TWDT Star Caps: Discussion

[qan]

New name is a decent low-tech solution. Could be standardized by using the same name + a tag such as (spider).

[ogron]

The games have been very competitive, people just love to complain about the 3 people out of 200 whose rating is 1* off.

There's enough granularity among veterans for all of us to conclude that adding .5 would be beneficial and make it a little bit better. I can name a bunch of people who are too good at 8 and will get benched at 9.

But there is no problem with the star cap system in terms of its effectiveness, the past four seasons have been good. It's more of a question of if people wanna go with a star cap system in the first place.

I'll take people with 20,000 hours eyeballing ratings over Euler's curve. The vast majority of the ratings are accurate and within .5 of the player every season.

There are 10 nerds in chat debating right now about whether turban is an 8 or 9 in warbird and analyzing his stats, how he's "better against good players in LD than actually carrying" and "being an elite support player doesn't make you a 9" and how he "isn't great at hunting low stars in TWDT-D" and doesn't "put up big numbers", basically that his skillset translates better to TWLD than TWDT-D, and the stats and his success level and testimonials bear that out, and how ro is a much better TWDT-D type player despite turban outperforming him in LD and how ro should be a 9 and turb should be an 8. If you wanted to add granularity you would probably put turb at an 8.5, but without .5s he's prob an 8.

These forums are populated by players have no idea how much actual work goes into ratings most seasons, and how accurate they are on the whole. I've had nothing to do with ratings for years and I think they've generally been good and produced strong seasons.

[Poseidon]

as i mentioned at the time, the ratings last season were p suspect. I referenced zidanes team above, which won the title last TWDT by taking advantage of the 6 understarred spiders on the roster; unsurprisingly, last TWDT, Ogron was on this team.

they win the championship and now u got the guy on here really saying "people just love to complain about the 3 people out of 200 whose rating is 1* off."

[Cape]

Hello nerds in chat, please give me 7.5 / 8.5 / 9 . Ty.

[qan]

I'll take people with 20,000 hours eyeballing ratings over Euler's curve. The vast majority of the ratings are accurate and within .5 of the player every season.

You've missed the point here, and if even you have missed it, I'm sure everyone else has as well. Nothing needs to change with ratings or how they are done. How those ratings contribute to the makeup of a team is what's at issue.

It's far too late to think about changing anything, but this is something to be worked on.

[zidane]

Poseidon the ratings last season were done by averaging out 6 peoples ratings, .5s were all rounded down. We then put the ratings on forums and let the public talk about them. Any subsequent changes that were made to ratings were then edited and clearly shown in the OP.

I took advantage by drafting old returning vets with high draft picks (sarger, mouse) whose current rating would be underrated based on their prime. It was a gamble and it worked out, however it had nothing to do with how ratings were done. This stuff has always happened and it's up to the captain to take that risk. For example weak just signed up, he hasn't played in 6 years.. if he ends up being as good as he was before he will almost certainly be underrated. There is nothing sinister about this, we cant predict the future and there has to be an aspect of rewarding good captaining or taking risks.

[Nipple Nibbler]

I agree with Claushouse that the manual ratings are pretty damn close and adding .5's should solve most problems with the current rating system.

qan That math thing you posted got me thinking, I am only 8/9* spider and I can take out two 6* spiders one handed. I have seen 10* Warbirds take on 2-3 opponents that are lower * and honestly it's a joke to watch. Maybe we just need larger gaps between the skill differences. Maybe instead of having 6-7-8-9-10* we could try skipping numbers something like (this is an example no math behind these numbers) 4-6-8-10-12? a 10* really should be able to carry, and I know for example Lasenza easily worth at least 3 of me in warbird and Zidane probably the same for Javelin. Having two 10* in a DT-D lineup back on Demacia was hilarious to watch and if I am to be honest should probably not be allowed for DT-D or DT-J with *caps.

[Nipple Nibbler]

Poseidon If I really think about it there are probably only 5-10 players that truly play all 3 ships, it would not be too hard to come up with ratings for us.

[ogron]

the only thing i would agree with poseidon on is that averaging out people's ratings is dumb, and produces the worst ratings.

i let 10 people do the ratings and averaged them out because i was tired of the complaining one season, and it was a huge mistake. literally the only upside is you get to shield yourself from criticism by saying "they were averaged from 8-10 people, they're not my ratings". that's it. people have different standards of what a 7* player is, and just as importantly, if you actually read all 10 player's ratings you'd realize how absolutely clueless half the people in this zone are when it comes to judging talent.

[beam]

I think you are missing the part where by adding a 10 star like those guys you say are worth 3 of you or whatever, you are also forcing yourself to add a 6 or have an 8 as your highest teammate, which is a pretty big liability, instead of adding a more balanced lineup of say all 8s or 9/9/8/7/7 for example. I feel a lot of 8s and 9s are capable of pulling off 1v2s vs a combination of a 6 and a 7, but I mean that's why they are rated higher they are supposed to be better. 6 and 7 star players are meant to be either newer players or people playing a ship that isn't their main and they should be a handicap to the team to detract from how great the 8s, 9s, and especially the 10s are. All the issues you guys have made a lot of sense when we were forcing mandatory star slot lineups but with an overall star cap it just doesn't make sense to me to say a 10 is too good and a 6 is too bad when to play the 10 you also have to play the 6.

