Upvote Upvoted 28 Downvote Downvoted
TF2Elo Predictions
posted in Projects
1
#1
0 Frags +

I created a project for predicting the outcomes of ESEA Invite matches. For each team in Invite I calculate an Elo score. This lets me get a probability of a team winning a given match. Additionally I simulate out the rest of the season to get the chances of each team making the playoffs.

The project can be found here: https://tf2elo.com
I will update it weekly with predictions for the next week's matches.

To get the starting Elo scores I used the results of last season. New teams start with a score of 1500.
Currently the ELO scores have not stabilized yet, as the season hasn't started yet. As such the probabilities are fairly conservative, being closer to a 50/50 outcome.

I created a project for predicting the outcomes of ESEA Invite matches. For each team in Invite I calculate an Elo score. This lets me get a probability of a team winning a given match. Additionally I simulate out the rest of the season to get the chances of each team making the playoffs.

The project can be found here: https://tf2elo.com
I will update it weekly with predictions for the next week's matches.

To get the starting Elo scores I used the results of last season. New teams start with a score of 1500.
Currently the ELO scores have not stabilized yet, as the season hasn't started yet. As such the probabilities are fairly conservative, being closer to a 50/50 outcome.
2
#2
14 Frags +

holy heck that's a really cool project!

holy heck that's a really cool project!
3
#3
2 Frags +

Thanks! I got the idea from 538's sports predictions (https://projects.fivethirtyeight.com/2018-nba-predictions/)

Thanks! I got the idea from 538's sports predictions (https://projects.fivethirtyeight.com/2018-nba-predictions/)
4
#4
5 Frags +

31% chance esketit. can't wait to see how it'll look once the season starts

31% chance esketit. can't wait to see how it'll look once the season starts
5
#5
0 Frags +

faint is mispelled

faint is mispelled
6
#6
14 Frags +
telephone_fanfaint is mispelled

thats not faint gaming that's just myanimelistdotnet's dumb name

[quote=telephone_fan]faint is mispelled[/quote]

thats not faint gaming that's just myanimelistdotnet's dumb name
7
#7
0 Frags +

we get 8th

we get 8th
8
#8
6 Frags +
bleghfarectelephone_fanfaint is mispelled
thats not faint gaming that's just myanimelistdotnet's dumb name

wow thats fucking stupid

[quote=bleghfarec][quote=telephone_fan]faint is mispelled[/quote]

thats not faint gaming that's just myanimelistdotnet's dumb name[/quote]

wow thats fucking stupid
9
#9
0 Frags +

I found an issue with how I was handling ties in the simulation, which lead to teams with the same elo having wildly different chances of making playoffs. That has been fixed.

A couple questions though:
Is TFCrew the same(ish) roster as The Sparkle Gang from last season?
And is the Feint gaming roster completely new?

I found an issue with how I was handling ties in the simulation, which lead to teams with the same elo having wildly different chances of making playoffs. That has been fixed.

A couple questions though:
Is TFCrew the same(ish) roster as The Sparkle Gang from last season?
And is the Feint gaming roster completely new?
10
#10
5 Frags +

tfcrew is the same as sparklegang except they have KTB instead of etney on pocket .

feint gaming is the same roster as myanimelist / feint gaming from last season except I think they traded goldfish pocket and trip roamer for tojo roamer and trip pocket

tfcrew is the same as sparklegang except they have KTB instead of etney on pocket .

feint gaming is the same roster as myanimelist / feint gaming from last season except I think they traded goldfish pocket and trip roamer for tojo roamer and trip pocket
11
#11
0 Frags +

Awesome, thanks! I updated the site to reflect TFCrew's score from last season.

