I always had my doubts about the "enemies will randomly hit" in Battlegrounds, because i observed that certain minions like Baron Rivendare, Soul Juggler or Mechano-Tank are either the very first target after taunt is killed, either they are avoided as hell until they boosted enough to stand a chance/win the game.
I also tried to avoid the bias and put everything on pure luck, but i also see a lot of other people complaining about the same behaviour. Since i'm not a guy that takes such obvious patterns as lucky coincidences, i started using my skills to create a simple bot that reads, parses and analyses HS logs. The idea was simple: how many rounds my key minions are killed as a primary target when there is no taunt on the table, and how many times the key minions luckily survive.
The results are pretty interesting and either i am the unluckiest man alive, either this game is scripted as fuck.
Out of almost 600 games, which generated 7745 rounds, i had Rivendere in 278 games and 1946 rounds. 76% (!!) of these rounds, my Rivendere was the very first minion to be sniped when the taunt was gone (or not at all on the table).
Soul Juggler. I played 167 games against Soul Juggler, more precisely 948 rounds. Here are the % of my hits on it:
7 cards on table - 1% direct hit 6 cards on the table - 1% direct hit 5 cards on the table - 2% direct hit 4 cards on the table - 9% direct hit (!) 3 cards on the table - 27% direct hit 2 cards on the table - 60% direct hit
How it is possible that Rivendare is sniped at a rate of 76% with 7/6 cards on the table, while the Soul Juggler is avoided 81% of the time with just 4 cards on the table? The very very similar stats we get with Mechano-Tank.
Yes, i know, it's a bias. I also know "Why would Blizzard script the game? What is the point?" rhetoric. However, keep in mind that Battlegrounds is nothing more than a slot machine mode, where the chances of hitting one of the 3-4 meta builds are actually very low in a game of 8 players, especially if you apply the same math that it's used to explain the visible patterns of key minions sniping. Therefore, the game engine compensates by providing very rare hits/sequences, generating spectacular lucky moments almost every game if not every round in some situations.
Behind "is a luck game" you can easily hide scripted behaviour. It's too easy and the benefits for Blizzard are way too huge to avoid this. Battlegrounds is literally a dopamine machine that survives only because you can achieve super mambo jambo combos, but if you'll step back and analyse the ods of this happening so often, you'll understand that's not random at all.
Now this is certainly an interesting study you did. I'm always reluctant with calling out things, when random is included, but you got some pretty radical numbers with a lot of observations. I like that you are still pointing out biases too. What if you did a statistic analysis? I'm not an expert on this but maybe a chi square would do? Just to see what the actual odds are for getting such a result. Let's not forget how many games are played and you might have done what's equivalent to rolling a double six and isolated that. Still, great that you tried this and for now you have an indicator of a scripting of odds.
Assuming that this script is a thing:
Is this rigging? Sounds like this script favors specific builds rather than people, perhaps for balance purposes even.
Do I experience it myself? Not really. There's always that game where baron lives forever and that where he's killed first and of course I feel like the game hates me ;)
This has certainly some more merit than the usual "its all rigged"-posts but as YourPersonalGuru pointed out a proper statistical analysis of the data would be helpful. But that aside i wanted to point out that baron dying earlier than other minions could also be because of its usually lower health compared to a lot of other minions. Baron rarely has more than its standard 7 health while most minions in the later stages tend to have much more.
Are these 948 rounds with no taunts on the board? Or are these the total rounds, but the stats are only from the rounds with no taunt. If so, how many cases were used for each percentage?
It‘s quite a nice analysis but I think it‘s the partially also the usual problem with statistics and too much data, which can‘t always be used
for example: when you say „they sniped it“, do you mean they killed it or does it also account for when it‘s hit…. because then your answer for the difference could be, that minions on tier 3/ a board with max tier 3 minions have a much bigger problem killing a 5 health minion (juggler) in one hit than tier 5 minions/tier 5 board have with killing baron in one hit… also there is then a chance to be zapp‘ed (which is in general quite difficult for jugglers) especially nowadays wherw zapp isn‘t the first minion often to play around unfrogged rats/rat packs
I think it would be easier to record say 100 rounds of Rivendare being in combat and analyze the footage, that way you can get a better idea of what may be causing any oddities.
I can't say too much on the topic, i'm not as regular a player - but i do spend a LOT of time when i do play being annoyed that my minions flat refuse to take out the soul juggler, like, ever :P
Add in zaps chance of attacking baron late game and it adds in other variables. I often pocket one for people with barons. Also I would argue soul juggling is more mid build, it's pretty useless after that so again available minions and consideration of baron being t5 and soul t3 etc.
This is interesting. You got the bot's output recorded (i.e. the actual parsed data to be analyzed regarding number of rounds)? First time I'm seeing anything of this ilk actually backed up by data.
