Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[math]

I just won a game on Goko which finally pushed my pro rating over 6000!  I have been aiming for this, and failing, for several weeks.  I am currently #59 on the leaderboard, although I probably won't stay that high for long.  I'm just making this thread to celebrate/brag.

It's ironic that I only got my rating up to 17 on Isotropic before it shut down.  I wonder how high I could have gone there if I had been able to keep going.  Does anyone have an estimate as to what 6000 would be on iso?

[ashersky]

Congrats!

[jsh357]

seems to be equivalent with about level 35+ to me.

[Polk5440]

seems to be equivalent with about level 35+ to me.

I would agree with this.

I am waiting for the day when someone breaks 9000....

[Robz888]

seems to be equivalent with about level 35+ to me.

I concur. I'm currently hanging out in the 6300~ range, which seems not so far off my Level 42~ on Iso.

[Qvist]

I think iso and Goko levels aren't comparable. I was already at over 7000 on Goko, but never managed to break level 38 on iso.

[qmech]

I think iso and Goko levels aren't comparable. I was already at over 7000 on Goko, but never managed to break level 38 on iso.

I wonder if this is an intrinsic property of the rating system, or just because the system is still finding its feet.  I hovered around 40 on Iso and am currently 6450 on Goko.  I find that, playing whoever joins my games, I'm typically gaining 20 points for a win, and losing 60 for a loss.  The standard is sufficiently low that's it's still possible to grind out a steady rating increase.  Has your experience been the same, or is your 7000 mostly from higher skill players?

[RTT]

i´m 6000 now already on goko and never got over 25 on ISO. I think reaching higher levels on Goko goes much faster.

[Qvist]

I think iso and Goko levels aren't comparable. I was already at over 7000 on Goko, but never managed to break level 38 on iso.

I wonder if this is an intrinsic property of the rating system, or just because the system is still finding its feet.  I hovered around 40 on Iso and am currently 6450 on Goko.  I find that, playing whoever joins my games, I'm typically gaining 20 points for a win, and losing 60 for a loss.  The standard is sufficiently low that's it's still possible to grind out a steady rating increase.  Has your experience been the same, or is your 7000 mostly from higher skill players?

I expect that I'm just lucky so far and as I only played ~100 Pro games on Goko so far the uncertainty factor is still high.
I play almost exclusively against people I know (that means mainly against all of you from the forums and also IRL friends), but I don't know how much this could affect the rating.

[Tables]

I had a lucky stream when I first started playing pro games, leading to the image in my sig. Now, I've just come off of a losing streak, and I'm below 5K. I think the system is far more temperamental (at first, at least) compared to Iso. I probably will end up hovering around 5.5K-6K, which is about level 30 on Iso.

[markusin]

My Goko pro rating is above 6000 now, but on Iso I had a hard time even maintaining a rating past 25. The thing is I mostly just play with anyone that joins my games. I'll still get like 20 points against someone with a rating below 1000 or something.

[jsh357]

I play against anyone but guests (it takes too long to wait on high skill players), and I have a very hard time making progress.  Winning gets me 2-10 points generally, but losing will set me back by 20-70.  I would assume the players who are on top primarily play people with high skill or just lose so little that it doesn't impact their ratings much.  I haven't seen any of the top guys on, so I haven't been able to look at their win/loss records to tell.

[WanderingWinder]

I play against anyone but guests (it takes too long to wait on high skill players), and I have a very hard time making progress.  Winning gets me 2-10 points generally, but losing will set me back by 20-70.  I would assume the players who are on top primarily play people with high skill or just lose so little that it doesn't impact their ratings much.  I haven't seen any of the top guys on, so I haven't been able to look at their win/loss records to tell.

Both. From what I understand, it's basically impossible to get very high playing people who are very far below you - part of why there are 6000-rated people who are super strong and some who are moderately good. It's a problem with the system. To be fair to them, it's a tricky problem to solve totally, ad iso (as well as many other rating systems) had the same problem to an extent, though not to this extent.

[Polk5440]

I think iso and Goko levels aren't comparable. I was already at over 7000 on Goko, but never managed to break level 38 on iso.

I wonder if this is an intrinsic property of the rating system, or just because the system is still finding its feet.  I hovered around 40 on Iso and am currently 6450 on Goko.  I find that, playing whoever joins my games, I'm typically gaining 20 points for a win, and losing 60 for a loss.  The standard is sufficiently low that's it's still possible to grind out a steady rating increase.  Has your experience been the same, or is your 7000 mostly from higher skill players?

I think Qvist has just gotten better. Presence of DA cards matters a lot, too. Some people have learned how to use them better than others.

[Qvist]

I think iso and Goko levels aren't comparable. I was already at over 7000 on Goko, but never managed to break level 38 on iso.

I wonder if this is an intrinsic property of the rating system, or just because the system is still finding its feet.  I hovered around 40 on Iso and am currently 6450 on Goko.  I find that, playing whoever joins my games, I'm typically gaining 20 points for a win, and losing 60 for a loss.  The standard is sufficiently low that's it's still possible to grind out a steady rating increase.  Has your experience been the same, or is your 7000 mostly from higher skill players?

