More on A Curious Case Study in KGS Ranks

[Bantari]

And one more thing... speaking of noise.

One account which sometimes plays seriously and sometimes casually - this increases noise.
Two separate accounts, one of which is always serious while the other always casual - this reduces noise.

Makes me also doubt you understand what "noise" is.

[oren]

One account which sometimes plays seriously and sometimes casually - this increases noise.
Two separate accounts, one of which is always serious while the other always casual - this reduces noise.

The "Two separate accounts" is my strategy right now.

[RobertJasiek]

We are looking at your actual games, and your actual win rates.

I see.

Your claims of long strings of high win rates on kgs simply don't exist

It seems to depend on what we call "long".

I forgot how many weeks it was, but it was long enough to call it "long" when referring to my high win rates in serious periods of reaching a higher rank BY PLAYING (in contrast to the other case where reached a higher rank almost BY NOT PLAYING).

When I speak of long strings more generally (other in the cases when I actually had them), I mean also those strings I (or other frequently playing players) would need to reach the next rank again by playing and as soon as possible. After every period of winning roughly 50%, then losing a rank within days (e.g., after having had a few bad nights), a frequently playing player then needs a couple of weeks to go up one rank again, if immediately he starts and maintains a serious period of beating the rating system by greater stanima.

How well you perform in real life games should have as much weight on your kgs rating as how well you play backgammon or how well you cook a ribeye.

I know. Server ratings are DEFINED to be independent of real world games. I say: instead of taking pride in creating different ratings with little meaning (and much system theory to hide this failure as well as possible), any server rating system ought to be as meaningful (within its volatility) as player strengths in the real world. A server rating system doing worse (e.g., because of modelling only 2/3 of all players well) is a failure. Any rating system must model ALL players well, even the bot players with totally unexpected but well described behaviour.

[RobertJasiek]

If you are concerned that your fun games are dragging down your rank, have you considered using free games instead of rated games for that purpose?

Of course. But... for me a not rated game lacks sufficient motivation; I would win even less. Playing only rated games keeps a good motivation.

(There are exceptions, such as special tournament games, where motivation comes from the tournament, while additional rating would disturbe it.)

[RobertJasiek]

Start beating 5d players regularly (or just play them more and beat them), stop losing to 3d players (important!), and you will get to 5d. That's the trick.

This is INSUFFICIENT! Let me repeat: a frequently playing player also needs to win a high percentage of such games for a long period of time / for many successive games. So long / many that usually it can only be achieved with more mental energy and time investment than one can take away from, e.g., one's job and put into the rating fight. (Alternative: don't play for 3 months, then win literally only a few games.)

I don't see how you can expect the system to make you 5d when you lose so often to 3d players. Not to mention 4d players.

I expect the system to acknowledge serious periods during which I win frequently, acknowledge in a manner that is not so strict to be a real competition with the job etc. When I had no job yet, I could afford to spend a couple of weeks almost exclusively on sleeping and fighting for the KGS rank.

BTW, there is a tougher example: apukam alias Tibor Pocsai has been stuck at 4d for a longer time than I. He is stronger than me (I know: he demonstrated it in a real world tournament game), and he was European Champion. A rating system letting such players be 4d is a failure. There is no way around it. A failure. It is the system that is the failure, not the players.

If you play strong enough, you get there

And if you are not a frequently playing player frustrated by the rating system's too great stability and its requirement to invest more effort and time than one possibly has.

How many of your wins can be contributed to such external factors?

When I have time to do such statistics, I might do them. By interpreting the games tactically and strategically in relation to a player's inconsistency during the same game.

there are plenty of people who made it from 4d to 5d

Yawn. I did a few times, to mention an example. Therefore I KNOW how it works for usually frequently playing players:

Either do not play for months then win a few games or invest ALL one's effort and time in reaching the goal.

I also know how to become KGS 6d. Easy: I created a test account, just to see that it works.

people graduate up the rating ladder all the time on KGS, when they play strong enough to reach the next level from wherever they are. This includes frequently playing people, seldom playing people, tired people, drunk people, and anything in between. Even you cannot refute this simple truth which is easily verifiable.

Ugh, so - for the frequently playing people - they must have had months of free time to play the necessary games.

Since you don't seem to be able to make it, to me it clearly means that you are not strong enough.

No, it means that I lack the necessary effort and time, which I cannot subtract from my work.

(I played without problems as 5d on KGS for quite a few months. The problem is not to play as 5d, but the problem is only to BECOME 5d again.)

you could surely try harder

Currently no. At the moment, my work does not provide enough free effort and time for fighting against the KGS rating system.

And this does not change the fact that - the way you play, you play as 4d, period.

Yeah, games like winning very early, then being bored to death by the opponent, until he has made me so tired of boredom that he kills me (what I saw coming 100 moves earlier). That's what happens when the system forces to play against too many weak players. There are limits to which one can bear boredom to maintain concentration.

The last loss of this kind in a real world game was as real world 3d. My second last loss of this kind is even much longer ago. Such things can happen to me only on servers with bad rating systems, which cause boredom and frustration artifically.

once I start trying harder and playing stronger - my rating will go up.

