|
Rerolling chests doesn't seem like a good strategy during this event. Most of the holiday skins are epics, so the odds of getting them are slim. And while you will get the rares eventually (Malthael and Kharazim) you get a dozen animated sprays first. Better just to reroll for shard value I guess.
|
#
?
Dec 16, 2017 10:47
|
|
|
|
| # ? May 18, 2024 19:49 |
|
|
Why on Earth wouldn't you train the system with a month of data before launching instead letting it take weeks to figure out pros are good.
|
|
#
?
Dec 16, 2017 11:43
|
|
|
I am not losing as much playing with Hanjo as I do with one on my team in overwatch
|
|
#
?
Dec 16, 2017 12:07
|
|
|
Man this was a good week to have 0 time to play HotS  Having withdrawal, though. Just woke up from a dream where the camera was zoomed out way too far. Still managed to flip Butcher over our wall as Diablo. Woke up honestly surprised that it was a dream.
|
|
#
?
Dec 16, 2017 13:58
|
|
|
porfiria posted:I really don't get the attachment to Performance Based MMR. It's solely to shut people up who continually cry about "MMR hell", because now they can't put the blame on their team when the system gives them negative performance adjustments.
|
|
#
?
Dec 16, 2017 16:42
|
|
|
DisgracelandUSA posted:Why on Earth wouldn't you train the system with a month of data before launching instead letting it take weeks to figure out pros are good. They've been training it for the better part of a year with pro data. The AI is just bad.
|
|
#
?
Dec 16, 2017 17:54
|
|
|
wiegieman posted:PBMMR was doing things like giving out negative points to people who won games quickly. If one team stomps the other, they should get more MMR, so that they get up to the MMR band they belong at and not get matched up with people who can't compete with them and get stomped. If the system isn’t rating performance on a scale that includes game time in any way then I’m honestly baffled as to how it was designed. It seems like there’s some really obvious things that would take anyone looking at game data five minutes to chart out in terms of what variables you would score. And your scoring would absolutely have to include performance / time or else what the gently caress are you doing? The ultimate game score becomes those assholes who stalled a hanamura forever to farm 125 kills.
|
|
#
?
Dec 16, 2017 18:42
|
|
|
I just wish blizz showed you what PBMM evaluated on. Ideally, that would help you improve your game by knowing X metric was better or worse than 50% of other players. Of course, then people will try to game those stats, but now people are trying to game stats anyways and it's all a giant tire fire.
|
|
#
?
Dec 16, 2017 18:47
|
|
|
As I’ve understood it the system uses time slices to evaluate performance so it absolutely does include a time dimension.
|
|
#
?
Dec 16, 2017 18:47
|
|
|
Midnightghoul posted:Some of the stuff on reddit from when HL was up was pretty wild. Multiple pro players getting major penalties to ranked points for playing correctly (there was a good writeup about how the system intends for tanks to take as much damage and use cooldowns as much as possible, and penalizing efficient tank play), some high rank GM players getting moved back to Diamond/Plat/Gold, and vice versa This is exactly what I meant when I predicted players would find out how to game the system. Turns out the best way of ranking someone is whether they win or lose and attempting to use numbers to analyze and reward (or punish) performance is a waste of dev time!
|
|
#
?
Dec 16, 2017 18:57
|
|
|
Libertine posted:If the system isn’t rating performance on a scale that includes game time in any way then I’m honestly baffled as to how it was designed. It seems like there’s some really obvious things that would take anyone looking at game data five minutes to chart out in terms of what variables you would score. And your scoring would absolutely have to include performance / time or else what the gently caress are you doing? The ultimate game score becomes those assholes who stalled a hanamura forever to farm 125 kills. My wild guess: In really fast one-sided stomps the winning team usually doesn't do much structure damage (because they just blast through one lane) and racks up a lot of kills while not laning and mercing as much as they would in a normal game. To the algorithm it looks like they're scrubs that need to l2p rather than players who so outclass the other team that they don't need to do anything but fight.
|
|
#
?
Dec 16, 2017 19:26
|
|
|
Erdricks posted:I just wish blizz showed you what PBMM evaluated on. Ideally, that would help you improve your game by knowing X metric was better or worse than 50% of other players. Of course, then people will try to game those stats, but now people are trying to game stats anyways and it's all a giant tire fire. It evaluates basically everything, but the weighting changes periodically based on what wins games. They can't say "get lots of siege damage with X character" because in a few weeks that may not be what wins games as that character, it might be hero damage or taking camps. The idea of the system is you try to win games. Whether it's going to pan out or not remains to be seen.
|
|
#
?