I think Qvist has just gotten better. Presence of DA cards matters a lot, too. Some people have learned how to use them better than others.

Well, I lost a lot of games today, so my guess might be right that I'm not that good. 

[Fabian]

Uhm speaking of rating.. I just beat some guy a few times in a row, I gained like 60 points or 50 points or whatever per game.

Then I beat him 47-46 and I gained... 0 points.

I can't really find words to describe how bad this rating system is if that just happened (and it did)

Edit: In picture form:

[Watno]

Yeah, I think there's a bug that occasionally games are ignored for rating. Happened to me a couple of times, but I don't think I ever reported it.

[Fabian]

Ok if that game was ignored because of some bug, I guess that's.. not as bad (lol imagine someone typing this sentence anywhere outside of goko)

[Watno]

I reported this on getsatisfaction with your pic now. Hope that's ok.

[qmech]

It's possible that the game deliberately doesn't reward very long winning streaks against the same player to discourage gaming of the leaderboard.

[WanderingWinder]

No, it's not that. I have had this happen before, too, where for some reason, seemingly randomly, I got 0 points off of a win... from someone who at the time was higher rated than me.

[Blueswan]

At what rating would you say that a player stops sucking? (hoping that I no longer suck too hard - my pro rating is currently 4098 and fluctuating wildly. I pretty consistently beat all the bots in 2-player, but lose just about whenever I play humans, part of the explanation for that is that I play far too quickly for my own good against humans, because I don't want to make them wait for me to consider my options).

[Twistedarcher]

If the rating system is an ELO-type rating system, which I believe someone from Goko said it was in a FAQ, then there's no true way to match ratings on goko to levels on isotropic.

The more active players there are, the higher the top ratings will be. More overall rating points will be in the system, allowing the scores to become slightly more spread out. As more and more people create accounts and participate in the rating system, the points will slowly gravitate towards the top players, increasing the top possible ratings ever so slightly.

The ratings that can be reasonably achieved, therefore, are determined mostly by the number of players. The only way someone would ever be able to reach 9000+ is if many, many more players joined in the rating system. If the number of players remains steady, it will be difficult for any player to get above the highest current ratings.

[WanderingWinder]

At what rating would you say that a player stops sucking?

Best guess? 10,000.

If the rating system is an ELO-type rating system, which I believe someone from Goko said it was in a FAQ,

Well, they said this, but in actual practice, it doesn't seem to have almost anything in common with the Elo system (by the way, I don't know why people capitalize all three letters of this name, but this is an aside), other than they are both rating system where higher numbers are supposed to represent better players.

[Twistedarcher]

If the rating system is an ELO-type rating system, which I believe someone from Goko said it was in a FAQ,

Well, they said this, but in actual practice, it doesn't seem to have almost anything in common with the Elo system (by the way, I don't know why people capitalize all three letters of this name, but this is an aside), other than they are both rating system where higher numbers are supposed to represent better players.

Well, it's hard to say without more information, but it does have some similarities, in that defeating an opponent ranked higher gives more points than defeating a lower ranked opponent, and vice versa with losses. The games with no gain are a mystery, to be sure, but it seems to have several components of the ELO system.

As to why ELO rather than Elo, I know that sports economics literature uses ELO rather than Elo. It probably should be Elo, since it's a guy's name and doesn't stand for anything, but I think both are acceptable.

[WanderingWinder]

If the rating system is an ELO-type rating system, which I believe someone from Goko said it was in a FAQ,

Well, they said this, but in actual practice, it doesn't seem to have almost anything in common with the Elo system (by the way, I don't know why people capitalize all three letters of this name, but this is an aside), other than they are both rating system where higher numbers are supposed to represent better players.

Well, it's hard to say without more information, but it does have some similarities, in that defeating an opponent ranked higher gives more points than defeating a lower ranked opponent, and vice versa with losses.

Two problems with this: first, this is true of, well, virtually every rating system, so it doesn't really make it Elo-like. Second, it isn't true! (as has been shown in numerous examples; generally, yes this happens, but only generally).

[qmech]

YOUR RATING IS



"ROCKING"

[Stealth Tomato]

At what rating would you say that a player stops sucking? (hoping that I no longer suck too hard - my pro rating is currently 4098 and fluctuating wildly. I pretty consistently beat all the bots in 2-player, but lose just about whenever I play humans, part of the explanation for that is that I play far too quickly for my own good against humans, because I don't want to make them wait for me to consider my options).

With no bots, I would guess about 5000 pro, 4000 casual. The casual ratings appear to be naturally depressed somewhat, and it's extremely difficult to get past 5000.

Bots make it strange because they're not players and they're naturally predictable.

[ashersky]

At what rating would you say that a player stops sucking?

Best guess? 10,000.

Don't you mean more than 9000?

[ashersky]

At what rating would you say that a player stops sucking? (hoping that I no longer suck too hard - my pro rating is currently 4098 and fluctuating wildly. I pretty consistently beat all the bots in 2-player, but lose just about whenever I play humans, part of the explanation for that is that I play far too quickly for my own good against humans, because I don't want to make them wait for me to consider my options).

