• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Dota 2 |OT2| Ellen Pudge starring in Beyond Two Throws

Status
Not open for further replies.
shira

shira

Member
Some Bo5 GG's today

live!
ibtSYOBEU5ZV7w.png

Liquid #8 vs #7 Fnatic
for ticket to Vegas/$2.5k guaranteed
http://www.twitch.com/d2l

wtf delayed?
t1386271800z1.png

icb0FJVAr27cu.jpg

Speed #4 vs #9 Sigma
for D2CL semifinals and $15k+6.25% guaranteed and a shot at [A]
http://d2cl.org/
 
U

Ultrabum

Member
LiQuid! said:
No lie, Carrie Underwood's voice is like nails on a chalkboard to me. I have to mute the intro to Sunday Night Football now or else my dick goes limp for one full U.S. week.
Click to expand...

U.s. Week different length of time compared to other countries.

Also, I have the same problem. More like Carrie NOMOREwood, am I right?
 
S

SteveWinwood

Member
oh god apparently ༼ つ ◕_◕ ༽つ GIVE FROSTIVUS !!! is a thing now

Pratfall said:
Cool, thanks. I considered aghs and then thought we needed a bit more right click and went butterfly. Now that I'm thinking about it, the stats and survivability from aghs would have helped us more probably. Also, whenever I got a 1-2 man gale into 4-5 man nova off, even if I died immediately after, we just crushed the teamfight so the Aghs would have helped there too.
Click to expand...

For the record, I don't think a satanic is a particularly good item on veno, but if you want tank and already have a helm, sure why not. The butterfly gives you SOME armor so it was probably a fine pick up anyhow. If you want some cheap armor, and make your ult even better, get that veil of discord. So crazy good. Everyone dies.

If you get that veil off for each tick of the ult with an aghs, you do 2160 points of damage. To every person. What.
 
B

balladofwindfishes

Member
Honestly, all Valve has to do for Frostivus is release a hero/heroes at a time between now and Christmas and release with it a wintery themed crate and I think the community would consider that fine.

I think the majority of Diretide stuff was not specific to the game mode, they just wanted anything at all.
 
Emerson

Emerson

May contain jokes =>
balladofwindfishes said:
Honestly, all Valve has to do for Frostivus is release a hero/heroes at a time between now and Christmas and release with it a wintery themed crate and I think the community would consider that fine.

I think the majority of Diretide stuff was not specific to the game mode, they just wanted anything at all.
Click to expand...

Give me the winter map and a Santa hat for Pudge and I won't complain at all.

Like you said I don't care much for special modes, I just love themed maps, themed cosmetics, etc. Just do something.
 
A

Anbokr

Bull on a Donut
Trasher said:
The problem with DotA is that there is no good way to quantify a rating. There are far too many factors and variables going on all at once. The one place where a rating does make sense is where it already exists in the game: TMM. At least in there you have a somewhat controlled sense of setup in having two teams actually working together. That isn't a given outside of TMM in normal MM. In regular MM you can have random mixed parties, people trying to teach their friends the game who are way below their skill level, people training on Meepo, people griefing their carries while playing support, and so on. There's no way to account for all the crap that goes on there.

The differences between TMM and regular MM are huge though. It's too bad that this is the state of the game, but the main issue is that the vast majority of players don't want to play in a TMM setting. Not only is it hard to even setup a team, but the pool of players to play against is too small for all of the different skill tiers. And I think that's what scares most of the lower skill-tiered players away from even giving TMM a shot. The TMM style of play is probably the best way to actually learn the game and get better though. Knowing how to actually play as a team is what most people are lacking in DotA. It is far too rare to actually find a good game in regular MM where you either A) don't have that selfish person/persons on your team that bitch and refuse to work with the team, or B) have someone on your team that virtually functions as cement shoes because they have no idea what's going on.