Are you, or are you not, a frequently playing player on KGS?

If you play like a 4d, this is where you land.

This is not the problem. The problem is that this is where I remain too long, because the system makes it by far too hard to improve a rank, and makes it by far harder than losing that rank.

I am not sure why would you even want something different

To make improving exactly as easy / difficult as dropping a rank.

It seems to me that what you are saying is that you can try hard and play like a 5d for short spurts, and therefore the system should upgrade you to 5d. This is not how it works, nor should it. Because, by definition, then you fall down again

When I am 5d and play like a 4d, I should drop and be 4d. When I am 4d and play like a 5d, I should raise and be 5d. But this is not how it works. Only the former applies "well".

If the rating system was so sensitive, lots of people would be overrated

There are lots of misrated players, regardless.

There is nothing wrong with creating two accounts, one for serious games one for casual ones.

Since the rating system failure has not been fixed for over a decade now, it seems that I need to use this and similar options, which I dislike, also because they contribute to making things worse.

You are smart enough to know all that, no?

I also know how to cheat the rating system well to get higher ranks on average. Many people know: they create a new account as soon as the old one burnt (stuck at a too low rating).

[RobertJasiek]

when you play only casual and fun games you play like [...], when you play seriously you play like [...], and when you play a mixture of both modes, you play like [...]. This is how it can hypothetically look when taken your history into account, and this is pretty much what you are saying as well. Now what you seem to want is a system which lets you generally play in the mix mode but ranks you as if you were constantly in the serious mode.

No. I want that the rating system acknowledges my change from a non-serious period to a serious period within ca. 100 successive games (69 wins : 31 losses) instead of ca. 400 successive games (276 wins : 124 losses).

but calls this level 5d rather than 4d

No.

The only other option is to implement a system

No. By far not the ONLY.

E.g., the system can be made 4 (reason see above) times as volatile for frequently playing players, while keeping (or reducing) volatility for the rarely playing players, with reasonable degrees in between.

[Boidhre]

I know. Server ratings are DEFINED to be independent of real world games. I say: instead of taking pride in creating different ratings with little meaning (and much system theory to hide this failure as well as possible), any server rating system ought to be as meaningful (within its volatility) as player strengths in the real world. A server rating system doing worse (e.g., because of modelling only 2/3 of all players well) is a failure. Any rating system must model ALL players well, even the bot players with totally unexpected but well described behaviour.

Server ratings are no more meant to model your over the board strength than the EGF system is meant to model your online strength. I cannot understand the issue, if you're not playing as well on KGS as in over-the-board tournaments, why should your online rating be higher than your performance merits?

[Uberdude]

I know. Server ratings are DEFINED to be independent of real world games. I say: instead of taking pride in creating different ratings with little meaning (and much system theory to hide this failure as well as possible), any server rating system ought to be as meaningful (within its volatility) as player strengths in the real world. A server rating system doing worse (e.g., because of modelling only 2/3 of all players well) is a failure. Any rating system must model ALL players well, even the bot players with totally unexpected but well described behaviour.

Server ratings are no more meant to model your over the board strength than the EGF system is meant to model your online strength. I cannot understand the issue, if you're not playing as well on KGS as in over-the-board tournaments, why should your online rating be higher than your performance merits?

Ego

[SmoothOper]

I wonder how many people have multiple accounts on KGS, and what does that do to a predictive ranking system, and would people playing a secondary weaker account be considered sand baggers?

[Mef]

I wonder how many people have multiple accounts on KGS, and what does that do to a predictive ranking system, and would people playing a secondary weaker account be considered sand baggers?

As long as each account is playing to win every game (e.g. you aren't intentionally losing games to your stronger account with your weaker account), it's a non-issue.

[RobertJasiek]

why should your online rating be higher than your performance merits?

Not higher, but as low or high. A rating that does not reflect high winning percentages (during successive weeks) properly does not measure performance properly. Instead, KGS measures EARLIER performance (history) too much for frequently playing players.

[Mef]

No. I want that the rating system acknowledges my change from a non-serious period to a serious period within ca. 100 successive games (69 wins : 31 losses) instead of ca. 400 successive games (276 wins : 124 losses).

As the previous thread showed, in the last 8 years your 100 game moving average win rate has only been 70% on four occasions. On all of them you got promoted. Your claims of 400 game requirements are utter nonsense.

[RobertJasiek]

Elsewhere you can read the figures why an order of magnitude of 400 is the reality. It depends on how frequent a frequent player plays and on where his rating was when he started. IIRC, I needed to play over 300 successive serious games to improve less than 1/3 of a rank to reach 5d again. For a full rank, many more games would be needed. (If you want to see the exact numbers, you need to search. Currently I lack time to dig out them.)

[Mef]

IIRC, I needed to play over 300 successive serious games to improve less than 1/3 of a rank to reach 5d again. For a full rank, many more games would be needed.

Yes, but you omitted an important detail: The reason it took 300 games is because you weren't winning them. If you have a win rate <.66 you shouldn't be expecting to promote.

[RobertJasiek]

IIRC, the win rate was ca. 70%, anyway enough to improve a rank, because I WAS promoted a rank.