With no bots, I would guess about 5000 pro, 4000 casual. The casual ratings appear to be naturally depressed somewhat, and it's extremely difficult to get past 5000.

Bots make it strange because they're not players and they're naturally predictable.

With enough games, my guess is that the "halfway point" from iso (I think lvl. 25) ends up somewhere around 4800 on Goko.  The problem is, 9600 isn't lvl 50, because I don't see anyone getting up that high yet.  I expect WW, et al. to push 7500?  Is that feasible?

[qmech]

At what rating would you say that a player stops sucking? (hoping that I no longer suck too hard - my pro rating is currently 4098 and fluctuating wildly. I pretty consistently beat all the bots in 2-player, but lose just about whenever I play humans, part of the explanation for that is that I play far too quickly for my own good against humans, because I don't want to make them wait for me to consider my options).

With no bots, I would guess about 5000 pro, 4000 casual. The casual ratings appear to be naturally depressed somewhat, and it's extremely difficult to get past 5000.

Bots make it strange because they're not players and they're naturally predictable.

With enough games, my guess is that the "halfway point" from iso (I think lvl. 25) ends up somewhere around 4800 on Goko.  The problem is, 9600 isn't lvl 50, because I don't see anyone getting up that high yet.  I expect WW, et al. to push 7500?  Is that feasible?

If 1000 is zero we get 50 being 7600 and 40 being 6280.  In the early days of Iso 40 was pretty much the ceiling, with the 50s only coming once the number of players had increased.

Your 7500 seems a reasonable guess.  If it's smashed then the rating systems likely aren't comparable in this way.

[heatthespurs]

My general feeling is that the Goko rating does not reflect player's strength as good as Iso did.

I am ~6000 in Goko and ~30 in iso. Back in the iso days, if I automatch and select "+/-10 rank", I could generally be able to play with someone that have roughly similar skill level with me (well, the lv.38 player is generally better and lv.22 is generally weaker than me. Though it is still very possible to beat a lv.38 or lose to a lv.22 in a game)

Now in Goko, I could be playing a 5500-6000 player which has skill level of roughly lv.15-20 in iso. And it happens quite often. Maybe it could be that everyone is still not familiar with Dark Age? (but so do I)...

[Stealth Tomato]

Now in Goko, I could be playing a 5500-6000 player which has skill level of roughly lv.15-20 in iso. And it happens quite often. Maybe it could be that everyone is still not familiar with Dark Age? (but so do I)...

Note that even a good rating system typically takes some time to stabilize, since new players on the system start at zero even if they are already very good.

[ooksoo]

I was around 5.6k until I decide to get to the top, My friend BrotherJay was on top for pro for a few day I think, and he is around 6700 ( i remeber this because it suprise the XXXX out of me when I heard he has highest ratting on pro) And I took him Down,when i played him, for the first few games i get about 100 points per game, when I loss is about 30-40, when I got up to 6200 and he is down to 5800 i get about 10-20 per game, and I started playing random people from there.

[ooksoo]

My general feeling is that the Goko rating does not reflect player's strength as good as Iso did.

I am ~6000 in Goko and ~30 in iso. Back in the iso days, if I automatch and select "+/-10 rank", I could generally be able to play with someone that have roughly similar skill level with me (well, the lv.38 player is generally better and lv.22 is generally weaker than me. Though it is still very possible to beat a lv.38 or lose to a lv.22 in a game)

Now in Goko, I could be playing a 5500-6000 player which has skill level of roughly lv.15-20 in iso. And it happens quite often. Maybe it could be that everyone is still not familiar with Dark Age? (but so do I)...

never played on iso, but I experienced more then a few games with iso players and they left for no reason, ( how do I know there are from iso? a lot player that have a very high rating with not that many of games,for example 5800 with 123-46-2 when I see players like that I would ask are you from iso?) and often iso players comes into goko and trash talk on the chat room saying how bad goko is, rofl

[ftl]

I don't know whether I trust the goko rating system right now. I'm #3 on it right now, and I really don't think I'm particularly good - on iso at least, I wasn't remotely close to the top 10 players, my rating was fairly unremarkable. So I suspect that me being #3 on the leaderboard is a symptom of some sort of problem with it, but I don't know what that problem is or how it affects everyone else.

[SCSN]

often iso players comes into goko and trash talk on the chat room saying how bad goko is, rofl

What's even funnier is that they're right 

[Stealth Tomato]

I don't know whether I trust the goko rating system right now. I'm #3 on it right now, and I really don't think I'm particularly good - on iso at least, I wasn't remotely close to the top 10 players, my rating was fairly unremarkable. So I suspect that me being #3 on the leaderboard is a symptom of some sort of problem with it, but I don't know what that problem is or how it affects everyone else.