Dec 16, 2017 20:12
|
|
|
lucifirius posted:There's a reason Overwatch doesn't use it (as much?). What? No, Overwatch absolutely uses performance based MMR, it's just that they black box it "so people can't game the system" but all this does is A). cause SR gains to be wonky at times whenever they adjust the balance of various features but fail to appropriately adjust their performance-based metrics and B). cause every player to have a million unfounded theories as to what the performance-based MMR grades on, which leads to people blaming their team/one-tricks/the phases of the moon for their losses instead of going "oh thank you game for giving me useful feedback to improve upon." porfiria posted:I really don't get the attachment to Performance Based MMR. Performance-based MMR is one of those things that sounds great on paper...why wouldn't you want to give players more SR for playing the game "well?" It encourages people to use good strategies and do useful things, right? But the problem, as it always is, remains that any performance-based MMR system is only as good as the people who program it. In other words, what performance-based MMR means is that you're relying on a bunch of Blizzard programmers to understand how to "properly" play the game and program the system accordingly and this never, ever, ever, ever, ever works in practice because there's a pretty significant gap between what a bunch of programmers like to think is the "right" way to play and how things actually work, especially as players get more skilled. Like the example that got brought up earlier re: tanks, to some guy at Blizzard it probably makes perfect sense that tanks should be graded on soaking damage and using cooldowns as often as possible, but it's not hard to imagine that actually it's probably a sign of better, more efficient play for tank players to not soak a bunch of damage they don't have to and to save their important abilities for when they're most impactful, but if Blizzard decides that their vision of the game is the correct one then guess what that means?
|
|
#
?
Dec 16, 2017 20:16
|
|
|
Kai Tave posted:What? No, Overwatch absolutely uses performance based MMR, it's just that they black box it "so people can't game the system" but all this does is A). cause SR gains to be wonky at times whenever they adjust the balance of various features but fail to appropriately adjust their performance-based metrics and B). cause every player to have a million unfounded theories as to what the performance-based MMR grades on, which leads to people blaming their team/one-tricks/the phases of the moon for their losses instead of going "oh thank you game for giving me useful feedback to improve upon."
|
|
#
?
Dec 16, 2017 20:27
|
|
|
Kai Tave posted:Performance-based MMR is one of those things that sounds great on paper...why wouldn't you want to give players more SR for playing the game "well?" It encourages people to use good strategies and do useful things, right? But the problem, as it always is, remains that any performance-based MMR system is only as good as the people who program it. In other words, what performance-based MMR means is that you're relying on a bunch of Blizzard programmers to understand how to "properly" play the game and program the system accordingly and this never, ever, ever, ever, ever works in practice because there's a pretty significant gap between what a bunch of programmers like to think is the "right" way to play and how things actually work, especially as players get more skilled. It's also possibly the case that metrics do an OK job separating mediocre players from horrible ones--someone who walks into towers and is dead 40% of the game is fairly easy to see statistically. But separating mediocre from good or really good seems way harder.
|
|
#
?
Dec 16, 2017 21:09
|
|
|
biznatchio posted:QM doesn't guarantee you a healer or a tank. It only matches like to like. (And recently it doesn't seem to be doing even that well.) So what do you play if you don’t have any friends who like this game but are interested in a balanced team? I only play QM because I’m an old fart and remember when I didn’t own enough heroes to play anything else. They eventually relented and let people pick like 20% of the roster to unlock, but I still keep playing QM mostly because I just focus on the daily gold quests until I unlock all my personal favorite characters. Which will take approximately forever, so I might as well ask about the alternatives now.
|
|
#
?
Dec 16, 2017 21:17
|
|
|
Kai Tave posted:Like the example that got brought up earlier re: tanks, to some guy at Blizzard it probably makes perfect sense that tanks should be graded on soaking damage and using cooldowns as often as possible, but it's not hard to imagine that actually it's probably a sign of better, more efficient play for tank players to not soak a bunch of damage they don't have to and to save their important abilities for when they're most impactful, but if Blizzard decides that their vision of the game is the correct one then guess what that means? Well, the best players know what they have to do and whether they care about being rank 1 world or just being ranked super high is on them. What matters most to them is going to be tournament play, and they’ll do what is most likely to win rather than what gives the highest MMR gains. The person at the very top had to be good to get there, even if they cheesed the factors the system grades on to get that spot. It’s sort of like how the #1 Ranked Hearthstone player day to day isn’t always the most valuable tournament player, but they’re clearly better than all but 500 or so people.
|
|
#
?
Dec 16, 2017 21:31
|
|
|
Craptacular! posted:So what do you play if you don’t have any friends who like this game but are interested in a balanced team? Play unranked mode. Which has the added benefit that no matter which character you play, every game counts toward completion for your daily quests.
|
|
#
?
Dec 16, 2017 21:50
|
|
|
That's technically not a good thing if you're trying to absolutely maximize your gold gain tho.
|
|
#
?
