Armada Player Rankings!

By shmitty, in Star Wars: Armada

I've added a new page over at SteelStrategy on Armada Player Rankings!

It's based heavily, blatantly stolen , from the ATP Tennis ranking system.

It is planned to be community driven, totally voluntary, and strictly for fun. To opt in, check out the rankings page and fill out the form linked there: http://www.steelstrategy.com/p/armada-player-rankings.htm

I currently have some of the newer Primes as well as Gencon and Nova in the data. If you have results from Worlds, Euros, or other Grand Championships I would love to get those in as well.

Have fun

I think I like the idea but I heavily dislike the particulars of the ranking system.

Quote

players are awarded ranking points based on both the tier of tournament that they played and their final ranking in that event.

sounds to me like a metric that mostly rewards access to tournaments rather than player skill, which I assume is what it is actually supposed so model.

A player that can, due to their geographic location, only attend one or two tournaments that q ualify for your APR system could easily score much lower than a player of worse skill that simply has access to more qualifying tournaments.

So I think it is inherently impossible for a system that awards additional points for every tournament you visit to be actually representative of player skill.

Btw. location is not the only problematic factor here, just the most prominent. This system would also punish you if you miss a tournament because you were lets say sick. If you take two players of equal skill that always visit the same tournaments but one time one of them gets sick all of a sudden your system would claim that he is now worse than the other player even though that is not at all the case.

The reason why I would guess that this system would work for professional tennis but not Armada is that in tennis you have literal professionals who make a living from playing (I assume. I know absolutely nothing about professional tennis) and can afford to travel to those tournaments so you can make the save assumption that everybody would roughly attend the same amount of tournaments whereas you can absolutely not make this assumption in armada, which is why this system would produce extremely misrepresentative results.

Long story short, I think this ranking system is terribly unsuited for armada.

Edited by LordCola

Yeah! How dare someone do something for the fun of it.

The NERVE.

Tier 0 Classic Ben

Maybe use a W/L ratio as the primary factor, then total games played, with a minimum number to qualify?

@LordCola has a point.

Anything else shoots reporting requirements through the roof, as well as infrastructure and anti-spam requirements. No the system isn’t great, but it’s not meant to be some Worlds-entry-granting mechanic either. It’s for fun. Definitely don’t take it seriously.

5 hours ago, The Jabbawookie said:

Maybe use a W/L ratio as the primary factor, then total games played, with a minimum number to qualify?

@LordCola has a point.

He is right. It is not measuring the objective skill of thr players, because it's impossible. It shows the success of the players in the differently challenging environments. It's still relevant.

While I do understand your criticisms of this system, it is not meant to a relative marker of player skill. It is meant to be fun and for those that do like attending lots of tournaments a way of tracking progress, etc.

I actually looked at many different ranking systems like ELO before settling on this one. My general feeling that was while ELO would be a great fit for a local Armada group to do a ranking ladder it didn't work as well for ranking players that would likely never actually play each other.

This is by no means a perfect system. It has many warts and failings, but it has the potential to be a great deal of fun for those that want to participate. There is nothing official about it and players will only appear if they opt in. It is also immensely practical in that it is easy to setup and administer. I didn't want "perfect to be the enemy of the good" in this case.

All systems for rankings have their flaw.

For FFG games it is really hard to get the data for it.
Basically it would be really great when FFG would have a good OP team, to do this themself. Other companies have done this already years ago. It should not be hard to do something in these days. It just cost some money, with no direct profit. This is why companies are not doing it. The indirect profit is not mensurable.

There is a really nice variant for ranking on the T3 . But it require alwasy the full data of all tournaments. So nothing that can be done easy.


Every wednesday morning (at 5:03 o'clock) the recomputation of the NTRs is started. The following algorithmic is used:
1. Depending on the tournament size the NTR-points for each acquired tournament are computed (tournament points).
2. Depending on the age of the tournament the influence of the tournament is computed. Each 150 days a tournament looses 25% of its value.
3. Every player now gets his/her points through the following formula for each tournament: PlayersPoints = (TournamentPoints / (Number of Participants - 1)) * (Number of Participants - Placement of Player) * TimeFactor
4. For a good placement you'll get a bonus to your points.
5. For each player his/her 5 highest scores (as in 3+4) are taken, sorted descending and valued at a rate of 24%/22%/20%/18%/16% (total 100%). The sum of the 5 valued scores is the number of NTR-points for the player.

