I will have an amazing match, then go on this streak of depressing games with 1-2 players just utterly causing chaos on purpose. Not going to lie tho, even the wins I get don't make me feel better since majority of the wins I get are from sheer lucky of enemy ignoring winions.....

May 8, 2018
15

05/08/2018 11:06 AMPosted by MattViWin/Loss streaks are a natural and unavoidable part of any matchmade multiplayer game, and the focus for us in the context of Heroes is doing our best to make sure those games feel as fair as possible.

Win / Loss streaks are a result of a system that is too aggressive at moving people.

Of course, they WILL happen, but they shouldn't be happening to the extreme that they currently do.

Just think about it for a moment. If the goal of the matchmaker is to give you a 50% chance to win a match, but you win or lose a string of 10 matches in a row, is the matchmaker really doing its job?

Let's do some really basic probability math. What's the probability that someone will lose or win 5 games in a row with a 50% chance to win them?

(0.5)^5 = 0.03125 = 3.125%

10 times in a row?

(0.5)^10 = 0.00098... = 0.098%

Even just losing 5 games in a row is a 3% chance, is it really "natural" that you'd have these streaks? No. Yet, they happen ALL of the time.

Looks to me like this isn't a natural result of a healthy matchmaker if the goal is to give you a 50% chance of winning a match.

Also, I'm glad I got to use discrete math for something, I was starting to think that class was a waste.

May 8, 2018
10

05/08/2018 07:03 AMPosted by LightningActions speak louder than words, and 2.0 was the biggest tell of where their priorities lie. 2.0 was not for the players, it was for the investors. Blizzard is no longer for the players, they are for investors.

Meh, tired of the same discussions, suggestions and comments falling on deaf ears . However, Good to see a BLUE response for a post of this nature, that is indeed rare, but just a whole bunch of TOO LITTLE TOO LATE, gonna on with HoTS. Even WELL before 2.0 posts of this nature were commonly found on these Forums, crying out for something, ANYTHING to be done about Ranked. 2.0 was just the blatant indicator of were the Dev's priorities were, spoiler... it wasn't concerning Ranked play.

Now we're in a place where even if they were to FINALLY do something drastic to help Ranked play, they've lost too many veterans like yourself, that matches will still be a Clown Fiesta simply because there's no longer a large enough pool of skillful and talented players to maintain a quality Ranked experience!

05/08/2018 04:15 PMPosted by Drothvader05/08/2018 11:06 AMPosted by MattViWin/Loss streaks are a natural and unavoidable part of any matchmade multiplayer game, and the focus for us in the context of Heroes is doing our best to make sure those games feel as fair as possible.

Win / Loss streaks are a result of a system that is too aggressive at moving people.

Of course, they WILL happen, but they shouldn't be happening to the extreme that they currently do.

Just think about it for a moment. If the goal of the matchmaker is to give you a 50% chance to win a match, but you win or lose a string of 10 matches in a row, is the matchmaker really doing its job?

Let's do some really basic probability math. What's the probability that someone will lose or win 5 games in a row with a 50% chance to win them?

(0.5)^5 = 0.03125 = 3.125%

10 times in a row?

(0.5)^10 = 0.00098... = 0.098%

Even just losing 5 games in a row is a 3% chance, is it really "natural" that you'd have these streaks? No. Yet, they happen ALL of the time.

Looks to me like this isn't a natural result of a healthy matchmaker if the goal is to give you a 50% chance of winning a match.

Also, I'm glad I got to use discrete math for something, I was starting to think that class was a waste.

Too aggressive? It took me tons of games playing mostly Arthas to move from silver to diamond, and I only did that to proof the point that the system is just !@#$ing bad (was beating GM's at silver).

Now I just play unranked and just laugh when the %^-* that kept me in silver happens: playing Alarak (who I don't like and am not very good with) and get 50% of team fight damage + 75% kills without being buffed, yet the game is clearly lost from the start.

May 8, 2018
14

05/08/2018 03:18 PMPosted by zenasprime05/08/2018 03:14 PMPosted by LightningStreaks will happen naturally. But if you think the streaks in HotS happen at a frequency and duration that is statistically natural, then you need the tin foil hat more than I do.

Define "statistically natural" please.

Also, before you answer, you may want to watch this video on "the Gambler's Fallacy". :)

https://www.youtube.com/watch?v=K8SkCh-n4rw

Not applicable at all.

That is the assumption that given a low chance that you're eventually owed a win condition. For example, if you have a 1% chance to obtain something, and you try for it 100 times, the gambler would say that after those 100 times they're guaranteed to get it. That's now how it works at all.

It actually works like this.

Actual Chance to Win =(1 - (1 - Chance to Win)^Tries)

So with 1% and 100 tries you get the following.

(1 - (1 - .01)^100) = .664 = 66.4%

This is because each event is independent of the last. In matchmaking, each event is NOT independent of the last, your MMR is changing with each result.

Again, I refer you to the math in an earlier post.

In order to determine the probability of a streak, it's simply the multiplicative of the odds of each chain in the event.

