|
|
On August 21 2013 23:29 Mataza wrote:Show nested quote +On August 21 2013 23:15 Oktyabr wrote:On August 21 2013 21:37 Mataza wrote:On August 21 2013 19:43 Numy wrote:On August 21 2013 12:52 Mataza wrote:On August 21 2013 10:27 TheYango wrote:On August 21 2013 09:16 Blitzkrieg0 wrote:On August 21 2013 09:08 CuddlyCuteKitten wrote:On August 21 2013 09:00 Kuroeeah wrote:
You can't really judge (on most instances) how good or bad a player is from a single game. You can make some general judgements pretty easily. For example in starcraft you could say that if after one game a player had >150 apm, mined x ammount of minerals per minute and never built up a bank of 500 for over 30 seconds he is probably at least decent. I am sure you could find similar metrics in dota to get a rough determination of player skill that works even if he trolls and/or his team lose horribly. Most of the statistics would just be clouded by variables that aren't even under the players control. If a player gets 800 GPM does that make him a good player? What if he was playing anti-mage and his team won the game 4 vs 5 while he just farmed all game? It certainly doesn't make him a bad player, but farming while you're under no pressure isn't impressive. At the same time you could have a player who gets five man gank trained the entire game while his spectre free farms. Does that make the player that got gank trained terrible since he died a lot and his team still wins the game because of spectre? Is the spectre good or bad because the other team ganked the wrong person? Too many variables are out of your control to have any sort of metric if a player is good. Your KDA and GPM don't provide any meaningful data alone. I don't think most of these things would be basic visible statistics. However, there are a lot of things not visible to you in any form that Valve could data mine from replays that in many cases are highly indicative of a good player. The best example that comes to me immediately is how often someone clicks on his opponents to view their items. This is such a bad example. You really only have to check items some few times. You check for boots if you plan an early gank. You check for dust/sentries/gem when you rely on invis. You check for bkb on heroes you want to focus in an upcoming teamfight. You check for dagger/shadowbladeforcestaff on heroes that can stun. For example I would never check a jakiros items if I don't have any invis heroes in my team. You can often tell most of what you need to know from the minimap and from just watching the enemy heroes attack/move. There are ways to tell if a player is good, but I know of no such way that can be expressed as a number. Erm that's not true at all. Checking a jakiros items are often very useful. You see if he has a tp which is important when prepping a dive. Can check if he has obs wards on him(Either unused or 1 stack) then if you keep checking you can get hints at where he has warded so you can either avoid the ward or remove it. I would never intentionally not check a heroes items. You should be checking their whole team constantly. Actually think checking for tps is one of the most important things you can do. Allows you to punish a mistake pretty easily that you wouldn't otherwise punish if you didn't check. Sigh. Then let's get more specific. Situation pub. You won your lane, it's 5 to 10 minutes into the game, your team has no wards. You already know the guys from your lane have next to nothing for the next 5 minutes. The enemy mid is a storm spirit, so you check his level with the scoreboard. You use smoke of deceit and move mid to gank. No item that he can have this early in the game can make it harder to gank him. Just by knowing what someone can have and cannot have at some points in the game reduces the number of times you need to check for items. I could check everyone's items every 3 minutes, but I don't. I check some enemy's items whenever I see them, some others I check once and then never again. I check less because I got better and know what I have to check for. It is similar to terran scouting in sc2. You could spend all your Orbital energy to scout your enemies base. Or you could check only at some selected timings at which actually stuff happens that matters. So you don't check his available mana pool, what rune he might have or how many bottle charges are remaining? Maybe he made treads by then or might have a magic stick with a couple of charges accumulated, all that counts to your estimation of how much your gang train needs to do in order to secure the kill. I've never regretted clicking on every single enemy hero that appears on my minimap whenever they go out to the lane to farm or they walked past river wards. You just get too much information on based on remaining mana pools alone, or ward counts that aren't even numbered that makes clicking on them worth it. I've made several successful ganks on the enemy carry before based off the knowledge that key members of the opposite team don't have tp scrolls. Like what Burningsera said, Yaphets is one sick player that makes way too many clutch decisions based on checking out inventories constantly. The big difference is whether or not storm is level 6. If he is 6, you have to stun/silence him til he is dead or the gank will fail. No items he has makes a difference in that because he will get away if he can cast his ultimate. Before the 10 minute mark no one in the game is gonna have an orchid or sheepstick, so again pointless. And yes, it is possible for a storm spirit to not be lvl 6 between 5-10 minutes. If I remember correctly the bench mark for lvl 6 is roughly 6 minutes, or not so roughly 5-8 depending on denies and number of people on the lane.
Okay, but as the person you quoted just said, there are so many other things you can check. Mana pool, if its low/gone makes it doesn't matter if he is 6 or not and you don't need to go overkill. Checking if he has a rune like invis/haste which can be used through silence or while a stun is in the air to get away to a good position. Or has been stated in other posts, checking his allies for tp scrolls so your gank doesn't go horribly wrong and you lose your whole team for a storm spirit. Checking his supports inventory constantly to know if they warded and sentry warded and might see your smoke.