My rating has relatively stabilized (I typically play 4500-5500 range players with occasional higher ratings, I'm around 6300 myself). I'm curious whether this is due to good design or if the ability to change your rating decays over time. I do know that my willingness to play matches against 4000-level players is likely hurting me overall, since it is difficult to beat competent Dominion players at the rate required to move up in level (some boards are sufficiently high-skill that a better player will thrash a lesser one; others don't offer much advantage for nuanced play).

The ratings are naturally going to be screwy with new-to-Goko players entering the system, since these players have a massive mismatch between actual skill and results available to the rating system. However, I don't see reason to declare the ratings are particularly bad.

[Fabian]

I don't have a strong opinion if the ratings are "bad" either, but it does seem like it's pretty necessary to avoid playing low rated players if you want your own rating to stay high. Like, I feel like it's pretty weird to have a rating system in 2p Dominion where one player gains 7 points after a win and loses 120 points after a loss. That seems kinda harsh, although maybe that just means you're not supposed to have a rating that high, or something, though that seems pretty weird too.

[WanderingWinder]

I don't have a strong opinion if the ratings are "bad" either, but it does seem like it's pretty necessary to avoid playing low rated players if you want your own rating to stay high. Like, I feel like it's pretty weird to have a rating system in 2p Dominion where one player gains 7 points after a win and loses 120 points after a loss. That seems kinda harsh, although maybe that just means you're not supposed to have a rating that high, or something, though that seems pretty weird too.

This is *potentially* fine, but it means that you should have a win percentage of... 94.5% against that other player.

The problem is that they don't have the system scaled right. To be fair to them, it's really hard to do - you have to have it be right for someone 5 points away from you and 500 or 50000 point away, and that's hard to scale properly. However, I think it is also fair to say that it's incredibly clear that they are way off, and they could adjust this....

[Schneau]

At this point, the only thing I'm convinced Goko does wrong with their rating system is report the changes after every game. It actually makes sense, as far as I can tell, that if you play someone much lower rated than you that you won't gain much by winning and will lose a lot by losing. I believe this happened on Isotropic as well, but since people couldn't check their rating changes until the next day, no one ever complained about it.

There may be other problems, but I'm not convinced of them at this point.

[Fabian]

I definitely don't think the imbalance was as big on isotropic. I do think there were plenty of competent lvl 25-30ish players which any 45ish player would have had a hard time staying even on in the long run (my RL friend who I played a ton being a good example), but it wasn't even close to the kind of stuff you encounter on goko when you're 7150. With that rating, it seemed pretty much every opponent was just impossible to keep your rating against in the long run. Imagine playing some top15 or top20 player and having to score 75-80% to break even, it's ridiculous (in the context of Dominion, anyway, obviously something like chess would be a whole different story).

Edit: Basically, this "it's incredibly clear that they are way off, and they could adjust this"

[dondon151]

My cumulative winrate on Iso was probably about ~75%, and most of that was playing opponents in the L25-35 range.

[Fabian]

75%? Councilroom link? I was at ~60% for all my isotropic life (past the first couple months) and that was good enough to be near the top of the leaderboard, although admitedly I played much stronger opponents (average level ~35.6).

[Stealth Tomato]

I definitely don't think the imbalance was as big on isotropic. I do think there were plenty of competent lvl 25-30ish players which any 45ish player would have had a hard time staying even on in the long run (my RL friend who I played a ton being a good example), but it wasn't even close to the kind of stuff you encounter on goko when you're 7150. With that rating, it seemed pretty much every opponent was just impossible to keep your rating against in the long run. Imagine playing some top15 or top20 player and having to score 75-80% to break even, it's ridiculous (in the context of Dominion, anyway, obviously something like chess would be a whole different story).

Edit: Basically, this "it's incredibly clear that they are way off, and they could adjust this"

Then again, 7150 would put you at #1 on the leaderboard, correct? A #1-on-the-leaderboard rating should be incredibly difficult to maintain. That's kind of like complaining it's difficult to sustain Level 50 on Iso.

Also, how far apart do the ratings have to be to require a 95% winrate? Because there are a lot of players against whom I really should win 19 out of 20 games.

[Fabian]

Yeah I was #1 literally for 1 or 2 games, that didn't last very long heh. What I mean to say by that is that 7150 on goko last week seems much tougher to maintain than level 50 on isotropic 2 months ago. The whole "lose one game, win 8 in a row to get the rating back" thing didn't exist on isotropic, I'm sure other top players on isotropic will agree with me. No one expects you to win 8 out of 9 games against half competent opponents, it's just not sustainable in dominion.

I haven't played a single "I should win 19 out of 20" opponent on Goko so far, and it surprises me you think they exist in abundance.

[qmech]

75%? Councilroom link? I was at ~60% for all my isotropic life (past the first couple months) and that was good enough to be near the top of the leaderboard, although admitedly I played much stronger opponents (average level ~35.6).

70%, which is still very impressive.

[ftl]

Heh, I guess we're not going to figure out whether the rating system is good or bad just by abstract talk... I'd assumed the opposite, that it was too easy to gain rating by playing low-level people, I don't think I've played many of the people in the top 20, just a few games vs them, but my rating got pretty high.

