G0tstats is back! (with more stats)

I’m not sure if the math works out the same regardless (it shouldn’t because it’s a ratio not a percentage), but what it’s actually saying is for each opponent you’ve played, you’ve had on average 1.75 games with them. (think of the cliche American dream where each family has 2.5 children).

Total no. of games played divided by total no. of unique users played.

You are most welcome :slight_smile:

OGS caps at 9 dan to keep with convention, but when the system was first introduced it caused a commotion because some of the bots rose to 11 dan (queue monocles popping out of eyeballs!) but AFAIK glicko can continue into infinity if you win enough games against strong enough opponents

Interesting :thinking:. How about on the low end, does it work going backwards as well? Could someone be 700K?

Yes, with a rating like 0.000000415 the rank would be 700k.
You can even get a rating low enough to break the OGS rating->rank formula getting imaginary ranks. (for ratings < 0)

4 Likes

Okay, that is a weird stat, since it is really isn’t the kind of thing people will find directly useful. Honestly, @Chinitsu I think it would be great if you could create a mouse hover text over each stat so that people could understand how the stat worked. Or at least for select stats, where it might be unclear, like this one.

Concerning your original question though, on how you could word it to be clearer, I would suggest: Average # of games played with each opponent. If that is too long, then: Average # of games per opponent, though this is still not much better than what you have now. That is, it I don’t think someone will read it and know exactly what the stat represents :thinking:.

tried it. 2 things i noticed that doesn’t seems right.

image
this is may stat in OGS page. I will take it as “official”


this is my gotstat page result.

  1. Note that I played 237 games in my OGS page, but it showed 293 games in Gotstats. Don’t know where the other 54 game came from.

  2. You can only see the %(48.5 : 51.5) but not the numbers in the Gotstats picture but when i move my mouse over it the # of games I played black vs white is exactly the number of my win/lose, ie 142:151. It would have been a very lucky coincidence if it is actually the case. But I would rather think it is some sort of a bug. (and form the OGS official page, my win/lose is actually 108:129)

1 Like

The number on OGS show “Ranked” games only, so if you uncheck the Unranked part you will now see that gotstats gives 265 ranked games. Still different from 237 on OGS profile page, the question should be where did those 28 games go? Probably one of OGS developers can help us answer that?

If you have doubt about the integrity of the stats, you can count the number of non-anulled ranked games in your profile page.

(technically th query is /games/?ended__isnull=false&annulled=false&ordering=-ended&page_size=50, meaning finished games that’s not annulled)

It have been a very lucky coincidence :stuck_out_tongue:

1 Like

now that you have mentioned it, i strongly suspect they are those time-out games that got annualed. i checked the math, i won all 28 of them. LOL.

Annulled game are excluded on both sites, so they should not be the cause.

Timed out games, though, might possibly be the cause.

I just checked my own stats, I have 1 game difference comparing with OGS profile, and coincidentally I also have 1 ranked game that I won by timeout. So maybe it’s the cause of the difference, I’ll need to investigate.

1 Like

I think this might be what is causing the discrepancy:

In short, games that are discounted by OGS due to the mass-timeout rule don’t actually get tagged as annulled.

4 Likes

OGS’s correspondence timeout is a little bit tricky, in case you don’t know it yet.
they are annualed but they won’t tell you, and are still marked ranked.

2 Likes

Hmm it would be hard to decide then, how do I know which one is timeout annulled, and which one is timeout but not annulled?

very hard. especially when you are importing it to a database.

the rule is: regular game timeout is counted, 1st game of a series of timeout in correspondence is also counted, but subsequent games are not.

the player can see that his rank is not changed when he won a timeout game, and as far as i know, that’s the only way to tell. if you are importing it to a database, you probably can’t tell the difference.

but then, maybe you don’t really to filter them out either. the rule is there to prevent “rank distortion”, but i think that probably would not be your concern.

2 Likes

If it’s rated and not anulled it’s in the rating history:

The rating history contains up to 5000 games of the selected category.

2 Likes

The link at the first post doesn’t work for me. Can someone post the link to the page please? Thanks.

Yeah the new link is https://avavt.github.io/gotstats/

@Everyone the scatter plot is up. I can already expect the complain about not being able to limit end date (to “zoom in”) but I’m having a little trouble with it T_T

4 Likes

The scatter plot is very informative. Thank you.

1 Like

@Chinitsu, it might be a good idea to edit your OP and change the URL to the current one, not-only-but-also because I just got this when I clicked the old link (b/c I wanted to know :sunglasses:)

Also, newcomers to the thread would probably anyway begin with the OP …

I tried to but I can’t, the edit button isn’t even there :sob:

1 Like

OK, I changed the link for you … let me know if you want any of the text changed also.

2 Likes

Thank you very much @trohde, the text is ok either way :smiley:

1 Like