[hyphen]

I was 4 stars on Soldier when ranked 250th on TF2PR with 0 stars in everything else. Then the 24th update came out (http://tf2center.com/changelog) and I was put down to 3 stars on Soldier and given 1 star in each of the other 8 classes. This is pretty weird because I've only played 2 lobbies that weren't on Soldier yet somehow now I'm qualified to play the other 8 classes despite having no experience or rankings.

It would be cool to have the algorithm calculating star rankings transparent.

omg le tf2 philosopher replied to my thread

I think its because of UGC experience and stuff, which kinda doesnt make sense. Im a #3 on Engineer, but 2 Stars on Engineer, because I'm on an Iron team (I know, #3 tier Engineer on an Iron team, weird), not some random bad one, but the current top team. Really does not make sense that any Iron Engineer who qualifies for Advanced Lobbies can be rated the same as the top Iron Engineer who also qualifies for Advanced Lobbies.

The system is broken, players can easily roster ride for 4-5 stars, but could be worse than an Iron Engineer rated higher than them on TF2PR.

EDIT: Doesn't that mean players without a team will be rated 1-2 stars no matter where they are on TF2PR?

Edited by Fyre, 25 June 2015 - 01:53 PM.

[Khil]

Also, how are Medics ranked? It shouldn't be by deaths or kills like other classes. IMO it should be by heals and not heals per minute because the tf2playerrankings heals/min is inaccurate. Also, if it doesn't go by Ubers to drops, it should. Medics cannot prevent dying 90% of the time if there team isn't very good; Medics are not a class that kills people. While dropping uber isn't always their fault, it can most likely be prevented in any case.

[MasterNoob]

I can understand that you guys would like to know a little more about what's going on under the hood. Things may be subject to change, so that's why we're in beta currently.

 

Stars are currently based on league skill (div) and TPR (http://beta.tf2playerrankings.com/). The latter is optional and totally depends on whether you have a league profile anyway (ETF2L, UGC, ESEA (soon)). We're currently using 50% league experience (all classes) and 50% TPR (per class).

 

With the update of June 24, all UGC stats were imported, providing 50% of the stars. So this explains why it would go up from 0 to 3 stars.

 

We will play with these ratios during the beta until we find the best value; for now it's 50/50.

 

MN

[hyphen]

I can understand that you guys would like to know a little more about what's going on under the hood. Things may be subject to change, so that's why we're in beta currently.

 

Stars are currently based on league skill (div) and TPR (http://beta.tf2playerrankings.com/). The latter is optional and totally depends on whether you have a league profile anyway (ETF2L, UGC, ESEA (soon)). We're currently using 50% league experience (all classes) and 50% TPR (per class).

 

With the update of June 24, all UGC stats were imported, providing 50% of the stars. So this explains why it would go up from 0 to 3 stars.

 

We will play with these ratios during the beta until we find the best value; for now it's 50/50.

 

MN

So it is impossible for any Iron Player to get 5 stars? I mean, I shouldn't be in Iron anyway, but winning Iron seems fun to me, but having that 2 stars on Advanced Lobbies while being a #3 on TF2PR seems quite odd just because I'm in a League I really shouldn't be in. Oh, well, this team is going Steel next season for sure. I guess 5 star Engie will have to wait until fall.

[Dagothur]

Hi MN,

 

Thanks for your response. Is it possible for you to outline the stars awarded based on division, as well as the stars awarded for each percentile range on TPR for the current iteration of the algorithm?

 

E.G.

 

Iron -> 0 stars

Steel/Silver -> 1 star

Gold/Plat -> 2 stars

 

TPR percentiles:

 

1%     5 stars

2-5%  4 stars

6-10% 3 stars

11-15% 2 stars

16-20% 1 star

 

Then your total stars would be div+TPR...