I have *got* to find time to join in council room development and pull in goko logs, then I can poke at their ranking system and compare actual win percentages and such... ...my thesis defense is on may 16th, I think after that I'll have time to devote some evenings to it.

[dondon151]

75%? Councilroom link? I was at ~60% for all my isotropic life (past the first couple months) and that was good enough to be near the top of the leaderboard, although admitedly I played much stronger opponents (average level ~35.6).

I played more often on an alt: http://councilroom.com/player?player=cbaka

And since I played fewer games and most of those games were spent going up the leaderboard, there's no doubt that I played weaker opponents on average. Even so, I "only" had a 72.6% win rate on this account, which justifies your dissatisfaction with Goko's rating system.

[qmech]

Heh, I guess we're not going to figure out whether the rating system is good or bad just by abstract talk... I'd assumed the opposite, that it was too easy to gain rating by playing low-level people,

That's how I got my rating high.  You don't want to play anyone rated 2000 because the fluke losses set you back too much, but you can definitely get over 6500 grinding out wins against weaker players.

As soon as I start playing WW rated games my Goko rating takes a hit.

[Stealth Tomato]

I haven't played a single "I should win 19 out of 20" opponent on Goko so far, and it surprises me you think they exist in abundance.

Have you ever accidentally played a career Lv10 or so on Iso? They're horrendous. They make repeated elementary mistakes on fairly straightforward boards, to the point where even bad boards are easy wins (because they try for things that lose consistently to Big Money). I could probably win 80-90% against them playing five-card kingdoms.

[Fabian]

"Have you ever accidentally played a career Lv10 or so on Iso?"

I played 5k or 6k games or something, and no, no I can't say that I have.

[Stealth Tomato]

"Have you ever accidentally played a career Lv10 or so on Iso?"

I played 5k or 6k games or something, and no, no I can't say that I have.

Fair enough. I had a few encounters due to screwing up automatch settings and the not-displaying-level bug (I looked up their level after accepting) and a bunch more when Google authentication was down so I created a Yahoo alternate. Suffice to say that until about Lv15 the players are not remotely competent. 20 is around the point where they start to play relatively well on uncomplicated boards.

[Fabian]

ST,

See that just blows my mind. All I ever did was play level ~30+ players, except when I played my lvl 25 RL friend, who was somehow better than any sub 35 player. How he was level 25 while being very very competent while a level 20 player is apparentely close to incompetent to the point of 95% winrates is very strange to me. Then again I guess we just had different isotropic experiences.

[dondon151]

I'm not even sure that a 94% winrate is possible against a L10 player if you go second. Just the likelihood of bad luck could set you back.

[Fabian]

I'm 100.0% convinced it's not, fwiw.

[ftl]

There were a bunch of ways to perpetually end up level 10-20 on iso. One is to play a lot of games but be bad at it.

The other is simply not to play much. If you only play a few games a week online, you'll never get above level 10-20 no matter how good you are.

[Polk5440]

my thesis defense is on may 16th

GOOD LUCK! (Do you want to brag and share a link to your paper? I might even read it.)

[Kirian]

I'm 100.0% convinced it's not, fwiw.

Indeed.  That level 10 player would have to play significantly worse than BM+X most of the time, and still an inferior strategy can win from the first seat, though it takes good luck.

It might be possible to achieve 75% win rate over a level 10 player.

[Jiriki]

I was hovering at 35 before Iso went down and have now cleared 6K playing 2-3 games a day +/-1K. I would say the quality of competition over 5K on goko is closer to the 20s on Iso than 30+. Seems like some combo of the serious Iso players didn't all move over, figuring out DA (I feel like I've been way ahead of my opponents on average using these cards), and that the ratings haven't had the time to stabilize like on Iso.

[meandering mercury]

my thesis defense is on may 16th

GOOD LUCK! (Do you want to brag and share a link to your paper? I might even read it.)

My thesis defense is in 90 minutes!!

Why am I lurking these forums again?!

(and good luck for your defense too)

[synpax]

I think I was in the low teens on Iso but have been on a 20+ game winning streak on Goko and am at 5200+ on the pro board now. A few close games. I'm not sure what the equivalency is, and I never look at my opponents score/rank before we begin the game.

I'd like to think I'm getting better. I was new to Dominion as Iso was shutting down. And on ISo, I was in the habit of watching videos and browsing 9gag while I played. At least now I pause the videos long enough to figure out a strategy.

[ragingduckd]

I'm no expert on the math (and I'd be grateful is someone who is would point me in the right direction), but...

It seems to me that ELO is inappropriate for games of chance.  ELO models win rates as
    P(player 1 wins) = P(x1>x2),
where x1 and x2 are Gaussians random variables with means r1 and r2 representing the ratings of the two players.  IMO, a more appropriate model for Dominion is something like
    P(player 1 wins) = 10% + 80%*P(x1>x2),
the same model, but with a 10% chance of automatic victory for each player.  That 10% represents the odds of opening Mountebank-Chapel, or getting all the prizes in a tourney game, or winning the Duchy split in a Duke game, etc.

