Kimarite trends website

[Octofuji]

Some of you might be interested in this interactive graph showing Kimarite trends over the years.

 https://sumo.stuartmcgill.org/

The data is from @thatsumoguy's  https://www.sumo-api.com/ and I'll refresh after each basho.

It just about works on a mobile but really the graphs are best viewed on a bigger screen. If you find any bugs please let me know - I'm sure there will be some :)

[serge_gva]

Extremely useful and interesting link, thank you!

I'd like to suggest two functions: a curve symbolizing the trend to reduce the variability that can exist from one basho to the next, and/or the possibility of choosing, on the X axis, a longer time (for example kimarite per year, or even per decade) for each point.

[Octofuji]

On 27/01/2025 at 09:00, serge_gva said:

Extremely useful and interesting link, thank you!

I'd like to suggest two functions: a curve symbolizing the trend to reduce the variability that can exist from one basho to the next, and/or the possibility of choosing, on the X axis, a longer time (for example kimarite per year, or even per decade) for each point.

Thank you very much for the feedback!

Grouping by year should be fairly easy to do. Maybe that should be the default option if the period selected is more than e.g. 10 years.

With the curve it's a little tricky. The idea for this tool was to provide something impressionistic, to give an idea of trends over time. It doesn't allow you to drill down into the data like e.g. Yubinhaad's posts. So definitely, reducing variability feels like the right thing to do. But at the same time, it feels odd for the curve not to pass through the actual data points... I think though that the "grouping by year" idea will help with this.

The other thing it really needs is URL query params, to allow external links to a selection of criteria (similar to Sumo DB queries). Again that shouldn't be too hard.

I'll post back in here when these are done.

[Octofuji]

I've added support for URL query parameters - so you can click on the "Share" button and it will copy a URL with the selected criteria to the clipboard.

I think next I'll try and add a regression line to go with the dataset, which (assuming it's appropriate) will smooth out the variability.

https://sumo.stuartmcgill.org/?display_as_percent=0&from=2015-05&selected_divisions[0]=Makuuchi&selected_types[0]=Hikiotoshi

Edited May 5 by Octofuji

[shimodahito]

very interesting... no hikiwake

[Octofuji]

2 hours ago, shimodahito said:

very interesting... no hikiwake

Not since 1974

 

[Octofuji]

On 27/01/2025 at 09:00, serge_gva said:

Extremely useful and interesting link, thank you!

I'd like to suggest two functions: a curve symbolizing the trend to reduce the variability that can exist from one basho to the next, and/or the possibility of choosing, on the X axis, a longer time (for example kimarite per year, or even per decade) for each point.

I've added the option to include linear regression plots, which should help to see the wood from the trees.

My statistics are very rusty, but from what I remember this is a reasonable use case.