There are so many more things that are important that you are obviously just throwing out the window despite them all being pointed out to you.
|
Could lorthar be heard through fog? I know manta can.
|
On August 21 2013 22:40 Mataza wrote:
My point is that there is a peak. It is sensible to check inventories x times over the course of a game. Every time more than x that you check inventories is superfluous. Wouldn't you agree that checking inventories of all other players every 10 seconds would be a bit overkill?
Its not overkill if you're capable of doing it whilst doing everything else you need to do...
More information is ALWAYS better than less. If you're able to keep tabs on everyone else's inventory constantly whilst simultaneously last hitting, zoning out the opponents in your lane, monitoring runes, etc then its a great thing to do.
Realistically most people probably can't do it more than once every few minutes. But if you have the chance to do so then why not take it every time? Its free information.
|
On August 22 2013 00:19 CustomKal wrote:Show nested quote +On August 21 2013 23:29 Mataza wrote:On August 21 2013 23:15 Oktyabr wrote:On August 21 2013 21:37 Mataza wrote:On August 21 2013 19:43 Numy wrote:On August 21 2013 12:52 Mataza wrote:On August 21 2013 10:27 TheYango wrote:On August 21 2013 09:16 Blitzkrieg0 wrote:On August 21 2013 09:08 CuddlyCuteKitten wrote:On August 21 2013 09:00 Kuroeeah wrote:
You can't really judge (on most instances) how good or bad a player is from a single game. You can make some general judgements pretty easily. For example in starcraft you could say that if after one game a player had >150 apm, mined x ammount of minerals per minute and never built up a bank of 500 for over 30 seconds he is probably at least decent. I am sure you could find similar metrics in dota to get a rough determination of player skill that works even if he trolls and/or his team lose horribly. Most of the statistics would just be clouded by variables that aren't even under the players control. If a player gets 800 GPM does that make him a good player? What if he was playing anti-mage and his team won the game 4 vs 5 while he just farmed all game? It certainly doesn't make him a bad player, but farming while you're under no pressure isn't impressive. At the same time you could have a player who gets five man gank trained the entire game while his spectre free farms. Does that make the player that got gank trained terrible since he died a lot and his team still wins the game because of spectre? Is the spectre good or bad because the other team ganked the wrong person? Too many variables are out of your control to have any sort of metric if a player is good. Your KDA and GPM don't provide any meaningful data alone. I don't think most of these things would be basic visible statistics. However, there are a lot of things not visible to you in any form that Valve could data mine from replays that in many cases are highly indicative of a good player. The best example that comes to me immediately is how often someone clicks on his opponents to view their items. This is such a bad example. You really only have to check items some few times. You check for boots if you plan an early gank. You check for dust/sentries/gem when you rely on invis. You check for bkb on heroes you want to focus in an upcoming teamfight. You check for dagger/shadowbladeforcestaff on heroes that can stun. For example I would never check a jakiros items if I don't have any invis heroes in my team. You can often tell most of what you need to know from the minimap and from just watching the enemy heroes attack/move. There are ways to tell if a player is good, but I know of no such way that can be expressed as a number. Erm that's not true at all. Checking a jakiros items are often very useful. You see if he has a tp which is important when prepping a dive. Can check if he has obs wards on him(Either unused or 1 stack) then if you keep checking you can get hints at where he has warded so you can either avoid the ward or remove it. I would never intentionally not check a heroes items. You should be checking their whole team constantly. Actually think checking for tps is one of the most important things you can do. Allows you to punish a mistake pretty easily that you wouldn't otherwise punish if you didn't check. Sigh. Then let's get more specific. Situation pub. You won your lane, it's 5 to 10 minutes into the game, your team has no wards. You already know the guys from your lane have next to nothing for the next 5 minutes. The enemy mid is a storm spirit, so you check his level with the scoreboard. You use smoke of deceit and move mid to gank. No item that he can have this early in the game can make it harder to gank him. Just by knowing what someone can have and cannot have at some points in the game reduces the number of times you need to check for items. I could check everyone's items every 3 minutes, but I don't. I check some enemy's items whenever I see them, some others I check once and then never again. I check less because I got better and know what I have to check for. It is similar to terran scouting in sc2. You could spend all your Orbital energy to scout your enemies base. Or you could check only at some selected timings at which actually stuff happens that matters. So you don't check his available mana pool, what rune he might have or how many bottle charges are remaining? Maybe he made treads by then or might have a magic stick with a couple of charges accumulated, all that counts to your estimation of how much your gang train needs to do in order to secure the kill. I've never regretted clicking on every single enemy hero that appears on my minimap whenever they go out to the lane to farm or they walked past river wards. You just get too much information on based on remaining mana pools alone, or ward counts that aren't even numbered that makes clicking on them worth it. I've made several successful ganks on the enemy carry before based off the knowledge that key members of the opposite team don't have tp scrolls. Like what Burningsera said, Yaphets is one sick player that makes way too many clutch decisions based on checking out inventories constantly. The big difference is whether or not storm is level 6. If he is 6, you have to stun/silence him til he is dead or the gank will fail. No items he has makes a difference in that because he will get away if he can cast his ultimate. Before the 10 minute mark no one in the game is gonna have an orchid or sheepstick, so again pointless. And yes, it is possible for a storm spirit to not be lvl 6 between 5-10 minutes. If I remember correctly the bench mark for lvl 6 is roughly 6 minutes, or not so roughly 5-8 depending on denies and number of people on the lane. Okay, but as the person you quoted just said, there are so many other things you can check. Mana pool, if its low/gone makes it doesn't matter if he is 6 or not and you don't need to go overkill. Checking if he has a rune like invis/haste which can be used through silence or while a stun is in the air to get away to a good position. Or has been stated in other posts, checking his allies for tp scrolls so your gank doesn't go horribly wrong and you lose your whole team for a storm spirit. Checking his supports inventory constantly to know if they warded and sentry warded and might see your smoke. There are so many more things that are important that you are obviously just throwing out the window despite them all being pointed out to you.