In my experience, the ELO/Goko expectation that a 2000-pt difference match go 5-1 in favor of the higher rated player just isn't realistic.  The weaker player wins much more often than the rating system expects.  Consequently, weak players improve their ratings by playing against stronger players and strong players benefit by limiting their opponents to other strong players.

If this assessment is correct, then opponent selection may have a major effect on Goko ratings.  A 5500 player who plays mostly with 4000-ish players might actually be just a skilled as a 6000 player only plays against the top 25.

[DStu]

I don't think it's so easy.  There is no automatic win in Dominion, someone who has never played this game will not win 100% of the games against you when they open 5/2 vs 4/3 on a Mountebank/Chapel board.  They probably will even lose most of the games.

How much variance a kingdom has depends on the board, but I don't really see a good way on how to get this into a model.  It also depends on the player skills, e.g. say Horse Traders/Duke is dominant on a board without other support, that's more or less a 50/50+first player advantage if both players know the strategy, but if not the board might be much more complicated.

[ragingduckd]

I don't think it's so easy.  There is no automatic win in Dominion, someone who has never played this game will not win 100% of the games against you when they open 5/2 vs 4/3 on a Mountebank/Chapel board.  They probably will even lose most of the games.

I agree.  That sort of formula couldn't apply for all ratings differences.  If a 10% adjustment were appropriate for a 1500-pt difference, then you'd need a smaller one for a 3000-pt difference.  But that's clumsy and inelegant, and I'm not seriously proposing it as a modification of Elo... it's just an example of how using Elo for Dominion might generate a bias in favor of the lower-rated player.

Personally, I find it impossible to break even against anyone more than 1500 points lower than me.  I've also jumped about 300 points since I changed my match criteria from 5000+ to 5500+.  But maybe that's just noise, or maybe it's the variance in the quality of my play that's higher than Goko's rating system expects.

How much variance a kingdom has depends on the board, but I don't really see a good way on how to get this into a model.  It also depends on the player skills, e.g. say Horse Traders/Duke is dominant on a board without other support, that's more or less a 50/50+first player advantage if both players know the strategy, but if not the board might be much more complicated.

Actually, I really like this idea.  I doubt you could map kingdoms to variances, but modifying Elo so that the variance is drawn from some random distribution might capture the key difference between a game like chess and a game like Dominion.  It feels like it might even be mathematically tractable.

I'm less confident about modifying the variance based on the players' skill, but maybe that could work too.

[SCSN]

or winning the Duchy split in a Duke game, etc.

With all the curses in your deck...

[Stealth Tomato]

I don't think it's so easy.  There is no automatic win in Dominion, someone who has never played this game will not win 100% of the games against you when they open 5/2 vs 4/3 on a Mountebank/Chapel board.  They probably will even lose most of the games.

I agree.  That sort of formula couldn't apply for all ratings differences.  If a 10% adjustment were appropriate for a 1500-pt difference, then you'd need a smaller one for a 3000-pt difference.  But that's clumsy and inelegant, and I'm not seriously proposing it as a modification of Elo... it's just an example of how using Elo for Dominion might generate a bias in favor of the lower-rated player.

Personally, I find it impossible to break even against anyone more than 1500 points lower than me.  I've also jumped about 300 points since I changed my match criteria from 5000+ to 5500+.  But maybe that's just noise, or maybe it's the variance in the quality of my play that's higher than Goko's rating system expects.

Remember, 300 points is ~3 levels on Iso, which is a perfectly normal fluctuation over time.

I think the overall point that luck isn't sufficiently accounted for is valid though. Although I think that comes out more in matches with closely-ranked opponents than dramatically superior/inferior ones. I can beat a Lv3000 9 times out of 10 regardless of board. On a substantial proportion of boards, I'm flipping coins against anyone 5500+, where maybe one of us can eke out a 10-20% edge with strong play but 2/3 of games will be determined by pure chance.

[WanderingWinder]

I'm going to make a few miscellaneous factual points. First of all, I don't know of any implementation of the Elo system currently that uses a normal distribution. Elo's original formulation did, though it can actually use *any* curve for the Win Expectancy. And all the big ones now (so far as I'm aware) are using the Logistic curve.
Secondly, Goko doesn't use Elo. Further, we don't know *what* their underlying WE curve is.

Okay, opinions. Goko's system would be much much better if it were the Elo system (with a reasonable choice of K value and underlying WE curve) - of this I am pretty confident. Well, okay, it's really more to do with the choice of WE curve as well as their updating procedure. But I am pretty sure their curve - if they're even using one, which I'm not totally sure they are - is DREADFUL, much worse than a Gaussian would be. And actually, if you look at what they are saying, they have to pervert their system to not allow WE of over 100%/less than 0% in some games, which is prima facie quite bad.

Okay, the 10% sure win but-then-change-that-based-on-how-big-a-rating-gap-there-is is really wanting a different curve. Which is fine, of course, though finding exactly which one is best is EXTREMELY tricky.

