EPT[HoN/DotA] Let's Play~!! - Page 1375
| Forum Index > General Games |
|
JeeJee
Canada5652 Posts
| ||
|
GhostKorean
United States2330 Posts
| ||
|
JeeJee
Canada5652 Posts
Standard "error rate" usually is something like 5% or 1% with 95% confidence. Perhaps you are more or less tolerant to error, so the numbers can change accordingly. I suppose you could ask your teacher on how to approach such a problem. It's quite a fun and very practical subject  | ||
|
Nevuk
United States16280 Posts
It's also really hard to say in hon because of how variant people's skill be on some heroes. Like I have <40% win rates on several heroes like ss, but >70% winrates on glacius, plague, hellbringer. | ||
bkrow
Australia8532 Posts
| ||
|
Etherone
United States1898 Posts
On July 08 2011 12:10 bkrow wrote: Lol Nevuk you just got mocked on Testie's stream :p you've been called out it was me, and all i said was i wanted anyone with a 70+ winrate on hb plague glacius on my team. and testie should invite him to play with us edit: also 2000+ mmr players with 66% winrates that dont play are dubious | ||
|
JeeJee
Canada5652 Posts
On July 08 2011 11:44 Nevuk wrote: Winrate doesn't really normalize to 50% ever, there just aren't enough players after a certain skill level. You'll see 2000+ mmr players with 66% winrates that just don't play on those accounts because their queues take hours. It's also really hard to say in hon because of how variant people's skill be on some heroes. Like I have <40% win rates on several heroes like ss, but >70% winrates on glacius, plague, hellbringer. Well we're talking from an average player's point of view. If you routinely play 1950+ then I don't think you'll get an accurate mmr reading anyway, simply due to lack of players at that level. But someone below 1900 should be totally fine. I also don't think individual hero skill matters much. Yeah you're better in some heroes and worse on others. It averages out to whatever your average skill is. It's like some players in BW that only play one build or one map. They're pretty damn good at it, but overall, not as much. | ||
|
Nevuk
United States16280 Posts
On July 08 2011 12:13 Etherone wrote: it was me, and all i said was i wanted anyone with a 70+ winrate on hb plague glacius on my team. and testie should invite him to play with us edit: also 2000+ mmr players with 66% winrates that dont play are dubious Meh, not really: http://www.heroesofnewerth.com/player_stats.php?aid=3293359 . He has more than one account above 2000, if you play with a team of fairly good people it's not impossible. The reason my win % on some heroes is so high is mostly because I play those heroes with people while I play the other ones more on solo. Someone posted a thing a while back (probably 30-40 pages) where you could see what your win% on an individual hero is, if you saw my overall stats you might change your mind. Some of them were also heroes I haven't played since the 1700 brackets like plague rider. | ||
|
Ganfei2
473 Posts
On July 08 2011 12:13 Etherone wrote: it was me, and all i said was i wanted anyone with a 70+ winrate on hb plague glacius on my team. and testie should invite him to play with us edit: also 2000+ mmr players with 66% winrates that dont play are dubious Sounds to me like you're one of those people who likes to say the opposite of things that are easily verifiable without any prior knowledge or research whatsoever, but what the fuck, the world is full of people like that so I guess I was just trying to say welcome. Winrate will never go to 50% at the highest ranks when the player pool above 1900 is 500 people. Also, the word dubious is never used to describe people as a standalone adjective. | ||
|
wonderwall
New Zealand695 Posts
There are better pushers like defiler or polly and better tanky initiators like Kraken. I just don't see what his role is meant to be? | ||
|
Dead9
United States4725 Posts
u can tell because he is very red and angry + Show Spoiler + he's kinda like defiler where you push really fast and get super tanky, except worse he's pretty good in lane tho (him + dsham is one of the strongest dual lanes, but he's pretty good with any stunner/slower) his ulti is pretty nice in teamfights, though usually it doesn't hit 80+ until halfway through the teamfight =\ also synergy doesn't matter if the silence is super good, it's a 4s silence with a dot attached that spreads to more targets what more can you ask for | ||
|
Nevuk
United States16280 Posts