Dec 16, 2017 22:04
|
|
|
But hero league is the easiest way to do quests?
|
|
#
?
Dec 16, 2017 22:08
|
|
|
John Murdoch posted:That's technically not a good thing if you're trying to absolutely maximize your gold gain tho. Pretty great if you just log in every few days to clear the quest backlog as quickly as possible though.
|
|
#
?
Dec 16, 2017 22:16
|
|
|
Good Canadian Boy posted:But hero league is the easiest way to do quests? How is hero league any different from Unranked Draft here?
|
|
#
?
Dec 16, 2017 22:19
|
|
|
Craptacular! posted:Well, the best players know what they have to do and whether they care about being rank 1 world or just being ranked super high is on them. What matters most to them is going to be tournament play, and they’ll do what is most likely to win rather than what gives the highest MMR gains. There's no reason for "things that give you good MMR gains" and "things that actually make you a better player" to be operating on separate axes, is the point.
|
|
#
?
Dec 16, 2017 22:20
|
|
|
Kai Tave posted:Performance-based MMR is one of those things that sounds great on paper...why wouldn't you want to give players more SR for playing the game "well?" It encourages people to use good strategies and do useful things, right? But the problem, as it always is, remains that any performance-based MMR system is only as good as the people who program it. In other words, what performance-based MMR means is that you're relying on a bunch of Blizzard programmers to understand how to "properly" play the game and program the system accordingly and this never, ever, ever, ever, ever works in practice because there's a pretty significant gap between what a bunch of programmers like to think is the "right" way to play and how things actually work, especially as players get more skilled. None of that is how machine learning works though. The entire idea is that Blizzard doesn't determine what makes players good. An algorithm looks at a lot of data and finds patterns that good players follow. And that will get better over time as it looks at more and more data. You can argue over how well that works but none of what you said applies to machine learning.
|
|
#
?
Dec 16, 2017 22:37
|
|
|
DisgracelandUSA posted:Why on Earth wouldn't you train the system with a month of data before launching instead letting it take weeks to figure out pros are good. My understanding is that they did.
|
|
#
?
Dec 17, 2017 00:17
|
|
|
QM matchmaking is... good. 
TwoWordName fucked around with this message at 01:32 on Dec 17, 2017 |
|
#
?
Dec 17, 2017 01:23
|
|
|
Love the winking lootbox overlooking that
|
|
#
?
Dec 17, 2017 01:35
|
|
|
I was going to say, I relate to that Stukov, but then I realized how terrible his healing was, what the hell. 60k would be below average for a game that long. Laffo @ Zarya's shoields being more valuable over all.
|
|
#
?
Dec 17, 2017 02:35
|
|
|
Even less healing than Tyrande. But then I noticed which award he got and it all makes sense now.
|
|
#
?
Dec 17, 2017 03:44
|
|
|
Do I need to have some software running to capture replays? Or is there something existing I just need to extract?
|
|
#
?
Dec 17, 2017 04:12
|
|
|
Check if you have them here C:\Users\(your login)\Documents\Heroes of the Storm\Accounts\(some number)\1-Hero-1-6683\Replays\Multiplayer
|
|
#
?
Dec 17, 2017 04:43
|
|
|
You can also just open the "watch" tab in game and click "view folder in explorer" (some wording like that) They save automatically and can play on any HotS client, though if they're from a previous patch it may take a while for them to load. If you want them in standalone video format for youtube or w/e, you'll need to play them while using a screen capture like OBS.
|
|
#
?
Dec 17, 2017 05:29
|
|
|
Newbie update. Been playing for about two weeks now. I got Gul'dan and Brightwing to 5 today. Gul'dan seems pretty fun. He can clear waves, and has good sustain. His ult is good for teamfights too. In the uncoordinated world of quickmatch, he seems pretty solid. My win rate with him was quite good. Brightwing is...well...I apparently can't figure out how to brightwing effectively. I might even be the worst brightwing player of all time, as I lost every single game. I don't seem to do particularly good healing, or particularly good damage. Enlighten me good sirs, how do I not be bad with the hungry faerie dragon?  I tried Sgt. Hammer after brightwing, but was getting sick of my losing streak, so I went back to butcher to get back to winning. Doing a butcher build focusing on block and charge abilities seems effective. Thinking Butcher is definitely going end up being a gold purchase for me after he goes off the free rotation. Have saved up about 12000 gold so far from this week. Filthy Monkey fucked around with this message at 06:10 on Dec 17, 2017 |
|
#
?