You also probably don't want to have automatic cut-offs like that. Let the win% float from 0 to 100, just have a good enough system that (if it's unreasonable to expect a 99% winrate) makes it very difficult for a 99% win-rate to emerge. Which actually isn't all that hard to do, though again, finding the best curve is incredibly difficult.

[ragingduckd]

And actually, if you look at what they are saying, they have to pervert their system to not allow WE of over 100%/less than 0% in some games, which is prima facie quite bad.

Yikes, that sounds awful.  Okay, I just found the Goko wiki entry and the Funsockets Q&A, but they don't really say anything useful about their system, except that it's a TrueSkill variant.  Can you link me to where they talk about perverting their system?

Okay, the 10% sure win but-then-change-that-based-on-how-big-a-rating-gap-there-is is really wanting a different curve. Which is fine, of course, though finding exactly which one is best is EXTREMELY tricky.

Ah.  That makes sense.  Ok, so there's no reason that Elo with a non-Gaussian WE has to have the bias I'm thinking about.

[SCSN]

And actually, if you look at what they are saying, they have to pervert their system to not allow WE of over 100%/less than 0% in some games, which is prima facie quite bad.

Yikes, that sounds awful.  Okay, I just found the Goko wiki entry and the Funsockets Q&A, but they don't really say anything useful about their system, except that it's a TrueSkill variant.  Can you link me to where they talk about perverting their system?

https://getsatisfaction.com/goko/topics/summary_of_rating_calculations#reply_11914215

Specifically:

That said, we no longer allow you to lose points by winning. Instead, we're perverting the rating system a little to prevent this, only because it upsets players who have a hard time believing it's the best thing to happen in order to model their ability.

Personally, I find it impossible to break even against anyone more than 1500 points lower than me.  I've also jumped about 300 points since I changed my match criteria from 5000+ to 5500+.

Interesting. I still play anyone >5000, but I'm gonna try this out for a bit. My guess though is that Goko expects me to win 65-75% against 5000-5500's, which doesn't sound terribly unreasonable?

Does any of you have the slightest idea as to how rating points are awarded, btw? I sometimes get 7 from beating a 5800, and other times 26 from winning against a 5200. There's certainly a correlation between level difference and points gained/lost, but there's also a ton of inexplicable variation.

[ragingduckd]

I was around 5.6k until I decide to get to the top,

I like this quote a lot. 

[Blueswan]

At what rating would you say that a player stops sucking? [...]

With no bots, I would guess about 5000 pro[...]

Getting sooo close to not sucking. 



Except that this is just from playing bots, so I guess it doesn't really count.

I wanted to post this, since I'm sure I'm gonna f*** up the next game.

[Blueswan]

Oh my. Amazingly I did not f*** it up. Infact, I'd say that Lord Bottington just got his ass handed to him. Oh well, won't bother you with this anymore, I reached my goal for now. Now if they could only fix the lagging so I could start playing real humans.

[Davio]

Okay, I eventually cracked the 6k mark on the pro leaderboard myself, my addiction won over my annoyance I guess.

And I have to be honest, it wasn't at all that hard and it didn't take that many games, so I don't know what to make of it. It seems like I'm better at pro games than casual games oddly enough.

I found the level of play overall to be pretty okay, but not spectacular, like how 30-ish was back on Iso. But then again I played mostly against 5k's and not so much against 6k's. I suspect 6k is comparable to Iso's 40 and Rabid's 7k is comparable to Iso's 50+. It's good to see many of the known forum players high on the leaderboard though, it shows to me their rating system isn't entirely bogus.

The players I've played against seemed to have a decent understanding of the fundamentals of the game, but didn't quite know when to switch to greening mode (they kept building up for too long) or how to manipulate the end game. Quite often I've surprised my opponent with a sneaky 3-pile ending which I've been eyeing up 5 turns in advance.

Many boards are just about getting control of the game and I tend to sacrifice a lot of early momentum to get the control I want through the mid- and endgame, while my opponents just seemed to only have a short term plan while lacking a strategic game plan for the entire game. They would get a Witch and then just start playing a glorified big money type game.

[Lekkit]

I agree with Davio. A lot of the players seems like they haven't been playing multiple thousands games. I guess this is bound to change as time goes by.

[DG]

Goko opponents may play differently if they are accustomed to playing against multiple opponents or bots.

[Stealth Tomato]

There's been a definite skill drop since Iso. There was a point at which I was arguably world-class, and I never got above #3 on Iso and typically hung around the middle of the top 10. I am certainly not world-class now and yet I'm sitting #3 on Goko.

[ragingduckd]

There's been a definite skill drop since Iso. There was a point at which I was arguably world-class, and I never got above #3 on Iso and typically hung around the middle of the top 10. I am certainly not world-class now and yet I'm sitting #3 on Goko.

Is it just that some of the best Iso players haven't transitioned?

There was (surprise, surprise) a huge increase in Goko skill when the Iso players transitioned in March and April.  None of Goko's top-10 from early March is higher than 30 today.

[Stealth Tomato]

There's been a definite skill drop since Iso. There was a point at which I was arguably world-class, and I never got above #3 on Iso and typically hung around the middle of the top 10. I am certainly not world-class now and yet I'm sitting #3 on Goko.

