AoE2 Tournament Points

[robo]

First seen here and here and analyzed here, I'd like to present a finalized version of the ATP (AoE2 Tournament Points) rankings, which are based on the ATP system used in professional Tennis and modified to try to suit Age of Empires 2.

Before I begin, huge thanks to @Tarsiz for initially creating the spreadsheet and kickstarting the idea, @Degaussed , @TheCapybara , @King_Boo , and the rest of the GB Discord for providing feedback and suggestions as new features were implemented, and @Adico for allowing me to shamelessly steal borrow ideas from the AoE IV ATR sheet.

• 1v1 ATP Sheet
• TG ATP Sheet (Note this is still WiP and some features from the 1v1 sheet have not been carried over yet)

Simply put, the ATP rankings are intended to be a way to track player tournament performance and provide an alternate method for seeding them in tournaments as accurately as possible, taking into account a wide range of criteria.
Players earn points based on the end position in a tournament (more points for first, less for 33-64th) weighted by the 'modifier' of the tournament which is impacted by the size of the prize pool, number of players, number of invited players, and other restrictions.
Currently, players do not earn any points if they do not complete all of their matches, that is, they withdraw or are otherwise disqualified from an event regardless of where they actually ended up, and any matches/games that were played are not counted towards their win/loss results.

The Team game sheet has a small delay in adding features compared to the 1v1 Rankings, as I want to make sure any changes are functional and completed before spending the time duplicating them onto another sheet. There is also significantly less data (54 completed 1v1 tournaments vs 14 completed tgs) so any inference from the rankings is dubious at best. I mostly created it to see what the numbers would look like. Players only earn points if they actually participate and play in games, being on the roster is not enough.

If anyone has any feedback, ideas or suggestions please do share them.

Features include:

• Almost all S, A, & B-Tier 1v1 tournaments since the release of DE are included. Over 200 players & 1300 tournament results included and counting!
• Tournaments are only counted for 1 year to represent their current form, but older data is still available.
• Match and Game win rate for most events within that time period. (some qualifiers are yet to be included)
• Various versions of the rankings (Regular, S-Tier only, all historical events, all historical S-Tier only, Country ranks)
• Difference/Change in ranking in the last 30 days to show how recent results have affected players. (recently added)
• Player search to see all of their results in the recorded tournaments. (You need to make a copy of the sheet to use this)
• Visual representation of top 20 players over time. Updated at the end of each month.

Ideas that are being thought about:

• Adding a decay algorithm.
  • Instead of points being worth 100% all year round, then 0 after the year is up, they would decrease slowly, picking up speed as the year progresses, to better represent their current form.
• Limited Rankings
  • Only include a limited number of results from A & B-Tiers to minimize players gaining lots of points in numerous smaller events vs weaker competition instead of performing against their peers.
  • The first attempt didn't have the desired effect and was almost identical to regular rankings, which may indicate that the idea isn't useful, or the implementation needs drastic improvement.
• Increasing the prize split bonuses (and negative) and adding more breakpoints.
  • This is to make the largest tournaments even more impactful and reduce the impact of smaller events, possibly in combination with the above Limited Rankings.
  • Maybe something like $100k+ = 1.2, $75k+ = 1.0, $50k+ = 0.8, $25k+ = 0.6, $15k+ = 0.4, $5k+ = 0.2 $5k- = -0.1, $1k- = -0.4, $500- = -0.8

(MbL falls out of the top20 after not playing in a tournament for the last 5 months)​

 

[PROject_Belgium]

Great job guys!

 

[Socksyy]

Legends!

 

[Adico]

If you want to implement decay, especially non-linear decay), here is the formula I use for the ATR:

points-(points*((TODAY()-Tournament end date)/365)^decay)

Points: is the flat amount of points that would be given out.
Tournament end date: is when the tournament is supposed to end, decay will start at that date.
Decay: is how strong you want the decay effect to be. for the ATR it's 1.8 resulting in this curve:

You can visualize different decays here https://www.desmos.com/calculator?lang=es

The full formula I have is:

=iferror(if(F2-(F2*((TODAY()-A2)/365)^1.8)<=0,"",F2-(F2*((TODAY()-A2)/365)^1.8)),F2)

 

[Elvaenor]

• Visual representation of top 20 players over time. Updated at the end of each month.

