Gentle Rodriguez
Member
SYNDERBAIT
-
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.
DOTA2 |OT14| i give up like your pubs do
- Thread starter Artanisix
- Start date
JawesomeJaw
Member
Let's go for 11 games in a row.
lv2 pidgey
Member
NormalFish
Banned
Inkls just working with the FBI agent to identify problematic individuals.
As DotoGAF's lawyer, I recommend not posting your report cards.
As DotoGAF's lawyer, I recommend not posting your report cards.
Gentle Rodriguez
Member
Snitch detected
SecretMoblin
Member
there's no way to see the commendation breakdown anymore (leadership/friendly/etc) right
I've been handing out lots of commends lately, especially to good players who have had to tolerate a miranda or drow on their team.
I've been handing out lots of commends lately, especially to good players who have had to tolerate a miranda or drow on their team.
lol
I'm just trying to look into what the "report rate is higher than % of players" is between 4 reports and higher.
Player score has me thinking as well. Does player score affect matchmaking? I've been thinking about this for a while because I hear lots of examples of people constantly paired up with flamers/trolls/ragequitters/etc but going in soloq both in unranked and ranked I've met nowhere near that same amount. So I'm thinking theres possibly some kind of "soft" low prio that you can get sent.
Of course there's also the possibility that mmr, player sensitivities, region, etc affect this and that what I'm saying is nonsense, but this is just one possibility I'm interested in exploring.
i would think so, it probably considers balance vs. queue time
Gully State
Member
Why can't they apply this to language as well?
waxinlyrical
Member
How broken is Jungle Beastmaster? <_< I don't really see it in my scrub tier matches but I don't buy Nahaz's "OMG IMBA BROKEN" analysis at face value.
When he says "play styles", he named those examples, but is he also referring to stuff like what kind of heroes people play, how they play them, etc?
If you want 10 minute queues they could. Otherwise chat wheel and alt+click exist.
waxinlyrical
Member
Not sure if serious. <_<
He's clearly saying they're trying to put abandoners/cheaters etc in the same matchmaking pool.
I understand that, but he said "play styles" and not characteristics or tendencies. I doubt it actually does that (the hero thing), but I was curious of it did
no its true, all op hero spammers are put in a special pool
User 85168
Banned
Pros get lvl 6 by like 5:45, and it's hard to shut down. That's almost a guaranteed gank and can snowball after that. Only cost is weaker lanes.
SecretMoblin
Member
What's the jungle BM strat? Stack and axes? The boar certainly can't tank a lot. When I play him I try and lane but send the boar to stack camps that I later take out, but I don't start in the jungle.
Is Enigma still the fastest jungler assuming optimal play?
User 85168
Banned
Iron Talon with boars is insanely fast
The same one as everyone else?
Iron talon boar lvl 1, passive lvl 2 then max boar/hawk and axes. Constantly scout with hawk while jungling. brown boots and vlad rush. You can almost always get a kill at 6 mid with a smoke.After that most ppl like getting necro but tbh I like aether lens and bots. Aghs and blink r also super good on him
I'm sure someone like swift could give better advice.
SecretMoblin
Member
interesting, thanks
I like blink/aghs but I know how powerful necro is on him, usually because I'm on the wrong end of it
I wonder if additional burst or lockdown is good, make him a clowny megaganker with lvl 5 dagon or ult -> hex while necro units feast
I like blink/aghs but I know how powerful necro is on him, usually because I'm on the wrong end of it
I wonder if additional burst or lockdown is good, make him a clowny megaganker with lvl 5 dagon or ult -> hex while necro units feast
Sticky Fingers
Member
Empire mid LC
rip milkman
I think they do, but if it takes too long to matchmake you its probably one of the criteria that get dropped relatively early.
Monochromaticx
Member
Empire mid LC
the proper way to play lc
6.87c better nerf beast master PROPERLY
reduce roar range ,reduce boar slow, remove invis from hawks, reduce passive attack speed or something.
SecretMoblin
Member
boar slow not through immunity should be enough?
NormalFish
Banned
i think there are just weird 'pockets' of MMR where lots players end up, and if there are some really obnoxious people, it permeates. I struggled to climb in the 2800 range, but I really enjoyed it because I played with some cool people who consistently played USE at the same timeslot.
However, once I hit 3k, I fell in to a similar sort of group of people, except most of them were super rage-y dickbags.
Though, that said, me and TUSR regularly managed to run in to absolutely terrible team mates and opponents whenever we duoed, so maybe I'm cursed.
Nah, decrease its overall slow speed in addition to making it not go through immunity.
I'm an absolute nightmare in pubs, and these people we end up queuing against put me to shame.
User 85168
Banned
He will never get picked
Here's a question for the mathematicians among us: How would you compensate for the inflated winrate on certain items on Dotabuff? So for example, Octarine Core has a really high win rate on a lot of heroes, but we don't consider the item OP because we have to account for the fact that if you get an Octarine on certain heroes, it's because you're already far ahead and it's a luxury item. So how would to compensate for that and find "true" item winrates?
if you could calculate the gold/objective/experience differential between the two teams at the time that the item was purchased, you could account for it somewhat. As in, Winrate while ahead, winrate while behind, winrate at parity.
I don't think item winrates have really been considered much of a factor by themselves. Now if it has a high winrate on like 90% of the hero pool, that is something to be concerned with.
It's like saying that Shiva's guard is OP because it has a 60% or higher winrate on every hero except Techies and Lycan (and on those two it's 58%). Bigger items like Octarine are bought if you're farming well. They're more expensive, and usually the team that has those types of things is snowballing.
Wok
Member
Something along the lines of conditional probabilities of winning with item X given that your gold advantage is Y at time T. Then compare conditional probabilities given same given gold advantage Y at same time T.
Gold advantage should be computed at the time of the item purchase of course.
One could already try to estimate p(item=X, goldAdvantage=Y, purchaseTime=t) based on yasp.co database, since there are gold graphs and times of item purchases. Similarly p(goldAdvantage=Y, purchaseTime=t) could be estimated. Then apply Bayes'rule.
That's what I'm saying: how would you alter the math so that we can see the how successful Shiva's is on given heroes accounting for that snowball effect?
NormalFish
Banned
I'm not sure this would be much more useful, as the amount of data would end up skewing the results (if Octarine is bought 80% of the time on X hero when ahead, then we have quite a bit less data for buying it without a significant gold differential or deficit). In theory I agree with you, but I'm not sure it'd work out, practically.
This is probably the best way to go about telling if an item is OP. For example, if Armlet is OP on one patch, you'll start seeing supports building Armlet and winning, which indicates the item is a bit over tuned. Inversely, if an item is built on only a handful of heroes, and isn't very succesful on those ones, it's not really worth building.
That said, it's still not perfect, as luxury items will naturally be built when a team is ahead, and early/midgame items will be more situational and difficult to assess.
I'm not sure there's a good solution for this. Probably best off watching a shit ton of dota and paying attention to which items meet success. Take a sample of 50-100 pro matches in which a given item is built, and see how it plays out.
Again, I think this makes sense in theory, but dota is a bit too complex to only include those factors.
SecretMoblin
Member
Similar threads
- Locked
- Locked