Anbokr isn't aware of these issues because he never tries to play with people below his skill tier. He only enjoys playing in try-hard 4-5 stacks where milkman carries his ass to victory.
Click to expand...

1) milkman rarely mms with us
2) Oh I'm aware. It's just like any other team game, CS:GO, LoL, HoN, etc... of course the rating isn't perfect, but that doesn't mean it isn't useful. In HoN, 90% of the time someone with a higher rating than you was generally better than you. Same deal with CS:GO. Hell, there were solo queue pub stars like CRUSHED with higher elo than pros who stacked (and it showed, he CRUSHED them pretty hard). I've also heard good things about the LoL system from friends though I can't speak from personal experience there.
3) TMM is complete dog shit for many reasons you outline. Also, starting off teams anywhere from 0 to 5000 kind of defeats the purpose of a leaderboard. You play 3 games to determine your rating and your done. I appreciate the premise, but it's just not going to work. Hell, maybe make the ranked mode CM/CD only to maintain some of the TMM feels idk.
4)
EQQdafw.png

ur garbage kid get good
 
M

M.D

Member
So, I'm playing with an alch who got tranq boots and managed to stun himself even tho he had at least 2 targets to stun...
 
Wok

Wok

Member
Haly said:
He describes his models as "hyperdimensional", which I take to mean it's more than just a line plot and he's factoring game time as well.

I agree with Cyborg in that he is clearly not confident in his method enough to reveal the formulae (although he might want to preserve it for commercial use), and the constant changes support this as well. This is very much a work in progress and DOTA2 players are the guinea pigs.
Click to expand...

Anbokr said:
Also he could simply be hiding his formulas/method in order to protect it. Valve crushed DBR fairly quickly and if he was completely transparent with how he calculates DSR, Valve could attempt to find a way to quash that or throw a kink in there.
Click to expand...

It is a machine learning algorithm, there is no commercial use whatsoever in the sense of a patent. He has just applied it, okay.
http://en.wikipedia.org/wiki/Support_vector_machine

It is a formidable blackbox, but it can totally fail if the training is not done with care.

I take it he used the games of the top players of dotabuff as training examples of pro games to allow a supervised clustering.
 
S

Satch

Banned
Nirvana said:
Are you playing Dota on Xbox Live? I remember this happening when I played Xbox, but Dota players are generally more concerned on whether or not you bought boots first instead of regen.
Click to expand...

im always shocked by it too tbh

most people are usually not crude but when there are assholes present, they are HUGE assholes. there is no middle ground for some reason
 
B

BigAT

Member
Anbokr said:
1) milkman rarely mms with us
2) Oh I'm aware. It's just like any other team game, CS:GO, LoL, HoN, etc... of course the rating isn't perfect, but that doesn't mean it isn't useful. In HoN, 90% of the time someone with a higher rating than you was generally better than you. Same deal with CS:GO. Hell, there were solo queue pub stars like CRUSHED with higher elo than pros who stacked (and it showed, he CRUSHED them pretty hard). I've also heard good things about the LoL system from friends though I can't speak from personal experience there.
3) TMM is complete dog shit for many reasons you outline. Also, starting off teams anywhere from 0 to 5000 kind of defeats the purpose of a leaderboard. You play 3 games to determine your rating and your done. I appreciate the premise, but it's just not going to work. Hell, maybe make the ranked mode CM/CD only to maintain some of the TMM feels idk.
4)
EQQdafw.png

ur garbage kid get good
Click to expand...

Too bad that all the games of Team Trash carrying him in the mid tier tournament don't count, otherwise he might have a half decent rating and not be in garbo tier.
 
A

Anbokr

Bull on a Donut
Wok said:
It is a machine learning algorithm, there is no commercial use whatsoever in the sense of a patent. He has just applied it, okay.
http://en.wikipedia.org/wiki/Support_vector_machine

It is a formidable blackbox, but it can totally fail if the training is not done with care.

I take it he used the games of the top players of dotabuff as training examples of pro games to allow a supervised clustering.
Click to expand...