[Poseidon]

ogron: so which did u think is the least understarred out of zids line last szn? Rab 8* Mouse 8* Hellkite 9* Sarger 9* i 4get the rest of em p sure u had 2 more understarred spid.
then u have kassius and azuline getting 6*. could go on for a while if i bothered to look it up
lol zid those people had been BDing for months before the draft

the ratings last szn were p bad and it was obvious zidane was far better able to take advantage of the situation than the other captains. its not zid's fault, hes just being a smart drafter - but he understands the game better than most captains so when u fuk up on the Star Ratings ppl like zid and dreamwin are the ones who really benefit. I called this out and the ratings out before the draft even happened. so did others such as nowon.

luckily if we do .5 ratings, it will mitigate the ability of some captains to take better advantage of misstarred players.


-----~----~----~-----~-----~----~--~----~----~---~-----~----~--~


Last TWDT I proposed forming a triumvirate of 3 trusted and knowledgeable peeps to rate, one group per league, and i still think itd be a good idea. They'd make the ratings as a group, not by taking an average (this is i believe what ogron is referencing above), but by consensus, or 2-1 vote if it came down to it. They'd just need to get together for a few hours w the list and go through it together... 90% they'd probably agree on w/o discussion, and any discussions themselves may help the raters to more fully consider the player and apply a more accurate rating than if all the ideas of that person were from their own memory and imagination and bias.

it should be easy enough to come up with three people everyone can feel comfortable with assigning this task

[qan]

qan That math thing you posted got me thinking, I am only 8/9* spider and I can take out two 6* spiders one handed. I have seen 10* Warbirds take on 2-3 opponents that are lower * and honestly it's a joke to watch. Maybe we just need larger gaps between the skill differences. Maybe instead of having 6-7-8-9-10* we could try skipping numbers something like (this is an example no math behind these numbers) 4-6-8-10-12? a 10* really should be able to carry, and I know for example Lasenza easily worth at least 3 of me in warbird and Zidane probably the same for Javelin. Having two 10* in a DT-D lineup back on Demacia was hilarious to watch and if I am to be honest should probably not be allowed for DT-D or DT-J with *caps.

Yep, this is the point of creating a formula that maps star levels to actual skill levels. You don't need to change the stars themselves. Just have a formula that translates what it means to be a 10 star vs a 6. Right now we use a linear formula. If the linear formula were accurate, two 5 star players should equal one 10 star, so that approximately 50% of the time in a 2v1, the two 5 star players would win. This is absolutely not the case right now, and demonstrates why using a linear formula doesn't work.

Translation to an actual skill level via formula, yet still keeping star values exactly as they are, is all that needs to be done. The translated skill level is then what would be used for calculating "star" caps (which would actually be a "translated skill value" cap, or whatever you'd want to call it).

It wouldn't even need to be a formula; it could be a simple lookup table. For example, a 10 star is worth a skill value of 20, a 9 star is worth 15, etc. (Just throwing those out as an example. Those are not actual numbers.) Someone just needs to figure out the translation. Requires some thought, experimentation, and a small bit of math. Then swapping out an 8 star for a 6 star would have roughly the same impact as swapping out a 10 star for an 8 star (for example). Right now those do not have the same impact. This would also allow ratings lower than 6 star.

[Rab]

Ah now I see what you're saying.

The way I see the ratings atm is a person is worth 2 people 2 stars lower than them. 1x10=2x8, 1x9=2x7, etc.

10* = 4.0
9* = 3.0
8* = 2.0
7* = 1.5
6* = 1.0

I think this is a useful way to think about it, even if people don't agree exactly.

edit: and the way you know this is with questions like:
- if you can pick 1x10* or 2x9* which would you pick?
- if you can pick 1x10* or 2x8* which would you pick?
- if you can pick 1x10* or 2x7* which would you pick?
etc.

[ogron]

i just don't agree with that, the best player in warbird isn't Faker playing Zed and dunking people with complete mastery of all his QWER, flash, item prioritization, CS'ing, and outleveling you to boot creating an even bigger advantage over time.

it's a ship with 2 buttons.

the curve between 10s, 9s, 8s, and 7s is fairly linear. the only non-linear jump is between 6s and the field, where they can be so atrocious and outclassed you can sometimes say there is an extremely non-linear difference between a 6 and a 7. you're also not FORCED to use a 6, and most lines eschew it.

there's also a high level of variance in performance from the SAME player. I've seen racka go 200-100 in TWDT-D and 81-80. mood, interest, practice, luck, and synergy between teammates all factor in.

exponential curves have little use here. there is no problem, the star cap system works really well and all the games are competitive.

[Nipple Nibbler]

Claushouse We can't all be clueless, I submitted some basing ratings along with markmrw DT or two ago, and honestly most of them ended up being what people were rated.