On July 08 2011 18:15 wonderwall wrote: I got Balph in an AR game. I don't really get how he's meant to be played. He has good early pushing potential with the whole minions+vomit synergising really well but, after the first 10-15 minutes of the game, what do? The silence and the ult seem counter-intuitive, you get more damage by them casting spells, so you stop them from casting spells? He isn't a great initiator with the ult doing mediocre damage even when fully charged and he's kinda squishy with a low str gain. There are better pushers like defiler or polly and better tanky initiators like Kraken. I just don't see what his role is meant to be? If you haven't won or gotten an advantage by 15-20 minutes you probably won't win. He was nerfed really hard a few patches ago and is really subpar. You don't have to rax them by the 20 minute mark or anything, but you need at least 3 towers by this point to win generally and even with that it'll be an uphill fight if you're behind in kills. As for how to play him : His strength is in teamfights moreso than pushing. The aoe on the ult is so large that it's really hard to miss with (it also moves with you). After level 6 your ult should be up for every teamfight, just carry tps and tp whenever you think there is going to be a fight somewhere and ult and slow. His ult is extremely strong if you build him very tanky and can reliably get off an 80+ ult every fight, the minions are pretty much a non factor besides for pushing after about level 12, the slow is decent and the silence can be extremely useful to disable a key member of their team, and you can get around nullstones by casting it on a nearby player and spreading it. The point of it kind of is that if their team is so silenced that they aren't casting spells then there's no way they're going to win that fight (assuming they don't have 5x your farm). The standard item build to go on him really is mana battery, power supply, phase boots (plated greaves if you're on a push lineup or just like them more), helm of the black legion, shaman's. (The progression is 2 totems, shield, tango, hp pot, a mana pot (those are just my starting items), lifetube, red boots (either order on this is fine), mana battery, hotbl, vestments, phase boots, shamans). At this point most people would go frostfield since he has mana issues and it gives him a massive survivability boost (+15 armor is massive, 90% increase to effective hp vs physical attacks), more aoe slow and benefits teamfights through the aura, and also can prevent people from running out of ult range. Also acceptable is getting an idol before frostfield, but that's really dependent on how well your team is staying alive and what damage the opposing team deals. Vs magic heavy team get idol, vs physical or mixed generally get frostfield. | ||
|
Etherone
United States1898 Posts
On July 08 2011 14:26 Nevuk wrote: Meh, not really: http://www.heroesofnewerth.com/player_stats.php?aid=3293359 . He has more than one account above 2000, if you play with a team of fairly good people it's not impossible. The reason my win % on some heroes is so high is mostly because I play those heroes with people while I play the other ones more on solo. Someone posted a thing a while back (probably 30-40 pages) where you could see what your win% on an individual hero is, if you saw my overall stats you might change your mind. Some of them were also heroes I haven't played since the 1700 brackets like plague rider. Did you mean the average of a 50% winrate across all players in every skill bracket? if so then it is 50% considering for every increase there must be a decrease. very few players who smurf and get team stomps organized to get very few accounts to 2000 mmr at a 66% winrate won't affect the average. if you mean players at 1900 don't have 50% winrates, then ill just have speedo screen cap tomorrow the next games he complains that there are mostly people with 50-54% winrates on both sides, with a few exceptions, and most of those exceptions are team players. well obviously heroes that you have a 70% winrate with over an insignificant sample shouldnt enter the conversation when you are trying to establish skill discrepancies with different heroes, especially when you are talking about heroes you haven't played for 100+mmr ( assuming you are 1800+) | ||
|
Nevuk
United States16280 Posts