It only use the participants (not tier) and only take the 5 highest scores. With a time decay as well.
Their system does not favor players who go on many tournaments, but these who are successful on big ones.

But without results from worlds, Euros, EU Grands, this list will be really limited 😊 😉

But when it helps your, i have these:

Euro Continental Standings after day 1. Sadly i don't have a picture of the day 2.

y4mV-UoFQ0Kono0H9U0Tg_TK5-YCFSIy2ltiOLlI

In the end with points in ():
1. Dennis (23)
2. Piotr S. (18)
3. Daniel (18)
4. Premyslav (18)
5. Calum (16)
6. Alessio (15)
7. Thomas (14)
8. Piotr P. (?)

For worlds i have to look through my pictures. I might have a standing after the last round as well.

But:

They had the full lists, and maybe as well the standing in the end.
@Karneck should as well have at least the top 10. because of his interviews, he did.

German Grand Championship:
I will have to check as well if i have any pictures left from it.

5 minutes ago, shmitty said:

While I do understand your criticisms of this system, it is not meant to a relative marker of player skill. It is meant to be fun and for those that do like attending lots of tournaments a way of tracking progress, etc.

I actually looked at many different ranking systems like ELO before settling on this one. My general feeling that was while ELO would be a great fit for a local Armada group to do a ranking ladder it didn't work as well for ranking players that would likely never actually play each other.

This is by no means a perfect system. It has many warts and failings, but it has the potential to be a great deal of fun for those that want to participate. There is nothing official about it and players will only appear if they opt in. It is also immensely practical in that it is easy to setup and administer. I didn't want "perfect to be the enemy of the good" in this case.

ELO is not really possible without full data (and it means even all pairings).

As you said, it is good when you have everything. But for Armada it might be still a bit bad. Because you also need to see the score or mov for a good ELO (someone who is winning 10-1 should be valued total different than someone who barly won with 6-5).

And ELO does not allow anyone to do fun tournament or test new lists. Because the ELO will go down way to much, when someone is testing something and loosing on a regional or store.

2 minutes ago, Tokra said:

There is a really nice variant for ranking on the T3 . But it require alwasy the full data of all tournaments. So nothing that can be done easy.

But when it helps your, i have these:

For worlds i have to look through my pictures. I might have a standing after the last round as well.

German Grand Championship:
I will have to check as well if i have any pictures left from it.

Thanks @Tokra

I was actually able to track down the final results from Worlds, so those are in there. I will get these results from Euros added in tomorrow. That helps too.

That T3 system sounds pretty interesting, but perhaps a little more coding than I can do to setup.

Would it be to difficult to track all games not just tournaments? At least at 400 points.

1 hour ago, ScottGilbert25 said:

Would it be to difficult to track all games not just tournaments? At least at 400 points.

It is not difficult, when all TOs would upload all the tournament data to one local database (like it is done at other games/companies).

But for Armada it is already hard enough to get the results from a tournament. No way at all to get the full data.

Does X-Wing have anything like this?

Thanks again for doing this, super fun. It is a great motivator for me to get out to other tournaments.

Edited by icedbecker2007

Hm. Instead of player ranking being just tourney A points + Tourney B points, Can a weighted average apply?

Something like .5*(most recent placing)+.5*(average of all placings in last year).

Though that doesn’t reward more active players and can make a one-off win look more impressive than consistent placing. So possibly add a modifier for number of events in last year?

(.5*LastPlacing + .5*YearAverage)*1.05*NumberOfEventsPlayed

Spitballing ideas. Feel free to trash or ignore them

Playing in my first prime in a week. Excited to weigh down the average

9 hours ago, Tokra said:

ELO is not really possible without full data (and it means even all pairings).