This is getting tiresome.
The point I was trying to make is that there is a maximum frequency of checking someone's everybody's inventory that makes sense.
Which means that just data mining the number of times someone checks other player's inventories is not a useful metric to gauge a player's skill for the mmr system.
If I check every enemy on the map every minute, I get x mmr.
Now do I get more mmr if I check everyone every 45 seconds?
Do I get more if I check them every 15 seconds, am i then 4 times as much mmr as the 1 minute guy? What about checking every 5 seconds?
Should the mmr system award someone more points for a game if he looked at other heroes more than his own hero?
|
United States47024 Posts
Mataza, you're simply not respecting the level of precision to which many kills happen and to which pros judge them.
It's pretty commonplace among the highest levels of play to keep tabs down to how many magic stick charges someone has (if you've watched YaphetS or 2009 play, they are making note of this all the time). That level of precision is useful. Virtually any item that someone could send to themselves has greater impact than a single magic stick charge.
On August 22 2013 01:12 Mataza wrote:
This is getting tiresome.
The point I was trying to make is that there is a maximum frequency of checking someone's everybody's inventory that makes sense.
Which means that just data mining the number of times someone checks other player's inventories is not a useful metric to gauge a player's skill for the mmr system.
If I check every enemy on the map every minute, I get x mmr.
Now do I get more mmr if I check everyone every 45 seconds?
Do I get more if I check them every 15 seconds, am i then 4 times as much mmr as the 1 minute guy? What about checking every 5 seconds?
Should the mmr system award someone more points for a game if he looked at other heroes more than his own hero?
You're forgetting the original context, which was smurf detection. As a precise MMR function, no it's not useful. As a broad indicator that can be used to assist smurf detection? Yes, this is useful.
|
Lalalaland34507 Posts
Oh that's what spawned this discussion? In which case, yes tracking how often people check opponents' inventories is certainly a very nice way of doing it.
|
Will European teams bother flying out to columbus for a chance at 15k?
|
I'm sure the appeal of Columbus will be enough to bring teams from hundreds of yards away.
|
On August 22 2013 01:30 EmilA wrote:
Will European teams bother flying out to columbus for a chance at 15k?
I don't see anyone but Alliance flying over, if even them.
|
I don't see anyone from EU/CN flying over to the US on their own, but if MLG wants to get some numbers on their stream they better invite some european and/or chinese teams and pay for their tickets at least.
|
|
|
Apparently MLG also offers an invitation to compete at the Fall Championships, so that might at least attract TOP euro teams to fly over.
MLG did mention that three will receive automatic invitation and five teams will be decided from open qualifiers. I don't know how MLG has done things previously, but if they offer some kind of financial help to fly teams over from EU, then i can see more Euro teams going. I hope i understood their press release right.
|
If alliance goes there there will be a fierce fight for second place in the lower bracket finals
|
On August 22 2013 01:43 Qbek wrote:
If alliance goes there there will be a fierce fight for second place in the lower bracket finals
Exactly why they shouldn't
|
Somehow doubt Alliance will want to fly to America after just winning the biggest tournament in Dota history. Imagine they just want to chill for a bit.
|
Netherlands45349 Posts
Arguably flying in the new Virtus Pro or w/e will give MLG the most viewers without making the tournament a stomp.
|
Czech Republic18921 Posts
Let's be honest, any team from outside NA that makes the trip probably takes the first place easily. Not saying all non-NA teams are better than the best NA teams, but all teams that would even consider flying over definitely are, imo.
|
what's everyone's thoughts on magic wand specifically as an item?
|
On August 22 2013 02:05 Kuroeeah wrote:
what's everyone's thoughts on magic wand specifically as an item?
It's good. You should buy it a lot.
|
On August 22 2013 02:06 PassiveAce wrote:Show nested quote +On August 22 2013 02:05 Kuroeeah wrote:
what's everyone's thoughts on magic wand specifically as an item? It's good. You should buy it a lot.
95% of the time you should probably be buying this item.
|
|
|
|
|
|
EPT