The thing I didn't know I needed in life! Really satisfying to look at!

 

[Lightysword]

Really nice. Did I see it correctly that S Tier Tournaments are the equivalent of Master Events, so winning them gives 1000 points?
I think it could be debated to give some tourneys Grand Slam status, in terms of revelance HiddenCup, NAC and KOTD come to mind there - and the Redbull Lan Event obviously. I think the distrinction in prestige and in part, pricepool would justify that.

 

[robo]

Really nice. Did I see it correctly that S Tier Tournaments are the equivalent of Master Events, so winning them gives 1000 points?
I think it could be debated to give some tourneys Grand Slam status, in terms of revelance HiddenCup, NAC and KOTD come to mind there - and the Redbull Lan Event obviously. I think the distrinction in prestige and in part, pricepool would justify that.

It's actually significantly more than that. The base point distribution is:

But every event has a modifier based on various criteria that increases or decreases the amount of points earned.

RBW5 for instance had a x2.0 modifier due to it's;
a) large prizepool +0.8
b) high number of participating players (inc qualifiers) +0.3
c) had a negative modifier due to some players being invited based on previous results -0.1
1.0+0.8+0.3-0.1 = x2.0 modifier.

You can read about the specific numbers and reasons why in the Details tab, and view what each tournament's modifier is in the Tourney Info tab.

 

[Breakfast]

This is fantastic stuff! Would it be possible to make Europe and America a little more distinct on the visual representation of the top 20 players for the colourblind among us?

 

[robo]

This is fantastic stuff! Would it be possible to make Europe and America a little more distinct on the visual representation of the top 20 players for the colourblind among us?

Updated, is this any better?

 

[Breakfast]

Updated, is this any better?
View attachment 200742

Much better, thanks!

 

[HappyLeaves]

Very good initiative.

 

[Tiger__Moth]

Great work !
I like that you can sort by country..... 'oven ready' for a nation's cup

 

[Musashi]

Looks amazing!

 

[The Bloodless]

I like this ranking because it puts me in the top 50.

 

[USC_kiky]

Great work!

To improve the visibility of this ranking system, I think it would be better if we could have an annual event designated for the top 4, 8, 16, 32 or 50 (just to include Bloodless ofc) players in this ranking

 

[Rustyiesty]

Idea for TG points - Last two years instead of last one year?

 

[vblytmp]

Great work!

I'm curious about how things look if we use most recent n tournaments, instead of a fixed cut-off of 1-year. Motivation is to normalize for variation in frequency of tournaments over time (in tennis, there are a relatively stable number of grand slams, so using a time-based cutoff achieves the same thing).

Only include a limited number of results from A & B-Tiers to minimize players gaining lots of points in numerous smaller events vs weaker competition instead of performing against their peers.

In addition, what I have in mind here is to use most recent n S-Tier tournaments, and most recent m A-tier tournaments etc at any time. This could help address the above issue in a more natural manner.

Adding a decay algorithm.

+1. Most-recent can be combined with a decay as well, and think it might give better results.

Disclaimer: this might have been proposed earlier / stray thought, so might have other known issues

 

[robo]

Idea for TG points - Last two years instead of last one year?

I think there is still the issue of not being enough events, 2 years of TGs is 13 events, and 1 year of 1v1s is 31.
And there also aren't enough smaller events for other players to grind their way up and make a name for themselves.

 

[robo]

I'm curious about how things look if we use most recent n tournaments, instead of a fixed cut-off of 1-year. Motivation is to normalize for variation in frequency of tournaments over time (in tennis, there are a relatively stable number of grand slams, so using a time-based cutoff achieves the same thing).

In addition, what I have in mind here is to use most recent n S-Tier tournaments, and most recent m A-tier tournaments etc at any time. This could help address the above issue in a more natural manner.