Awesome, thanks! I updated the site to reflect TFCrew's score from last season.
12
#12
5 Frags +

I don't mean to be rude because this a great project and I personally find it really cool... but what exactly is the difference between what you are doing and what's being done at tf2metrics? https://tf2metrics.wordpress.com/

I don't mean to be rude because this a great project and I personally find it really cool... but what exactly is the difference between what you are doing and what's being done at tf2metrics? https://tf2metrics.wordpress.com/
13
#13
2 Frags +
MysclickI don't mean to be rude because this a great project and I personally find it really cool... but what exactly is the difference between what you are doing and what's being done at tf2metrics? https://tf2metrics.wordpress.com/

First of all I had no idea tf2metrics existed, but it looks really cool!
There seem to be a few differences though.

  • TF2metrics gives rankings for individual players as well as teams. I only give rankings for teams.
  • TF2metrics' match predictions give an expected score for each team while I give a probability of each team winning. This is sort of a nitpick point though...
  • Also TF2metrics doesn't seem to give weekly playoff chance predictions, while TF2Elo will.

It will be interesting to see how the different ranking/prediction systems perform over the season.

[quote=Mysclick]I don't mean to be rude because this a great project and I personally find it really cool... but what exactly is the difference between what you are doing and what's being done at tf2metrics? https://tf2metrics.wordpress.com/[/quote]

First of all I had no idea tf2metrics existed, but it looks really cool!
There seem to be a few differences though.
[list]
[*] TF2metrics gives rankings for individual players as well as teams. I only give rankings for teams.
[*] TF2metrics' match predictions give an expected score for each team while I give a probability of each team winning. This is sort of a nitpick point though...
[*] Also TF2metrics doesn't seem to give weekly playoff chance predictions, while TF2Elo will.
[/list]

It will be interesting to see how the different ranking/prediction systems perform over the season.
14
#14
4 Frags +

Really cool website, allthough I question a few of the numbers, such as a team where no players have invite experiance having a 46% chance of beating last season's 4th place team.

Really cool website, allthough I question a few of the numbers, such as a team where no players have invite experiance having a 46% chance of beating last season's 4th place team.
15
#15
3 Frags +
Cornpop16Really cool website, allthough I question a few of the numbers, such as a team where no players have invite experiance having a 46% chance of beating last season's 4th place team.

clearly you didn't watch us play last night.

[quote=Cornpop16]Really cool website, allthough I question a few of the numbers, such as a team where no players have invite experiance having a 46% chance of beating last season's 4th place team.[/quote]

clearly you didn't watch us play last night.
16
#16
5 Frags +

it's because they're new to invite so until they win or lose 1500 will be their fixed elo

it's because they're new to invite so until they win or lose 1500 will be their fixed elo
17
#17
TF2Metrics
6 Frags +

Fellow who runs TF2Metrics here.

Seeing this inspired me to start making dedicated weekly match predictions again. Fight me.

For the sake of comparison, here's my own crystal ball's assessment of what Week 1 of Invite might look like, including most likely score outcomes and win chances. I think my system is quite a bit more cynical than yours by the looks of things.

I'll also point out that my system has the unfair advantage of about a year's worth of data on most of the players involved in Invite, plus it already has a little bit of knowledge about most of the new arrivals, too. This means if your system proves to already be more accurate than mine I'm gonna be grumpy.

Fellow who runs TF2Metrics here.

Seeing this inspired me to start making dedicated weekly match predictions again. Fight me.

For the sake of comparison, [url=https://twitter.com/TF2Metrics/status/954500576590647296]here's[/url] my own crystal ball's assessment of what Week 1 of Invite might look like, including most likely score outcomes and win chances. I think my system is quite a bit more cynical than yours by the looks of things.

I'll also point out that my system has the unfair advantage of about a year's worth of data on most of the players involved in Invite, plus it already has a little bit of knowledge about most of the new arrivals, too. This means if your system proves to already be more accurate than mine I'm gonna be grumpy.
18
#18
6 Frags +

0.17% chance I like those odds

0.17% chance I like those odds
Please sign in through STEAM to post a comment.