 

e.g. a gold demo in the 13th percentile would be 4 stars. (Gold/Plat -> 2 stars + 11-15% 2 stars)

[hyphen]

Hi MN,

 

Thanks for your response. Is it possible for you to outline the stars awarded based on division, as well as the stars awarded for each percentile range on TPR for the current iteration of the algorithm?

 

E.G.

 

Iron -> 0 stars

Steel/Silver -> 1 star

Gold/Plat -> 2 stars

 

TPR percentiles:

 

1%     5 stars

2-5%  4 stars

6-10% 3 stars

11-15% 2 stars

16-20% 1 star

 

Then your total stars would be div+TPR...

 

e.g. a gold demo in the 13th percentile would be 4 stars. (Gold/Plat -> 2 stars + 11-15% 2 stars)

I agree with where you're going, but the percentiles should be a bit different in my opinion. Mainly the fact that Division should only add to your star count, rather than drop your star count. A roster riding Gold Scout might be worse than a Steel Scout (according to TF2PR?), but with the current star rating, the Gold Scout will have 1-2 more stars than the Steel Scout.

[Dagothur]

I wasn't suggesting what I think the algorithm should be. I was asking if he could define what it is, with an example. 

[MasterNoob]

Maybe I forgot to mention this, but the stars are merely an indication of skill. The TPR team balancer does not take any league experience into account, as a constant factor anyway. 

[hyphen]

Maybe I forgot to mention this, but the stars are merely an indication of skill. The TPR team balancer does not take any league experience into account, as a constant factor anyway.

If a higher rated Engineer on TF2PR plays in a lower division such as me, the league stats that were added on the 24th would make my star rating drop, showing I lack skill, which is the complete the opposite as shown on TF2PR.

I just think what League you are in should add to your star count, not remove to it. It gives a better idea of skill if league experience was only additive to your star count.

[Mother Tereza]

Iron -> 0 stars

Steel/Silver -> 1 star

Gold/Plat -> 2 stars

 

TPR percentiles:

 

1%     5 stars

2-5%  4 stars

6-10% 3 stars

11-15% 2 stars

16-20% 1 star

 

Then your total stars would be div+TPR...

 

e.g. a gold demo in the 13th percentile would be 4 stars. (Gold/Plat -> 2 stars + 11-15% 2 stars)

 

I like this idea. How about a negative stars for a particular percentages below 20%?

 

E.g.

 

20-50% - minus 1 star

50-99% - minus 2 stars

 

or maybe constant minus 1 star for everyone below 20%?

[Lolacio]

The feeling I get with the current system is that every player in high/prem has  4+ stars on every single class. Which hopefully will get tweaked a bit. But then on the other side you have open players with 4-5 stars on classes as well. And then they play like shit and don't talk.

 

I personally do like the idea of integrating TPR but not like this

 

Hi MN,

 

Thanks for your response. Is it possible for you to outline the stars awarded based on division, as well as the stars awarded for each percentile range on TPR for the current iteration of the algorithm?

 

E.G.

 

Iron -> 0 stars

Steel/Silver -> 1 star

Gold/Plat -> 2 stars

 

TPR percentiles:

 

1%     5 stars

2-5%  4 stars

6-10% 3 stars

11-15% 2 stars

16-20% 1 star

 

Then your total stars would be div+TPR...

 

e.g. a gold demo in the 13th percentile would be 4 stars. (Gold/Plat -> 2 stars + 11-15% 2 stars)

 

Actual experience like being in gold should earn you more stars then having a good ranking on TPR. Grinding out stats in normal centers is way too easy for people to get a good rank on it as it currently stands. If I can get in the top 5% on pyro just by using backburner then any iron/open player can get a good rank by trying or just by stacking with friends.

 

So I would propose something like this:

Iron: 0

Steel: 1

Silver: 2

Gold: 3

Platinum: 4

 

(Not sure if you can differentiate the bottom/top of open)

Bottom open: 0

High open: 1

Mid: 2

High: 3

Prem: 4

 

PLR

Top 10%: 1

11% - 20%: 0

21%-75%: -1

76%-100%: -2

Unranked: -2

 

High/Prem players would on their worst class still have at least 1-2 stars purely because they have the gamesence to accommodate for it. Open and Iron players can play advanced lobbies if they try hard to get a good TPR on that class. While Steel/Silver players can play the classes they have proven themselves on.

