Nut Shot's Conqueror's Cup Player Rankings - UPDATED

[rockfist]

If you make it to the semi finals, much less the finals, you are playing someone good and if you don't bring your "A" game its not going to end well. Its much harder than an early round on average. That's my opinion on the matter.

[Extreme Ways]

If you make it to the semi finals, much less the finals, you are playing someone good and if you don't bring your "A" game its not going to end well. Its much harder than an early round on average. That's my opinion on the matter.

Yeah, this.

With all due respect, TOFU won first CC11 matchup 39-4. That should likely still be worth less than winning 31-24 in semis, or FALL 32-18 in quarters. Theres quite a bump there, but we have no training data to train parameters on.

Edit: thats relevant because it means I rate a 8-5 finals score a lot higher than 11-1 in the first round

[rockfist]

8-5 is a very good score in the finals.

[niMic]

Yep I agree. Honestly some finals I would be happy with an even score. And I'm sure there are finals or other late rounds where I didn't even win half the games, yet still felt better about my play than going 7-1 in round 1.

I guess one complication by increasing the worth of later rounds too much is that good players in mediocre clans won't be rewarded for taking games off TOFU in round 1, for example. But if he was going to account for everything it might turn into an even more massive job than it currently is.

[rockfist]

I'm never happy with an even score, but I've been below even in 2/4 finals We are 2/4 in finals I've played in and those happen to be the two team losses. This year I will do better. I promise.

[MTIceman41]

Agree there should be a “bump” for finals and semis…but I think there is just a general law of averages so to speak…last year we played ascension/ LHDD/ tofu…

Player A goes 6/10. 7/11. 8/20 Let’s say
Player B goes 4/10. 6/12. 11/18

Technically player B has higher ranking with the formula but same wins against 3 top clans to me it’s a numbers game…
That being said more points in beginning of the game vs at end when it’s crunch time I get that argument…

I would prob do slight bump for semis/finals but keep round 1and2 the same
Incrementally 5-10% increase each round might be skewed slightly…
If I did a formula I would weight more for higher map variety, teammate variety..
Regardless CC needs to have a NSS holiday for the work done here

[Nut Shot Scott]

If I did a formula I would weight more for higher map variety, teammate variety..

I tried teammate variety early on and the data collection was too much. I'd still be working on CL11.

Map variety is interesting. Hadn't thought of that.

Oh, and I would take a general contribution medal

[MTIceman41]

If I did a formula I would weight more for higher map variety, teammate variety..

I tried teammate variety early on and the data collection was too much. I'd still be working on CL11.

Map variety is interesting. Hadn't thought of that.

Oh, and I would take a general contribution medal

That’s something we can all agree on…maybe rockfist can do a nutshot to the ones in charge of medals

[Nut Shot Scott]

CCX has been completed. Top 25 are posted in the original message.

Here are the Top 25 per ROUND averages for CCX

Spoiler

1 - JPlo64 - FOED 2 - josko.ri - S & M 3 - Tin Trumpet - ATL 4 - kjg21 - S & M 5 - Donelladan - LHDD 6 - loutil - TOFU 7 - IanG7 - S & M 8 - Astoriana - S & M 9 - samdemars - LHDD 10 - rockfist - TOFU 11 - Doc_Brown - TOFU 12 - GeneralTaken - ATL 13 - misteryforall - ATL 14 - mc05025 - FALL 15 - queen victorious - ATL 16 - shoop76 - ATL 17 - t4mcr53s2 - FALL 18 - niMic - S & M 19 - _Untouchable - LHDD 20 - GoranZ - FALL 21 - MTIceman41 - S & M 22 - Phil Gates - ATL 23 - groovysmurf - FOED 24 - Heads or Tails - ATL 25 - morleyjoe - ATL

[rockfist]

Nice work thanks!

[Nut Shot Scott]

Moving slower than I had hoped. Get working CC9 in a couple days.

[rockfist]

CC9 was the best I ever played, but I do not think it will show up in the stats the way you compile them, which I'm totally cool with. We had so many absences and vacations during that year it was ridiculous. I think at one point I was sitting for four people who were normally major contributors all out for various reasons.

[niMic]

Great job.

[Nut Shot Scott]

CC9 was the best I ever played, but I do not think it will show up in the stats the way you compile them, which I'm totally cool with. We had so many absences and vacations during that year it was ridiculous. I think at one point I was sitting for four people who were normally major contributors all out for various reasons.

Lol yeah, I don't think I'll be going in to figure out who was sitting what games. Better chance of a map matrix.

[Extreme Ways]

Dave and I the only ones with a sub50 win percentage in the top 60. Get in!

[Donelladan]

What does Average per game and average per round means ?

Jplo's score is pretty amazing, 80% win rate 100% game played.

Seeing the score the avantage you gave for reaching the finals seems to make the comparison between player from clans that reached the finals and other impossible. Therefore I don't really see the point of such ranking, you could just save time and rank only players from the 2 clans reaching the finals.
Imho it would be more interesting to have ranking without differences between players reaching finals and player not reaching it.