Or, for simplicity's sake, if you are supposed to have a 50% chance to win a match, it's simply (0.5)^N.

Considering the chance of flipping a coin (which is 50%) and getting heads 5 times in a row is 3.125%, there's nothing natural about it. The simple fact of the matter is with the frequency of streaks in this game, the matchmaker ISN'T creating matches where you have a 50% chance of winning.

05/08/2018 04:15 PMPosted by Drothvader(0.5)^5 = 0.03125 = 3.125%

10 times in a row?

Yes.

Here, watch this fun video on the Gambler's Fallacy: https://www.youtube.com/watch?v=K8SkCh-n4rw

Also known as the "Law of Averages".

Enjoy. :)

05/08/2018 11:42 AMPosted by RemusBeen here since Beta (lvl 1611). I wait for the Personal Rank system. :-) I hope they implement it this year. Right now i casually play some QM's here and there but even that i am losing interest because i am waiting on some heroes from my top 3 list which is:

Imperius

Mephisto

Baal

I am very close to take a several month long break. Depending on this next hero. I hope they take a break from the Overwatch heroes and give a real deep love to Diablo foremost and Starcraft.

That has been the top 3 for Diablo fans for a long time, (same results every time they do a reddit poll) sadly I think only Imperius has a chance, just cause D3.

05/08/2018 04:27 PMPosted by DrothvaderConsidering the chance of flipping a coin (which is 50%) and getting heads 5 times in a row is 3.125%, there's nothing natural about it. The simple fact of the matter is with the frequency of streaks in this game, the matchmaker ISN'T creating matches where you have a 50% chance of winning.

What evidence do you have that suggests that this may be the case, well other then your observation of some perceived pattern that may or may not exist?

Like, can you show us the MMR values against the results of the matches where players who are match together have significantly lower MMR values and thus are predicted to lose?

I'm just curious what kind of science you are basing these hypotheses on.

May 8, 2018
10

05/08/2018 04:29 PMPosted by zenasprime05/08/2018 04:15 PMPosted by Drothvader(0.5)^5 = 0.03125 = 3.125%

10 times in a row?

Yes.

Here, watch this fun video on the Gambler's Fallacy: https://www.youtube.com/watch?v=K8SkCh-n4rw

Also known as the "Law of Averages".

Enjoy. :)

Again, that's not applicable...

You very clearly have zero idea about what you're talking about.

It's simple combinatorics. If you're supposed to have a 50% chance to have an outcome, IE, a coin flip, you chances of a streak are (0.5)^N.

Considering a streak as low as 5 only occurs at a 3% chance, there's absolutely nothing "natural" about them.

You might make the argument that I can't say that because your MMR is moving every match, well, that's not relevant as the MMR of your teammates and opponents changes as well to match your new MMR value.

If the system is really designed to give you a 50% chance to win a match, streaks are entirely UNNATURAL.

May 8, 2018
-10

05/08/2018 04:27 PMPosted by DrothvaderThis is because each event is independent of the last. In matchmaking, each event is NOT independent of the last, your MMR is changing with each result.

This is not actually true. Your MMR changes very little between games. In fact, that's why PRA exists, to keep your Rank in line with your established MMR range.

SO, even if you do advance in up in Rank by a factor of 4-5 units, it's very unlikely that your MMR has made any significant change.

05/08/2018 04:39 PMPosted by zenasprimeThis is not actually true.

05/08/2018 04:39 PMPosted by zenasprimeYour MMR changes very little between games.

It's not true that MMR changes each match, but MMR changes each match...

You're just digging a deeper hole, dude.

05/08/2018 04:39 PMPosted by zenasprimeIn fact, that's why PRA exists, to keep your Rank in line with your established MMR range.

Again, what does this have to do with streaks? My argument is that STREAKS are UNNATURAL because math.

05/08/2018 04:39 PMPosted by zenasprimeSO, even if you do advance in up in Rank by a factor of 4-5 units, it's very unlikely that your MMR has made any significant change.

Again, this has absolutely nothing to do with streaks...

If you're of Master skill level, and you have a negative PRA, all that affects is the speed at which you climb, not your actual climb. The only situation where this wouldn't be the case is if matchmaking wasn't working correctly.

You're throwing all of this completely irrelevant information out here that isn't even making a cohesive argument. Again, we're talking about STREAKS. If the goal of the match maker is to give you a 50% chance to win your match, and you're getting win / loss streaks of 5 or more often (which happens to me as well, it's not just the OP) you're going against everything natural as there's only supposed to be a 3% chance that's happening.

Again, what does this have to do with streaks? My argument is that STREAKS are UNNATURAL because math.

But they aren't "unnatural". The error is in the belief that there is some "unnatural" pattern when it's really just chance playing out in a way that

**feels**unintuitive to your understanding of the nature of reality.

Here's another fun article: http://scienceblogs.com/cortex/2008/03/31/the-illusion-of-streaks/

05/08/2018 04:52 PMPosted by zenasprimeBut they aren't "unnatural".

At the frequency of which they occur, yes, they are.

