For the purpose of a machine learning algorithm, the term "variable" would be correct. Are you using it in a different way?

by machine learning, you mean linear regression right?

He doesn't know what he means.

Yeh the exact number depends on what you define as a variable. However, in my database I can see what happens to every player every minute of the match for every player. This allows me to more accurately quantify the value each player bring to the team.

Based on that, I can more accurately set up a predictive model.

Thus, the majority of the variables do not directly impact the final probability prediction, however the vast majority have a minor impact on some of the prior calculations.

Dissapointed with your lack of conspiracy theory. Give me your best shot. When I am reading your comments, I think of a flat-earther so I had hoped you would give me something comparable insane.

Why waste your talents on sportsbetting when you are doing something that even the game developers can't do?

Did you get denied for spelling errors?

Game-developers are not spending a lot of ressources on the analytics-side of esports. They definitely could do the same/better if they hired proper statisticans, however it would require a lot of ressources for not that much gain. Thus, they rather spend more ressources on optimizing the playing experience for the playerbase.

That's not to say that we won't see more analytics in esports-broadcasts in the future (we definitely will), but it will most likely be outsourced by tournament organizers or game-developers (whoever has the rights).

Further, Riot's stats team definitely has a much larger database than I do since they have access to server logs. However, unfortunately they are not very good at using the data in a productive way. The player ratings model they created is heavily flawed.

some ppl so toxic in sports betting section on here

Are you talking about me? lol at least be a man about it

I'm just having fun with a 5 post troll

pretty sure that some ppl came with a direct stare followed by an eye roll.

who else would i be referring to?

get your test levels checked son

stop this passive aggressive ****

Dota.sqlite3.zip

I'm not using this anymore so I figured I'd post it in case anyone else wants to take a look. This is a link to a SQLite database (250MB zipped) that contains basically every tournament level DotA game and many lower level matches played since 2017 through last year's International (Oct 2019) where data is publicly available. 2016 and before is only "mostly reliable" because the data availability is not good.

There's data on the tournaments (prize pool, start/end dates, quality, etc.), game patches, individual players/heroes, teams, series of matches (Bo3, Bo5, etc. but data is spotty), the picks and bans for the matches, as well as data on the matches themselves (duration, teams, heroes, scores, etc.).

The link should be active for at least 30 days.

This thread seems to be pretty dead but I've put some work into creating a model for the LoL seasons so maybe that can generate some interest. Anyone ever looked at props in LoL or things like first bloods? That might be where I go next.

Well league of legends is now the biggest sport on draftkings. Anyone else been firing the 2AM chinese league slates?

You're absolutely toxic in every thread you post in and I have no idea why you insist on posting zero content of value while degrading every thread you post in into some sort of garbage shitfest of you insulting other posters.

I can't think of a single person that enjoys your posts or likes you here. Does anyone here enjoy lvr's posting? Anyone?

Your posts are garbage in the MMA thread, they're garbage in the politics thread, they're garbage here. You add nothing of value to this forum.

There, was that man enough for you?

is there a general consens on what the sharpest book for esports is ? is it pinnacle aswell ?

Think youre being a little too hard on lvr to be honest.

Saying someone has a model that uses "hundreds" (as in somewhere between 100 and 999) of 'variables' is pretty lol. Either theyre talking about two different things, or this guy really doesnt know anything about modeling.

I've been using Pinnacle lines and line movements to judge my bets, I think theyre likely the sharpest book.

Reread the actual context of the discussion. I wrote I have hundreds of variables in a database (so not just wins and losses as Ivr was implying).

So for instance in League of Legends I have data on XP, gold etc. for each player on a per minute basis.

Nowhere did I write that I use hundreds of features as input in a single machine-learning model. Although indirectly the amount of variables/different types of data points (whatever we call it) that have been processed through the process before a machine-learning model generates the final output is probably in the hundreds for some of my models.

How immediate is something like this? As in, how soon will I be able to bet on LoL at an actual Vegas casino/app?

They are hoping to have everything setup by the time sportsbooks re-open.

I can't get statsfair to return any results. Am I just a noob? Is it just for eSports?

Thanks

Anyone else watching MSC? I'm essentially fading LCK entirely.

Skimmed a little bit i'm sure it's in the thread but didn't see the answer to this readily at hand:

Best US facing site for starcraft 2 betting of reasonable size? Not interested in the other esports, asking exclusively about starcraft. I'm using 5dimes right now but they limit me to $100 a bet and frequently move the line 25 pts after i bet it. I'd even venture i'm one of the only people betting some of these lines sometimes.

Really feels like i have an edge even so, but not sure if it's worth my time looking for a book that will accept larger wagers on this.