Seems like a reasonable idea

 

[USC_kiky]

" the 'modifier' of the tournament which is impacted by the size of the prize pool, number of players, number of invited players, and other restrictions. "
Not sure if this has been mentioned/considered, but I feel that another factor that should not be ignored when determining the modifier is how many top players signed up for the events.

A $10k events that none of Viper, Lierey, Hera, Yo, Daut or Jordan signed up may not deserve a higher modifier than another $1k event that have all of these top players, though it's possible that there are more players signed up for the $10k event than the $1k event.

A metric that captures how competitive / prestigious a tournament is, in my mind, can be the average ATP ranking (as of right before the tournament) of the top 8 contestants that signed up for the tournament. Clearly, when all the 8 best players in the world signed up for the tournament, this metric is 4.5. But as of now, if the top 8 contestants that sign up for a new tournament are Vinchester (3), Dogao (12), Nico(13), Bact(14), Miguel(15), ClassicPro(17), Barles (18) and Slam (19), then this metric would be (3+12+13+14+15+17+18+19)/8 = 13.875. Obviously, the smaller this avg ATP ranking of top 8 contestants is, the more competitive this tournament would be.

This metric can adjust the modifier by accounting for scenarios when some top players, e.g. Viper, Hera, Lierey, Yo, can't take part in a tournament due to conflict of schedules or VISA issues. It's always easier to win a tournament when none of the top 8 players show up

 

[nimanoe]

Great work!
I think the ranking is fairly accurate, Liereyy and MbL are obviously rated too low, but that's because they haven't been active lately, not really a fault of the system.

I really like the suggestion from @USC_kiky above to look at how many top players played in the tournament. I think a small flaw of the current system is that it gives significantly more points to the top players in Visible Cup than the players who qualified for Hidden Cup, which is kind of weird. For example Nicov only got 160 points from Hidden Cup 4, but saymyname got 400 points for getting 3rd in Visible Cup 4 and 40 points for the Hidden Cup 4 qualifier.

If you don't go with the solution from Kiky, maybe you could have a negative multiplier for tournaments that have restricted who can sign up.

 

[robo]

If you don't go with the solution from Kiky, maybe you could have a negative multiplier for tournaments that have restricted who can sign up.

At the moment they already have the 'Other Restriction' negative modifier (-0.4), but it isn't enough to overcome the benefits of the events (high participants +0.3, no invites +0.1), but I didn't really consider the impact of the event alongside/compared to HC4 which is a valid point. Probably will add the elo -0.4 soonTM.

Maybe in addition to, or instead of @vblytmp suggestion, each tier of event could decay at different rates.
Eg. S-Tier events decay very slowly for the first 6 months (only losing 10-20% of their points), then increasing to maybe 50% by the end of the year, while A-Tier events decay 30-40% in the first 6 months, increasing to 100% at the end of the year, and B-Tier is even faster, and maybe they stop contributing before the year is even over to further limit their impact.

 

[TheCapybara]

I'd personally be vehemently against the idea of modifying value retroactively based on participants. It's one of the things I strongly dislike about Liquipedia's notability criteria. I think it's far preferable to judge a tournament on its settings rather than which players choose to play it, just like the French Open is still the French Open regardless of whether Djokovic and Nadal are fit to play it or not.

 

[willdbeast]

I'd personally be vehemently against the idea of modifying value retroactively based on participants. It's one of the things I strongly dislike about Liquipedia's notability criteria. I think it's far preferable to judge a tournament on its settings rather than which players choose to play it, just like the French Open is still the French Open regardless of whether Djokovic and Nadal are fit to play it or not.

Depends what exactly you are trying to measure
If you're measuring achievement then a win is a win
If you're measuring skill then the cohort of opponents you were against is surely relevant

 

[TheCapybara]

Depends what exactly you are trying to measure
If you're measuring achievement then a win is a win
If you're measuring skill then the cohort of opponents you were against is surely relevant

How would you start determining which opponents merit a positive impact on modifiers and which lower it? I know there's the suggestion from kiky above, but that then becomes self-deterministic (i.e. a higher ranked player enters an event and therefore makes that event worth more because of their own presence in it).