 

The main reason behind my proposal is because comms get way more crucial towards top teams where in lower teams they tend to be clustered more. Seeing that all games are going to be mumble required comms will be a crucial part. Definitely because most players haven't played with each other before. Currently I have the stars plugin made by kengur on and I see a good deal of open/unexperienced players in the games and these players tend to be completely silent and bad. Really bad...

[MEEP]

Hi MN,

 

Thanks for your response. Is it possible for you to outline the stars awarded based on division, as well as the stars awarded for each percentile range on TPR for the current iteration of the algorithm?

 

E.G.

 

Iron -> 0 stars

Steel/Silver -> 1 star

Gold/Plat -> 2 stars

 

TPR percentiles:

 

1%     5 stars

2-5%  4 stars

6-10% 3 stars

11-15% 2 stars

16-20% 1 star

 

Then your total stars would be div+TPR...

 

e.g. a gold demo in the 13th percentile would be 4 stars. (Gold/Plat -> 2 stars + 11-15% 2 stars)

I would have to disagree with this. I know someone who was in the top 1% as well as a gold player but still had 4 stars.

[Cortana]

Excuse me if I'm being rude, but some of your ratings are delirious.

http://beta.tf2playe...ource=tf2center- dude has a k/d of 1.2 as heavy, 3 stars as heavy according to your system

http://beta.tf2playe...ource=tf2center(me) - I play nothing but heavy, I have an okay k/d of 1.93, 3 stars as heavy according to your system

 

I'm not salty or anything, I couldn't care less about my personal image compared to other dudes, but you might want to review how people get stars for other classes than their main.

[hyphen]

Another case here:
Max! - 4 star Pyro http://tf2center.com...561197989627594
Gabiduo - 4 star Pyro http://tf2center.com...561198036965261

League experience, especially Highlander, which seems to give every plat player 3-4 stars on everything, should not be such a large factor in the system.

Uncle Dane - 834/Tier 6 on Engineer - http://beta.tf2playe...ource=tf2center- 3 star Engineer http://tf2center.com...561198057999536

Me/Fyre - 50/Tier 2 on Engineer - http://beta.tf2playe...561198097422673- 2 star Engineer http://tf2center.com...561198097422673

Also, Sigafoo is a 5 star sniper and bary isnt???????

System really needs to be reworked and make stars based off of skill/stats.

Although this might not matter as als are down and take so long to fill.

[Cortana]

I'd also like to say that after joining an iron team's roster, my rating has dropped from 4 stars to 2 on Heavy (the only class I play on tf2c). So one can perform great on one class (at least, I like to think I perform decently in most lobbies) but if they're in a low ranked team's roster they shouldn't have a high rating? It denies the efforts one goes through to better themselves. Lobbies are the only thing I can play due to my relatively bad location to the rest of the world, and I've learned a lot from them (also from watching povs and the likes of those), and I don't feel like my ratings deserve to go down the drain because of a team I'm in. Identically, I don't feel like someone who's in a high ranked team deserves high ratings on classes they don't even play often-- some people have 3 to 4 stars on classes they rarely play just because of their roster placement.

 

I'm not going to tell you how to do your job, as I couldn't begin to imagine how much effort you pour in your work as tf2c devs. However, I ask of you the following:

 

• Either rework the stars system so that it penalizes people less based on their roster, and rewards them more for their performance

• Or change the purpose of the stars rating. I know a lot of people who are very good in terms of DM and gamesense, yet their ratings are low because of their roster.

 

Thank you for reading. I hope the stars rating can be changed before ALs are re-integrated into the website.

Edited by Cortana, 23 July 2015 - 12:09 PM.