I’ve worked with the data. It does tell you if a game was tournament or not. Also tells you which round the tournament game was played.
“tournament”: 2497,
“tournament_round”: 3, as an example…
When a game is not a tournament match, tournament: null and tournament_round: 0
People may discover in my they are tournament addicts and don’t put the same effort into friendly games while almost no effort goes into unrated games.
I will look at my code and see if there were any other ideas that might work you.
It would be really lovely if we could get the same statistics also for games against players of the same rank. In particular, I would like to know if I am better at black or white, but since often the weaker player is assigned black, the total statistic is worthless for this. Another thing that would be interesting would be the average rating of the opponents.
For example, on 9x9, my win ratio is very close to 50% because most of the games were games at the beginning when I played against players of the same strength. However, on correspondence and blitz the ratio is off. I think that on blitz my win ratio is higher because I play against lower-rated players on average, so it doesn’t really mean that my win ratio is above elo expectations. On correspondence however, I think that this is due to the fact that I rarely play correspondence games, so many of them are games that make my correspondence rank “catch up” with my live rank.
The best would be a win ratio that is weighted by relative ranks to see whether the player is underrated (i.e. sandbagging) or overrated.
Updates include performance in tournaments, statistic for even games only, and more stalker-friendly data like register date and average game per day.
The result looks quite promising, for example if you look at my statistics it’s quite clear that I’m a sandbagger, um, better at black than white, unlike what the total game charts show.
With this the planning queue is empty so I hope you guys can give me more ideas!
@Wulfenia: it’s actually very time consuming to get statistics for games against players of the same rank. Since OGS’ data only contains each player’s current rank.
In order to get games against player of the same rank at the time of the game I’d need to request information on each unique game to get the kifu. For a player with 5-6k games it would be a ridiculously long wait. So I could only provide data on even games, which is a good enough substitute I think?
The problem of whether a player is sandbagging is quite delicate and I’d rather leave it to the mods
Oh, interesting indeed…
I don’t know why is that exactly. Could be some kind of corruption of data when switching from nova to OGS. But then again my old account from 2011 wasn’t affected, it still show the correct registration date. So… I’m not sure xD
The players may be evenly ranked now, but they could have had a rank difference at the time of the game. Is this accounted for? I feel like it reverts back to the problem of not knowing the past rank details.
Don’t bother with a pie chart. Just list each flag and the player’s record against it.
By rule set… Korean, Japanese, Chinese, New Zealand, AGA
Rank difference per match… Am I punching above my weight? I’d expect to see something like +2.3 for the guy who is learning and picking on people his size or greater or -3.5 for the guy who hates to lose
Just wanted to get these ideas out on this pleasant Sunday evening.
I really resent your implication. You play against stronger players to learn and then disdain them for “picking on you”. On some other servers, it is almost impossible to get a game as a weak player because people will not offer to play against them at all. I make a lot of open unrestricted challenges and the recent batch had a solid 2/3 of them taken by weaker players, simply because there are more weaker players than stronger players. If you see a “guy who hates to lose” you might actually be seeing a guy who doesn’t take more than he gives from a community.
I blocked you now from my open game challenges to be sure that I will never be “picking on you”.
I think you’re being a bit too harsh here, @Wulfenia … maybe a misunderstanding? I had read @Professor_X’s comment as somewhat intentionally exaggerated:
On one side we have the player who ONLY plays (much) weaker players—could certainly be seen as somebody who doesn’t like to lose.
On the other side we have the player who ONLY plays stronger players (while at the same time stating that playing weaker players is “unfair” … as seen here recently).
Both kinds of players do exist … hopefully it’s just phases they’re going through, though.
I totally agree that it’s desirable to have a good balance of giving to and taking from the community … and I’m quite sure that Professor X sees it similarly.
Maybe I’m confused, but I read [quote=“Professor_X, post:56, topic:6524”]
+2.3 for the guy who is learning and picking on people his size or greater
[/quote] to mean that the weaker player is picking on the stronger.
That’s an unusual use of the phrase, since normally it’s the other way around, but that’s how it reads to me. I’m curious why @Professor_X put it this way.
I’m not a native speaker of English, but I have always thought that “to pick on somebody” not only means “to bully/harrass weaker ones” but could also mean “to challenge somebody” … was I wrong about that? (cf. “Pick on someone your own size!”)
As @trohde noted, I was giving an example of extremes. Most players would hover around plus or minus one stone.
You need to consider context:
Both @AlphaGo_2_0 and I were working on statistical programs independently. Each of us had no idea the other guy existed. One day, I discovered this thread. I was happy because I have no time to finish what I started. I had ideas that I thought @AlphaGo_2_0 would like so I shared them in the thread.
There are no implications. I happen to play exclusively correspondence tournaments. I expect that I would have something around +3 because I’m always getting my ass kicked. I thought it would give more insight for players. That is all.
Read deeply into your stones. Don’t read deeply into forum messages.
