Interest and brainstorming for a results-based analytical ranking system

[Joy]

Every system has issues, but there’s an infrastructure pipeline that needs to be up to mitigate the majority. The first is a global tournament registry that’s easy to interface with, which Sendou is currently working on. The next would be a player identification system, which we can do via splatnet. It’s one of the projects I’m working on still, here’s a screenshot (minus the Doxx Wintraders option, which I added for a joke)

This pulls matches, and will ideally be configured to make it so easy for matches to be uploaded that there’s no reason not to. That would push matches to a global match database, similar to stat.ink, but with full integration with sendou’s tournament system. This also verifies all players participating. From there, pulling results is automatic.

As far as issues, there’s going to always exist tradeoffs and the systems won’t be especially accurate. The system we have now (or more accurate the lackthereof) that’s results based has many issues as it stands, especially when players “estimate” their own skill level. Worst case, the rankings can be fuzzified into a tiering system where thresholds can be identified akin to LUTI divs

 

[Abyss]

Every system has issues, but there’s an infrastructure pipeline that needs to be up to mitigate the majority. The first is a global tournament registry that’s easy to interface with, which Sendou is currently working on. The next would be a player identification system, which we can do via splatnet. It’s one of the projects I’m working on still, here’s a screenshot (minus the Doxx Wintraders option, which I added for a joke)View attachment 8999
This pulls matches, and will ideally be configured to make it so easy for matches to be uploaded that there’s no reason not to. That would push matches to a global match database, similar to stat.ink, but with full integration with sendou’s tournament system. This also verifies all players participating. From there, pulling results is automatic.

As far as issues, there’s going to always exist tradeoffs and the systems won’t be especially accurate. The system we have now (or more accurate the lackthereof) that’s results based has many issues as it stands, especially when players “estimate” their own skill level. Worst case, the rankings can be fuzzified into a tiering system where thresholds can be identified akin to LUTI divs

That's really cool. Kinda wish Nintendo would give us an actual open API but oh well.

Is your project open source? would be kinda curious to poke around.

 

[Joy]

That's really cool. Kinda wish Nintendo would give us an actual open API but oh well.

Is your project open source? would be kinda curious to poke around.

It is! Here’s the splatnet API I developed for this: https://github.com/cesaregarza/SplatNet3_Scraper

And here’s my progress on the GUI: https://github.com/cesaregarza/DataZipcaster/tree/feature/gui

 

[DoubleDrewski]

Every system has issues, but there’s an infrastructure pipeline that needs to be up to mitigate the majority. The first is a global tournament registry that’s easy to interface with, which Sendou is currently working on. The next would be a player identification system, which we can do via splatnet. It’s one of the projects I’m working on still, here’s a screenshot (minus the Doxx Wintraders option, which I added for a joke)View attachment 8999
This pulls matches, and will ideally be configured to make it so easy for matches to be uploaded that there’s no reason not to. That would push matches to a global match database, similar to stat.ink, but with full integration with sendou’s tournament system. This also verifies all players participating. From there, pulling results is automatic.

As far as issues, there’s going to always exist tradeoffs and the systems won’t be especially accurate. The system we have now (or more accurate the lackthereof) that’s results based has many issues as it stands, especially when players “estimate” their own skill level. Worst case, the rankings can be fuzzified into a tiering system where thresholds can be identified akin to LUTI divs

This is beyond dope. Love that it's open source as well.

 

[Joy]

I’ve been working on this for about a year, I needed a production-grade Splatnet 3 scraper and s3s wasn’t up to the task, so it’s kind of a behemoth project to do alone (SplatNet3_Scraper alone is over 10 thousand lines of code!). Slowly working up to it though, a lot of my time is also taken up analyzing SendouQ’s performance and other miscellaneous projects.

The GUI is taking especially long since as a data science person, GUIs and websites are not my forte, and I have a vision for it before it can be released. It needs to update itself, and it needs to be an executable (no installing Python directly). Without these, it’s not ready to release. I appreciate any help I can get, I’m just a stickler for my code’s structure lol.

I’ll eventually make a post with my grand vision, but baby steps.

 

[DoubleDrewski]

I’ve been working on this for about a year, I needed a production-grade Splatnet 3 scraper and s3s wasn’t up to the task, so it’s kind of a behemoth project to do alone (SplatNet3_Scraper alone is over 10 thousand lines of code!). Slowly working up to it though, a lot of my time is also taken up analyzing SendouQ’s performance and other miscellaneous projects.

The GUI is taking especially long since as a data science person, GUIs and websites are not my forte, and I have a vision for it before it can be released. It needs to update itself, and it needs to be an executable (no installing Python directly). Without these, it’s not ready to release. I appreciate any help I can get, I’m just a stickler for my code’s structure lol.

I’ll eventually make a post with my grand vision, but baby steps.

I can’t say I’m well versed in GUI, but have you already tried outsourcing help for that aspect specifically?

also as a side note, every sane human being in the modern age should have python on their computer imo lmao. It’s just too useful.

 

[guster-animations]

this is a really cool idea and it’s much needed! it’s much easier to get than an estimate for a luti div, too