I'm not talking about protecting it for commercial or patent reasons. I'm talking protecting it from Valve privatizing even more stats in an attempt to quash it. I.E his data relies on X, Y, Z, P, A to calculate his rating, Valve disables X from dotabuff/profiles.
 
B

balladofwindfishes

Member
Dotaskill dude left his database and source code wide open for a few minutes.

It just takes your win rate, KDA, GPM, XPM and enemy's average win rate and figures out your "skill" via that.

Which means low kill, low farm games (for example, if you're a support) probably count against you even if you did everything right. It doesn't appear to take wards or anything into account, so it's very carry biased.

An entirely worthless stat.
 
S

Sibylus

Banned
balladofwindfishes said:
Dotaskill dude left his database and source code wide open for a few minutes.

It ust takes your win rate, KDA, GPM, XPM and enemy's average win rate and figures out your "skill" via that.

Which means low kill, low farm games (for example, if you're a support) probably count against you even if you did everything right. It doesn't appear to take wards or anything into account, so it's very carry biased.

An entirely worthless stat.
Click to expand...
What a surprise.
 
S

SteveWinwood

Member
1.09 said:
I would've rather not had 4 giant gifs in my dota thread
Click to expand...

i like em

balladofwindfishes said:
Dotaskill dude left his database and source code wide open for a few minutes.

It just takes your win rate, KDA, GPM, XPM and enemy's average win rate and figures out your "skill" via that.

Which means low kill, low farm games (for example, if you're a support) probably count against you even if you did everything right. It doesn't appear to take wards or anything into account, so it's very carry biased.

An entirely worthless stat.
Click to expand...

shocking
 
G

G.ZZZ

Member
balladofwindfishes said:
Dotaskill dude left his database and source code wide open for a few minutes.

It just takes your win rate, KDA, GPM, XPM and enemy's average win rate and figures out your "skill" via that.

Which means low kill, low farm games (for example, if you're a support) probably count against you even if you did everything right. It doesn't appear to take wards or anything into account, so it's very carry biased.

An entirely worthless stat.
Click to expand...

OH SURPRISE, THIS IS WHY TRASHVIMES HAS AN HIGHER DSR THAN ME, YOU FUCKING TRYHARD MIDPICKER PLZ POOL ME 4 TANGOS BECAUSE WE NEED ME MID TO WIN.

In the meanwhile, G.ZZZ create space everywhere and win games ez.
 
Wok

Wok

Member
Wok said:
Is there a Steam API to access stats about games played? The kind of things used by dotabuff and mydotaskill?

Edit: http://dev.dota2.com/showthread.php?t=58317

Edit: If I am not wrong, there have been more than 8 million matches for the last seven days. This is a lot to parse.
Click to expand...

The biggest obstacle to make one's own ranking consists in downloading data with such restrictions:
You must manually limit your requests to one request per second in order to reduce the strain on the servers

Please note that as written in the WebAPI T&C's, you are limited to 100,000 API calls per day.
Click to expand...
 
L

LiQuid!

I proudly and openly admit to wishing death upon the mothers of people I don't like
Liquid, come the fuck on bros....................
 
Status
Not open for further replies.

Similar threads

The Verge: We played Valve’s secret new shooter: Deadlock
Opinion News 2 3 4
199
14K
near replied
Eiyuden Chronicle: Hundred Heroes | Review Thread
Analysis Review 2 3
143
12K
BadBurger replied
Romancing SaGa 2: Revenge of the Seven | Review Thread
Review
22
2K
NotMyProblemAnymoreCunt replied
S.T.A.L.K.E.R. 2: Heart of Chornobyl | Review Thread
Analysis Review 7 8 9 10 11
506
28K
amigastar replied
  • Locked
Nickelodeon All-Star Brawl 2 | Review Thread
Analysis Review
2
1K
Draugoth replied
Top Bottom