As you said, it is good when you have everything. But for Armada it might be still a bit bad. Because you also need to see the score or mov for a good ELO (someone who is winning 10-1 should be valued total different than someone who barly won with 6-5).

And ELO does not allow anyone to do fun tournament or test new lists. Because the ELO will go down way to much, when someone is testing something and loosing on a regional or store.

I think an Elo system (it's Elo btw not ELO ;) ) could work and can also factor in the size of a win. Conveniently the ELO formula works on a range of 400 to -400. In chess it's 400 for a win, 0 for a draw and -400 for a loss. In Armada you can just use MOV weigh is going to be 400 to -400!

  1. For each win, add your opponent's rating plus 400,
  2. For each loss, add your opponent's rating minus 400,
  3. And divide this sum by the number of played games.

Players can choose to play a ranked game or not and Vassal games should also count in order to have some international "normalisation" between regions.

This also means that a new player getting a narrow loss to a top player can see their ratings go up.

For those (like me!) who can't make it to many tournaments it also gives more opportunity for competitive games. But of course this is all really about having a bit of fun!

When players meet in a tournament they can choose to also play games as ELO rated and just need to have a score sheet and ELO names that can submitted to ranking afterwards.

In a tournament head-to-head matches are needed and may result in changes to rankings differently to final tournament standings as skill of the opponents will also a factor.

As with chess, new players can have a provisional score for the first few games to protect established players with high rankings getting too big a hit on their rankings if they lose to an experienced player with a provisional ranking.

I will have a look at the last two years of Vassal tournaments as we have head-to-head scores for every game and see what the Elo scores look like from that as a little experiment and possible starting point for a ranking system.

This is strictly for FFG posted events? I was wondering if we would add event like Texas Galactic Open to the list. I can get the scores for this event if needed.

This is a cool idea for “funsies”. I used to get excited by the player rankings Decipher used to maintain for the Star Wars CCG (“an elegant game for a more civilized age”). If I recall correctly, it was based on the system used for chess. You got a boost from playing higher rated players. Higher rated players would hardly get anything by playing (and beating) lowered ranked players. Decipher’s system was slow as molasses to be updated, though it was reliant on paper tournament results being mailed or otherwise transmitted to be input into the ranking system.

50 minutes ago, ptownhiker said:

This is a cool idea for “funsies”. I used to get excited by the player rankings Decipher used to maintain for the Star Wars CCG (“an elegant game for a more civilized age”). If I recall correctly, it was based on the system used for chess. You got a boost from playing higher rated players. Higher rated players would hardly get anything by playing (and beating) lowered ranked players. Decipher’s system was slow as molasses to be updated, though it was reliant on paper tournament results being mailed or otherwise transmitted to be input into the ranking system.

You come at the king, you best not miss.

(Loses to @Tokra like 47 times in a row)

18 minutes ago, geek19 said:

You come at the king, you best not miss.

(Loses to @Tokra like 47 times in a row)

I am not throwing away my- *rolls blanks*

*rerolls into more blanks*

Edited by The Jabbawookie
6 hours ago, BozoLtD said:

This is strictly for FFG posted events? I was wondering if we would add event like Texas Galactic Open to the list. I can get the scores for this event if needed.

When I was discussing this during setup that was one example of a non-FFG tournament that I would like to add in. Tournaments like that and the Vassal world Cup I believe are worth considering.

10 hours ago, Church14 said:

Hm. Instead of player ranking being just tourney A points + Tourney B points, Can a weighted average apply?

Something like .5*(most recent placing)+.5*(average of all placings in last year).

Though that doesn’t reward more active players and can make a one-off win look more impressive than consistent placing. So possibly add a modifier for number of events in last year?

(.5*LastPlacing + .5*YearAverage)*1.05*NumberOfEventsPlayed

Spitballing ideas. Feel free to trash or ignore them

I've looked at some similar thoughts to that. What would probably be the easiest to implement would be to only count the 5 highest results for a player in a calendar year. Some of it is just the limitations of what I can reasonably automate in spreadsheet. Right now it is setup for the input to be minimal and the computing to all be automatic.