Dec 17, 2017 05:53
|
|
|
Filthy Monkey posted:Brightwing is...well...I apparently can't figure out how to brightwing effectively. I lost every game in which I played, so it is entirely possible that I am terrible with brightwing. I don't seem to do particularly good healing, or particularly good damage. Enlighten me good sirs, how do I not be bad? Blink Heal is fine, probably your go-to with most random QM comps, but I think Emerald Winds is the better ult. Like Gust, it doesn't do anything besides moving people around because, like Gust, it would be ridiculously overpowered if it did. With proper positioning, your ult is stealing a boss, turning a 5v5 into a 2v5 followed by a 3v5, or completely shutting down the enemy team as they try to engage. Plus she brings Bribe and Cleanse, which can be amazing in certain situations or against certain comps. Overall, Brightwing isn't going to top damage, or healing, or anything else. She is mostly a non-factor as a support against burst damage team compositions. But she brings more utility than any other character in the game I can think of. In the right situation, she'll counter an entire enemy team by herself.
|
|
#
?
Dec 17, 2017 06:19
|
|
|
Appreciate the advice. I agree with the brightwing waveclear being decent. I did often find myself on soaking duty, since I find that is often something people neglect to do. I tried both ultimates with Brightwing, and neither one really impressed me. I can see your reasoning behind emerald wind though. I definitely was picking up the increased polymorph duration talent. That one seemed like a no brainer. With butcher I've found more success generally not soaking myself, but instead trying to gank people soaking lanes. If they extend out a little too far, particularly if they aren't at full health, running at them and trying to hack their precious face meat off will sometimes net a kill. I am still pretty bad at the game, but he has been my most successful assassin character so far. In a couple of the butcher games today I had a Medivh buddies who seemed to like to stick with me. Definitely helpful having portals and shields for safer face carving. Also had some amusing tassidar walls, trapping enemies next to me for delicious dismemberment. What builds to people like for butchering? I've mostly been using something like this so far. I have been torn between enraged and bloody frezny for the 16 talent. https://www.heroesfire.com/hots/talent-calculator/the-butcher#kgOV Kind of trying to focus on some survivability, along with the charge skill talents. Filthy Monkey fucked around with this message at 07:33 on Dec 17, 2017 |
|
#
?
Dec 17, 2017 07:30
|
|
|
Filthy Monkey posted:What builds to people like for butchering? I've mostly been using something like this so far. I have been torn between enraged and bloody frezny for the 16 talent. I generally go for the Hamstring talents at 1 and 4, being able to use it quicker, cheaper, and at longer range synergizes well with Brutal Strike at 13 which empowers your next three autoattacks after using Hamstring, so being able to proc that more often can be a nice way to up your overall DPS, and it works no matter when you use Hamstring, so it's good against structures, mercs, etc, but hitting heroes means you can proc it again even sooner. At 7 if I'm really worried about heavy burst comps Meat Shield is good, otherwise I go for Insatiable Blade, then at 16 and 20 unless there's a compelling reason to take Bolt of the Storm I generally just take Blood Frenzy and Nexus Blades.
|
|
#
?
Dec 17, 2017 10:30
|
|
|
Kai Tave posted:There's no reason for "things that give you good MMR gains" and "things that actually make you a better player" to be operating on separate axes, is the point. But there is, because otherwise every change to talents, abilities, stats, etc doesn't change what the hero is supposed to do. This is the most outrageous example to prove the point, but if in theory they accidentally give Hammer the ability to be the best healer in the game in the hands of the most talented player, that doesn't mean she was intended to become a support. If you want a game where heroes actually shift their entire purpose every time a balance patch is out and the company threw the entire concept of roles out of the window, go play Dota2. You will find a game where last week's team-slaughtering high damage powerhouse is this week's preferred low damage disabler/healer, and the entire concept of roles has been abolished. And as much as I loved to watch TI4 Support Jugg, as pro-play so out of left field that other pros disregarded it as a prank at practice, I'd actually personally rather play a game where roles are strictly enforced. Because that kind of "you figure out what it should do" design adds a lot of complexity that, again just to me personally, is cool and awesome to watch and way too frustrating to actually play.
|
|
#
?
Dec 17, 2017 10:35
|
|
|
Tibalt posted:I think Emerald Winds is the better ult. Like Gust, it doesn't do anything besides moving people around It does actually do a lot of damage as well. I once won a 1v1 against Kharazim by surprising him with a close cast Q (it has no travel time if the target area is right next to you) and ult combo, doing like 900 damage in one burst.
|
|
#
?
Dec 17, 2017 11:17
|
|
|
Win or lose I’m having too much fun as Hanjo
|
|
#
?
Dec 17, 2017 12:53
|
|
|
|
| # ? May 18, 2024 19:49 |
|
|
i was so hyped for the MMR system changes i'm a drat fool
|
|
#
?
Dec 17, 2017 13:22
|
|