Is it just that some of the best Iso players haven't transitioned?

There was (surprise, surprise) a huge increase in Goko skill when the Iso players transitioned in March and April.  None of Goko's top-10 from early March is higher than 30 today.

I think it's a couple of things. Many players haven't transitioned. Many of those who transitioned have let their game slip. Dark Ages threw a lot of people for a loop.

[Kirian]

Also, the climb and fall speeds can be very fast.  I climbed about 50 places in three games to get back into the top 100, after a fall of similar magnitude the other day.

This also gives a sense of how many good players haven't transitioned; I was never in the top 100 on Iso.  Top 250, yes, back when there were fewer players, but generally around 400 recently.  So something like three in four good players haven't transitioned.

[Dsell]

I played my first two pro games ever on Goko earlier. I won both and gained almost 4000 points, which seems insane. I went from 1000 to 4800 and am in the top 500 now, which is...cool...I guess...but seems totally bizarre and wrong after just 2 games. I've seen lvl 5000 compared to lvl 20-25 on iso in this thread, so this is like a 15+ lvl jump in iso terms.

[StrongRhino]

I played my first two pro games ever on Goko earlier. I won both and gained almost 4000 points, which seems insane. I went from 1000 to 4800 and am in the top 500 now, which is...cool...I guess...but seems totally bizarre and wrong after just 2 games. I've seen lvl 5000 compared to lvl 20-25 on iso in this thread, so this is like a 15+ lvl jump in iso terms.

It's not even like he beat a 6000 or anything, he beat me, and I was like 5000 before the first, and 4600 after the second.

[RTT]

I played my first two pro games ever on Goko earlier. I won both and gained almost 4000 points, which seems insane. I went from 1000 to 4800 and am in the top 500 now, which is...cool...I guess...but seems totally bizarre and wrong after just 2 games. I've seen lvl 5000 compared to lvl 20-25 on iso in this thread, so this is like a 15+ lvl jump in iso terms.

the first few games dont say that much. you have a high deviation and gain a lot or loose a lot of points in 1 game.

[Watno]

You will notice that when you look at your name in the leaderboard, it doesn't say your position next to it, but "prov". This means your rating is provisional, and is just the best guess that can be made with the limited information available. You won't actually appear on the leaderboard (visible to others), once you have played more games.

[Warfreak2]

A rating is not a reward for playing well, it is for finding people of similar skill levels so that you can have reasonably evenly matched games. A rating jump is neither fair nor unfair; it is either accurate or inaccurate. Of course you can "gain" 4000 points for winning your first two pro games; if you beat someone around 5000 level in your first game, the system is not going to say "oh, I guess you should play with opponents around 1050 instead of 1000".

[Dsell]

You will notice that when you look at your name in the leaderboard, it doesn't say your position next to it, but "prov". This means your rating is provisional, and is just the best guess that can be made with the limited information available. You won't actually appear on the leaderboard (visible to others), once you have played more games.

Yep, that's all correct. Hopefully (if I play more in the near future) I can keep my rating up and get on the leaderboard. I am happy that my rating is "correcting itself" quickly, but I was totally astonished to see it change that fast. If it settles to a more reasonable pace of change when I have more games under my belt, I can work with that.

A rating is not a reward for playing well, it is for finding people of similar skill levels so that you can have reasonably evenly matched games. A rating jump is neither fair nor unfair; it is either accurate or inaccurate. Of course you can "gain" 4000 points for winning your first two pro games; if you beat someone around 5000 level in your first game, the system is not going to say "oh, I guess you should play with opponents around 1050 instead of 1000".

Like I said, I'm happy that I'll already be able to join 4000+ games since I am pretty sure that's where I'd end up eventually, I'm just comparing it with the only thing I really know how to compare it to (iso) and the dramatic change was startling. I hope it doesn't sound like I'm complaining that anything is "unfair," it's just very different from my only real point of reference.

[shark_bait]

Going 9-1 (only loss to stef) in my first 10 put me at #14 or so on the leaderboard.  Two losses then dropped me about 900 points but I suppose the system is very quick to reward play and correct with large fluctuations.

[Warfreak2]

Isotropic's rating system was similar, in that you had a skill level and also a confidence. The difference was (and I assume it is a difference, but I don't think Goko have said how their ratings work), Isotropic gave you the lower bound of a 99% confidence interval as your rating, so that people who played fewer games were biased towards having lower ratings. I presume that was because an Isotropic rating was meant to be a status thing as well as for matchmaking.

[Polk5440]

I'm just comparing it with the only thing I really know how to compare it to (iso) and the dramatic change was startling.

This is a little bit of a false comparison because, presumably, you were slowly getting better over time on Iso. You switch to Goko, and you still have your higher skill level, so the rating system should more quickly recognize your higher underlying TrueSkill (*ahem*). If you had started a fresh account on Iso at some point (like dondon did, I think?) you would have found that you would have achieved the same rating much more quickly the second time around.

Having said that, I wouldn't be surprised if there actually is a difference, though.