If I copied TLPD but for Smash players, would anyone use it?

[FoxLisk]

Hey all. If I built something like the TLPD for smash players, would people use it?

The TeamLiquid Progamer Database is a comprehensive, searchable history of roughly every game of Starcraft and Starcraft II ever played in a serious tournament. Please check it out if you have not done so before.

The most interesting things it offers, in my eyes, are player win rates, including by-matchup, by opponent, by map, etc.

It also offers an ELO rating (easily calculable with all that information centralized), recent and lifetime accomplishments, links to videos, some other stuff.

Smash doesn't really have anything like it. AllIsBrawl kinda supports some of these things, but my experience with it is that it is tremendously lacking, and, naturally, focused on brawl.


So I have two real questions:
1) Would the general smash player be interested in this information being compiled like this?
2) Would TOs be willing to help post more information?

For (1) I'm just trying to gauge interest.
For (2) I'm envisioning - this is still a gleam in my eye - that I could hopefully provide a way to extract this information automatically from TIO files. Stage and character information would be harder to support, at least initially. I don't know who made TIO but maybe I could get in touch with whoever did and talk about building additional optional support into the product or whatever. Otherwise this is obviously a lot to ask.

So. Should I build it?

 

[Sephirothxxx]

That sounds great, though pretty complex. I think people would use it.

 

[Gea]

Look at it this way. TIO was made and just about every TO uses it. If you added in things like seeding by variables and it was compatible with TIO it would be pretty amazing.

NealDT made TIO. He also founded AiB. Check there to contact him.

 

[Neji]

I think this would be an amazing idea. I would definitely use this, but it seems like it would take A LOT of work to make.

 

[Geenareeno]

This would be crazy good for the community but also tons of work and not perfect due to people not submitting results.

 

[Dan Salvato]

This would be crazy good for the community but also tons of work and not perfect due to people not submitting results.

I think most tournament hosts are likely connected to the Smash community pretty well and would thus be willing to cooperate. Plus, if the service was well integrated with TIO, eg. being able to easily upload results, then that could greatly expand the number of people using the service. I think it could become a great tool that provides insight and puts us another step forward in the fighting game community.

 

[Zodiac]

I have no idea how I would use it as a player, but it would be pretty cool!

 

[Teczer0]

This sounds similar to Nintendude's idea for an ELO ranking system and could be really cool.

Though it does sound incredibly hard to implement.

 

[Massive]

The hardest thing about this would be gathering sufficient data to create accurate scores.
You're going to need a large enough data set per player (statistically speaking it will need to be >31 data points per player) for the scores to approach a normal distribution and receive accurate predictive measurements of player ability/matchup.

There's also issues with regional balancing. There are so many smaller tournaments held that, if submitted, could skew scores inaccurately for players in a region/grouping. The way around this would be some sort of regional weight or tournament weighting system, and I can see that being a majorly contested thing.

We might want to consider an alternative to ELO ratings as well based upon to the fact that they could be skewed by matchup differences. ELO ratings were intially designed for chess, where nobody has any type of advantage based on what side of the board they're on. In smash, we have picks and counterpicks as well as stage differences. Creating an average score is a decent metric, but if someone is hard countered by a stage/character the ELO predictions/values would break down and yield less-viable data.

It's not impossible, but it would be quite arduous, good luck.

 

[Composeur]

Still, it would be a strong incentive to submit results if the data were to be used to calculate the seeding national tournaments. SSBM lacks a global centralized accounting of its tournaments and their contestants, and I think that having one would give the game's community much more legitimacy, both in the eyes of others and to ourselves.

It would definitely be a lot of work, but to those who have never met him, I know FoxLisk, and he is an extremely intelligent individual more than equal to the task. I'm virtually certain that he could follow through, with enough support.

 

[FoxLisk]

sweet. i guess i'll start working on it this week and see what I can do. I'll let y'all know when i have anything worth showing off.

 

[finalcloud13]

Yes yes and yes.

 

[Fregadero]

sweet. i guess i'll start working on it this week and see what I can do. I'll let y'all know when i have anything worth showing off.

Good **** dude this sounds like an awesome idea.

 

[Battlecow]

Yeah, I'd use this.

 

[Arcansi]

ELO ratings were intially designed for chess, where nobody has any type of advantage based on what side of the board they're on.

This is incorrect. Played perfectly, one side will always win. I'm pretty sure it's white but idk.

 

[FoxLisk]

This is incorrect. Played perfectly, one side will always win. I'm pretty sure it's white but idk.

This is not known (and arguably unknowable; it's possible that chess is unsolvable). White has a significant but not insurmountable (~4-6%, adjusted for ELO differences) lead in wins at very high levels of play.

About ELO scores:

I actually think it's kind of funny that everyone jumped on that aspect, since I think it's the least interesting part of the TLPD. However, I realize that the TLPD is a PRO database where this can easily be extended to basically everyone, which makes it a little more interesting. I haven't given it a ton of thought, but I will say a few things:

stuff about ELO ratings

I've typed and deleted a lot of words. Basically: Your concerns are all kind of right but mostly misguided, and I think they're missing the point of what an ELO rating system is for - that is, an estimate of each player's skill relative to the player pool. If you want to address anything in greater detail, PM me or AIM me or something.

 

[Massive]

If I'm not mistaken the white side advantage is not inherent to the white pieces, but based on the tradition of white moving first.

The only viable argument I've heard of chess being non-solvable is based on the amount of computation time required. It IS solvable, the problem space is just very, very large. I suspect it will be one of the first time intensive problem-sets to fall when quantum computing comes into its own.

I've typed and deleted a lot of words. Basically: Your concerns are all kind of right but mostly misguided, and I think they're missing the point of what an ELO rating system is for - that is, an estimate of each player's skill relative to the player pool. If you want to address anything in greater detail, PM me or AIM me or something.

I understand the want for a player rating metric and ELO is probably as good an option as any. I have a bad habit of thinking of caveats for most things, and it was not meant to dissuade you or anyone from implementing anything, more as a cautionary statement. Carry on, lol.

 

[zZz]

Two words.

#^$& and Yes

 

[Bones0]

Would this be linked to videos of said matches? I think it'd be cool if you could go search for a tournament (recent or old) and look at the bracket, and then each match within the bracket that has a video can have a little red recording dot or something in the top corner. You can then just click the corresponding branch of the bracket to pop up that match's video from YouTube. Obviously this would probably be pretty complicated, but I figured I'd just throw it out there. Good luck.

 

[FoxLisk]

I'm working on a parser for the Tio files at the moment, and if anyone has an example of a saved Tio file from a real tournament that I can test it on I would be very grateful.

I, of course, have files I've created for this purpose, but I've never hosted a tournament so I have no idea if I'm testing realistic use cases.


UPDATE: I'm on track to have a first pass of this ready in the next week (although i haven't looked into hosting it anywhere yet, that might take a little longer), and it won't have anything fancy - just players with lists of their recent sets, who they played against and whether or not they won. searchable by player & by tournament, with whatever other information & niceties I feel like throwing on it before getting it out for people to see.
Also it will look like **** because I have no design ability at all.

Would this be linked to videos of said matches? I think it'd be cool if you could go search for a tournament (recent or old) and look at the bracket, and then each match within the bracket that has a video can have a little red recording dot or something in the top corner. You can then just click the corresponding branch of the bracket to pop up that match's video from YouTube. Obviously this would probably be pretty complicated, but I figured I'd just throw it out there. Good luck.

Maybe. Certainly not in the first draft. I'd love this but it's not a top priority.

 

[Composeur]

Wow, ****'s getting serious, yo.