On July 08 2011 20:20 Etherone wrote: Did you mean the average of a 50% winrate across all players in every skill bracket? if so then it is 50% considering for every increase there must be a decrease. very few players who smurf and get team stomps organized to get very few accounts to 2000 mmr at a 66% winrate won't affect the average. if you mean players at 1900 don't have 50% winrates, then ill just have speedo screen cap tomorrow the next games he complains that there are mostly people with 50-54% winrates on both sides, with a few exceptions, and most of those exceptions are team players. well obviously heroes that you have a 70% winrate with over an insignificant sample shouldnt enter the conversation when you are trying to establish skill discrepancies with different heroes, especially when you are talking about heroes you haven't played for 100+mmr ( assuming you are 1800+) Glacius is my most played hero at 10%, but aside from him it's 3-4% max. So I don't think any of my heroes are technically "statistically significant." Regardless, the point was not about the variance in my individual skill at heroes but rather that you will have 1800 players who occasionally play like 1700 players or 1900 players depending on their hero and that mmr isn't particularly accurate except as an average. (And on some players it can be even more extreme of a difference). My own was just a somewhat anecdotal example, feel free to disprove it with a non-anecdotal one if you can find the evidence. (It may be hard with hon's xml down for maintenance though). Basically the way I think of it is that someone's mmr is very accurate if their win% is 50, but the further from 50 the less accurate it is - a 1600 player with a 55% winrate could probably make it to 1650-1700 if they played an infinite number of games without increasing their skill level and at that point would have a 50% winrate. However, no one plays an infinite number of games so you just have to guess. (I did mean that at higher brackets the win% tends to be above >50% because there are so few players at those ratings) | ||
|
Dead9
United States4725 Posts
But yeah you can grind out some pretty high mmr with some pretty low winrates lol + Show Spoiler + | ||
|
Ocedic
United States1808 Posts
Come watch! | ||
|
Nevuk
United States16280 Posts
On July 08 2011 20:45 Dead9 wrote: It's never gonna average to 50% because you have to win games to get up to the mmr where you have a 50% winrate But yeah you can grind out some pretty high mmr with some pretty low winrates lol + Show Spoiler + damn son. Anyways, matchmaking balance feels a little wonky the past few weeks. Anyone else notice massive weird mmr matchups? Like.... 3 1600s with 2 1800s, 1570s with 1850s, etc. (I mean when they obviously weren't locked together) | ||
|
anomalopidae
Slovenia549 Posts
| ||
|
Dead9
United States4725 Posts
Not sure if it's just me, but I'm watching a 1080p slideshow lol edit: new patch + Show Spoiler [2.0.38] + - Fixed hitting S or H while connecting to a server causing the game to never load - Updated text when disconnecting from Matchmaking to be more clear - Fixed players respawning attached to projectiles when Gore Taunted when "no respawn timers" is set - Fixed custom account icons counting as modified files for the purposes of Tournament Rules - Fixed exploit with Rampage and Blood Hunter reveal gadgets - Fixes to Emerald Warden bugs with Nullstone and Magic Immunity - Nullstone effects tweaked - Added new Alt Avatar: Phantom Defiler - Added new Alt Courier: Rat Courier == Items == - Alchemist's Bones cooldown/timer increased by 20 seconds - Astrolabe * Reworked Components * Fortified Bracer (510) * Refreshing Ornament (603) * Astrolabe Recipe (600) * Total Cost: 1713 * Effects * +8 Strength, +5 Agility, +5 Intelligence * +3 Health Regen aura * On use: Heals nearby allies for 200 Health. 100 Mana cost, 45 second cd * A hero/unit can still only be healed once every 25 seconds - Energizer * Components * Soulscream Ring (460) * Major Totem (540) * Energizer Recipe (700) * Total Cost: 1700 * Effects * +7 Strength, +10 Agility, +7 Intelligence * On use: Nearby allies gain 50 Movement Speed and Unitwalking for 5 seconds. 100 Mana cost, 45 second cd - Ring of Sorcery * Reworked Components * Amulet of Exile (485) * Pickled Brain (900) * Auto combines * Total Cost: 1385 * Effects * +300 Mana, +3 Strength, +3 Agility, +6 Intelligence * On use: Nearby allies gain 135 Mana. 25 Mana cost, 33 second cd - Shieldbreaker * Rework * Warhammer (1600) * Shieldbreaker Recipe (900) * Total Cost: 2500 / 3400 / 4300 * Effects * +30 / 45 / 60 Damage * -2 / -4 / -6 Armor debuff on target + Show Spoiler + i like the alch nerf, shieldbreaker rework, and semiport of jango i guess im okay with the manaring rework i hate the astro rework though, i guess it makes it more support oriented but i don't think it's going to be picked up much any more =\ | ||
|
Ocedic
United States1808 Posts
On July 09 2011 00:32 Dead9 wrote: Not sure if it's just me, but I'm watching a 1080p slideshow lol Hmm, yeah I noticed that. Gonna try to fix and resume streaming later. | ||
| ||