Just for my own sanity, I wrote a small snippet of java code to demonstrate that my math is correct. I rolled a random value and counted the number of times the value was below .5.

https://pastebin.com/8s5UTPJ0

Here's a couple outputs.

The chance of a streak of 5 is 3.212%.

The chance of a streak of 5 is 3.127%.

The chance of a streak of 5 is 3.0509999999999997%.

The chance of a streak of 5 is 3.113%.

So yeah, continue to tell me there's something "natural" about a streak of 5 when your chance of it happening is only 3.125%.

Streaks ARE NOT natural in frequencies that we're experiencing.

May 9, 2018
-13

05/08/2018 05:10 PMPosted by DrothvaderAt the frequency of which they occur, yes, they are.

[citation needed]

You need some statistical data to back up that claim. You can't base your statistical model on the casual observation of a couple dozen games or the antidotal evidence presented here on the forums by frustrated players.

May 9, 2018
-11

05/08/2018 05:10 PMPosted by DrothvaderStreaks ARE NOT natural in frequencies that we're experiencing.

Show me the data that indicates that there is a suspicious frequency of win streaks across the entire player base.

I'll wait patiently...

05/08/2018 05:17 PMPosted by zenasprime05/08/2018 05:10 PMPosted by DrothvaderStreaks ARE NOT natural in frequencies that we're experiencing.

Show me the data that indicates that there is a suspicious frequency of win streaks across the entire player base.

I'll wait patiently...

Then you're going to be waiting a while because this isn't worth the effort of going through people's match histories on HOTSLogs to prove my point. Because let's face it, that's the only way I could "prove it" without you coming up with some other asinine counterargument.

If you want to live in delusion, fine.

May 9, 2018
17

05/08/2018 04:52 PMPosted by zenasprimeAgain, what does this have to do with streaks? My argument is that STREAKS are UNNATURAL because math.

But they aren't "unnatural". The error is in the belief that there is some "unnatural" pattern when it's really just chance playing out in a way thatfeelsunintuitive to your understanding of the nature of reality.

Here's another fun article: http://scienceblogs.com/cortex/2008/03/31/the-illusion-of-streaks/

Your argument is that streaks don't actually exist because the idea of probability events as combined links in a chain is incorrect. That essentially each event in a probability chain is completely independent of the preceding or following links. The events in the chain (coin flips in this case) do not affect each other, so "streaks" are nothing more than coincidence.

The problem with this premise is that it does not apply to a situation like HotS. HotS is not driven by probability and it never will nor could it ever be. It is driven by an algorithm that attempts to simulate probability. But it will never simulate probability with 100% accuracy, and even if it could accomplish a 50% win rate, it would still not be the same method as a coin flip. In such a scenario, streaks are not illusions because the events in the chain are actually linked. What happened in the last event (game) influences the outcome of the next event. The algorithm changes the conditions of the next event based on previous events. Treating games of HotS like coin flips is a false equivalency.

So you need a new premise to argue on.

May 9, 2018
-11

05/08/2018 05:22 PMPosted by DrothvaderThen you're going to be waiting a while because this isn't worth the effort of going through people's match histories on HOTSLogs to prove my point. Because let's face it, that's the only way I could "prove it" without you coming up with some other asinine counterargument.

it's not worth it because it's a fools errand to try and prove this conspiracy theory that Blizzard is intentionally manipulating game outcomes by providing less than fair matches.

May 9, 2018
-12

05/08/2018 05:23 PMPosted by LunafreyaThe problem with this premise is that it does not apply to a situation like HotS. HotS is not driven by probability and it never will nor could it ever be. It is driven by an algorithm that attempts to simulate probability. But it will never simulate probability with 100% accuracy, and even if it could accomplish a 50% win rate, it would still not be the same method as a coin flip. In such a scenario, streaks are not illusions because the events in the chain are actually linked. What happened in the last event (game) influences the outcome of the next event. The algorithm changes the conditions of the next event based on previous events. Treating games of HotS like coin flips is a false equivalency.

o.O

05/08/2018 05:24 PMPosted by zenasprime05/08/2018 05:22 PMPosted by DrothvaderThen you're going to be waiting a while because this isn't worth the effort of going through people's match histories on HOTSLogs to prove my point. Because let's face it, that's the only way I could "prove it" without you coming up with some other asinine counterargument.

it's not worth it because it's a fools errand to try and prove this conspiracy theory that Blizzard is intentionally manipulating game outcomes by providing less than fair matches.

No, it's not worth my time. I couldn't just use one or two people, no, you would easily dismiss that. I would have to use hundreds, maybe even thousands of players match histories to prove my point. Since the raw data dump is NOT tied to player ID's, there's no way I could construct an individual's match history without just programmatically collecting the data myself.

It's literally not worth my time to spend several hours collecting data, and parsing said data just to watch you jump through some other hoop.

First your argument was that streaks are natural, I can easily prove they aren't, now you're saying that people are just pretending that they're getting win / loss streaks all of the time. You've essentially flip flopped on your stance and now want proof that streaks do in fact happen at a frequency higher than they mathematically should. I'm not wasting more of my time with such a flawed argument.