[rockfist]

There is truth to that but, how would you account for someone who goes 6-1 in round one and stops playing versus someone who goes 7-6 in the finals? Surely 7-6 in the finals is more valuable and the games are more difficult as we both know.

[MTIceman41]

I think 3-5 years of data this gets “closer” in terms of correct rankings…obviously heavy S&M/TOFU but I think more Fall/foed/lhdd players get into top 25 once sample size increases

[Nut Shot Scott]

Fair points.

1. Average per game is the average points someone scores per game.

2. Average per round is meant to partially address your issue - it's the average score of each player per round played. So if you played 2 rounds for a total of 50, your average is 25.

3. It is an individual ranking within a team competition that doesn't have a uniform number of game - bonus or no bonus going into the next round, you're going to score high if your team wins and you play a lot of games.

4. The more data that comes in, the more things should come together. That said, if you play on a team that keeps winning year over year, and you keep winning, it's indicative of you being a good player, thus you would be ranked high.

[Donelladan]

I think 3-5 years of data this gets “closer” in terms of correct rankings…obviously heavy S&M/TOFU but I think more Fall/foed/lhdd players get into top 25 once sample size increases

I doubt it S&M et TOFU have been way more often in the finals, so more data means less chance for the other clans to be in the top 25, unless there has been a big turnover in TOFU/S&M roster and many players only have one final, then yes, maybe some players from other clans that joined every event in the cup will rise, but then it's within a total point ranking that awards longevity.

There is truth to that but, how would you account for someone who goes 6-1 in round one and stops playing versus someone who goes 7-6 in the finals? Surely 7-6 in the finals is more valuable and the games are more difficult as we both know.

Should we then give negative results to someone going 7-6 in the first round ?
Like we assume than in the first round your win % must be 75%, and give points according to your win ratio relatively to what isexpected. If your win ratio is 70% in the first round then you have negative factor applied to your rankin.
assume than in round 2 you should have 65% win rate, round 3, 55%, and round 4, 50%.
Kinda similar to the D400 idea, you must perform very well against weaker opponent to be able to rise. Average performance against strong opponent is on the other hand valuable

Btw is clan ranking taken into account in the formula, and what's the formula ?

1. Average per game is the average points someone scores per game.

And this is based on dub/trip/quad and which rounds you played in, smthg else ?

[rockfist]

I personally don’t value the first round at all. I went undefeated in the first round this year. I’d trade all of those wins for two more wins in the finals, they are of no consequence, but 7-6 would be a shit performance in the first round if I posted it for TOFU. However at the end of the day we if I were 7-6 in the first round unless TOFU loses, it only matters for pride.

So I think it depends on who you are and what clan you are in. I said in our war thread that Caymanmew’s performance against us in the quarter finals was excellent and I believe that even though it was 8-8 not 15-5 or something along those lines it was a strong performance against a superior opponent (based on the fact we won not conjecture). It’s rare that you are going to see anyone post a strong positive record over a large number of games on the losing side, but those performances are really really great when they happen.

If I had to give my thoughts on how this should end up, put Josko number one, put you (Don), me, maybe Fydraca, or JP in a hat and pull us out numbers 2-5, but there are many other good players. I’d have had myself number two but my performance sucked last year.

[Nut Shot Scott]

The reality is that average per round is probably more indicative of individual performances.

As for taking clan ranking into account - lol. If you want to go through and give me the D400 rank of each clan at the time of the corresponding Cup, I'll work it into the formula. But I'm not doing that. It's time consuming enough to pull the data and sort it into home/away/dubs/trips/quads by round.

My take on it is this: if you're a good player on a shitty team, you should have a solid per game or per round average when it's all said and done. Your totals might be low. If you're a shitty player on a good team, you're totals should be decent but averages will be low. The system isn't perfect but over time, the cream should rise, so to speak.

[Extreme Ways]

Hmm I'm 5/14 in the CC finals last year, pretty bad. I quite like the idea of 'negative' scores but in reality we're trying to finetune parameters on a small dataset manually, while our training data is also our test data. In CC10, I performed worse than 50% and below par taking into account the the result of the war. I shouldnt really get points for that, but adding that has its own issues...

No system will be perfect

[Nut Shot Scott]

I'm not sure results on a Cup by Cup basis are going to say much beyond the players at the top. I have looked at this as a more cumulative effort than League. So when it's all done, the picture will look a bit more in focus.

I am always tinkering with the ideas thrown out here btw. Some I run into time constraints with, others I have incorporated and some I haven't figured out quite how or if to fit them in yet. So keep them coming. I just can't guarantee they will make it in.

The one I'd really like to make work is map variety, I think it's doable. Strength of opponent I think is generally covered by adding a % to each round. It isn't perfect, certainly, but it is something. Because I absolutely agree that going 9-1 vs the 17th clan in round 1 is not as impactful as going 8-7 vs the 3rd ranked clan in the semifinals. The first round for some is a stat padder but also, that's the whole point. If you're better than someone, you're better than them, and the results should and do show that when you go 9-1.

[niMic]

The one I'd really like to make work is map variety

Oh no, that's not good for me