Interesting Chess: A site for games that were "Interesting"

Hi all,
I have made a website to show titled players game streaks that are "interesting" based on the probability of such streaks occurring (a player winning too many games in a row). Currently I only show streaks from GMs to IMs.

interestingchess(.)com

Please let me know what you think of it.

I am sorry, I am not sure what you mean Bricks.

Yeah the hyperlink is giving me trouble, maybe it is a bug. But I have added it in the body. Many thanks Bricks

Hi Hilton,

It's very secure as there are no accounts, passwords, payments or anything. It's a simple dashboard of scrapped games.

nice website it is quite interesting (as the forum's title says)

BTW,if I may ask,where did you buy a server and that website name?

And, how much did you cost buying that?

Hi MistMint,
I bought the domain in spaceship(.)com but you can compare prices at tldes(.)com, that is how I found it.
Regarding the server I am a dev with AWS certs, the scraping runs on ECS daily and spins down. The website runs from a static website in an S3 bucket with cloudfront in front. The Route 53 DNS service redirects the domain to the cloudfront distribution.

This is one days cost, I think I could reduce it in the coming days by moving away from serverless, I was focused on get this puppy running

The image does not load but its like $.3 USD per day which .2 is just the scraping compute. I could reduce that significantly. AWS has a free tier so it should be free for 6 months.
The domain costed around 16 USD or something like that per year.

Interesting project )
Looks nice

Could you explain a bit why case with 24 wins from 2610 vs 1548 on average is interesting?
(second row in the table)

Yeah I think there is more work to do on the UI and algorithm that was not considered.
If you go into that specific game the winner rating is quite different probably because its a different game mode.
So I am missing displaying the winner rating for that game mode instead of just showing the overall. Maybe even having "game mode" aware streaks instead of just sequential.
Definitely a worth while improvement

Hi Alexey I have implemented these changes:

1. Only 'normal' chess and 960 game modes are counted.
2. The ratings from each game are displayed to avoid confusions

Thanks for the feedback!

Cool project, Alien!
How about adding a radio button or smth to filter for rated games only — I imagine those might be especially interesting when looking at probabilities.
From what I’ve seen, the JSON data for the games includes a "rated" boolean, so implementing this filter could be quite straightforward.

mmm I was using /games/{YYYY}/{MM} instead of /games/live/{BASETIME}/(INCREMENT)
the monthly games do not have a rated column, I looked and could not find it.
Maybe I can rework the code to use games/live/{BASETIME}/(INCREMENT) and get the rated boolean.
Thanks for the suggestion!

/games/live/{BASETIME}/(INCREMENT) is bonkers, it just gives you all the data from years ago!
Amazing

Thats true, i didnt even know this endpoint existed. Thank you!

That being said, i think its pretty easy to overestimate the probability of < 1%. I just looked up my greatest rapid winning streak and did the math and came to the conclusion that it had a probability of 0.16%. Maybe i had something even more unlikely, but i do not think so. (I remember most of my long winning streaks.) With over 2300 rapid games i should already have had 4 of these streaks, but i only remember 2.

The more games you are looking at, the more likely it becomes to find this insane streaks. You are looking at probably multiple 100s of accounts with 1000+ games. It would be interesting to multiply the probability of the rarest one with the number of games you looked at, to see how often you should be finding a streak like that.

Maybe you could add a feature to enter a username and see their rarest streak.

Also always keep in mind that the combination loss loss win loss win win win loss win is equally unlikely in exactly this order than my 9x winning streak. The only diffrence is that nobody is interested in a random combination of games like that.

Also do not forget, that being underrated makes it more likely to win while our math models do not get that at all.

Hi there is a search bar for players name. Also I only process games from 30 days ago