I presume all of those baron rounds were without taunts or Zap, and we're talking about baron being the first attacked, rather than just first to die? Just making sure we're all on the same page here!
So couple questions about this and the study you did:
Did you account for cards such as Zapp being played?
Is the minion stats of board accounted for throughout the rounds? As your board is typically much larger than your Baron, that means that they can sometimes tank extra hits whereas Baron is a small, weak-statted minion by the time he gets played and dies in 1 hit. To account for this, do you have statistics on how often your Baron, or other "key cards" are attacked first? Dying first is very easily skewable in that comparison.
Going back to your current metric of first to die - Baron being compared to Juggler, and to an extent mechano, is an iffy comparison. They go in different builds. Juggler survives to be a later minion alive with your analysis, sure, but if you think about how a juggler demon builds, that makes more sense that their board will be thinner, especially as voidlord is a common tool for jugglers. By the time that voidlord and all its spawned taunts die, your board is going to be down to less minions (4-5) on board by the time that Juggler can even be hit. Baron is more often played with beast or mech comps (more beast), where they don't have returning taunts. So once your first taunt is down, everything is immediately open, Baron immediately vulnerable. Mech builds are fairly similar to beasts in this way that they typically don't have returning taunts, once their initial ones are gone. Though it's thrown off more by divine shields. Mechano and Baron both in mechs are comparable, mostly, when used in mech comps, but not other comps. If you have them both on board in any of your games, you'll have to account for that, your board positioning, and cleave. As baron usually goes further back, closer to taunt, and more readily cleaved. So other things to consider with your program would be to account for cleave and how often that impacts your results as well as what comp it is being played in. If you have Baron and Juggler (or mechano) on board, which dies first more often? If you have Baron in a juggle demon comp instead of Juggler, does baron die noticeably more often?
Very interesting preliminary research and statistics though, has me curious for sure. Makes me want to see further data on this for sure! Some of those things I mentioned above I think would be particularly interesting to see accounted for in data because otherwise, currently, the data is interesting but has enough potential holes and confounding factors for me to consider it proof of non-RNG being used (yes, it is significant as of now, just not sure it's necessarily telling).
Quite interesting. At least someone can get data of some sort on this issue.
I'd have to echo what some other people have said though which is that given Baron's low stats at a point in the game when hit by pretty much anything it will die. Also, Juggler builds tend to have lots of taunts in the way since there are lots of demon taunts that spawn more demons for more Juggler hits. And yea, how many times was Baron "sniped" by Zapp?
Overall, I'd say it's a good start. However, it does seem to be biased. The bias in this case is trying to make the stats justify the conclusion without looking at (or just not publishing) all the details.
I would like to see a large amount of data collected on this. Lately I've been frustrated that my Sefin seems to always have taunt and getting attacked immediately, Once again there is the case of it being a low stat minion at that part of the game. So I am also just praying nothing hits it and I'm also emotionally invested in it by that point too which isn't good for rational thinking.
I will say this though. A while back Kripp had mentioned something about it being confirmed by Blizzard that cleave minions hit the corner minions more often than would be expected just in general. So there may be something there. Just need a reliably unbiased, and preferably open source data collection and analysis.
This is definitely a cool study and I would love to see if there was a way the community could easily track stats like this. It does make sense that while the battles are randomized, there would be some sort of "Priority" the AI takes when attacking certain stuff, which might lead to all those moments where an important card gets sniped.
I was just wondering a day or two ago about whether or not I had the energy to expend on proving that Soul Juggler is blessed during attack determination. It seemed so obvious, but I am also aware of biases and am data driven. Thank you for saving me the time and effort.
Yeah most people worth their salt go out of their way to protect their high value low statted assets, otherwise you just accept the risks of running those builds without proper protection.
But I win with Rivendare builds regularly. That woiuld be impossible if he was snped first 76% of the time. Beasts and Mechs rely on him heavily..
The thing here is that in the case of beasts, 1 taunt is usually enough to guarantee success cause that way you guarantee that your bird hits at least once and then with stuff like multiple (or golden) birds or wolf or by giving them reborn is usually enough for your comp to work and win games even if your Baron gets sniped immediatly after your one taunt dies.
In the case of mechs Baron can also be incredibly important but in this tribe he's usually protected by many taunts that have divine shield on top of it and some of them can even re-gain divine shield many times (deflecto and mecherel) so it's very rare for Baron to die early. The only cases I can think of being either your opponent having Zapp in which case bad luck and shouldn't be addressed in this study of sorts the OP is doing or is just straight up terrible positiong, in which case should be considered either cause it's player error.
Well, I am happy that you did this so I didn't have to because I have similar stats. I started just writing down statistics because I saw the possibilities were so far from RNG at most times you just had to laugh especially on certain cards that were targeted and this went both ways since I also noted when I was in so-called luck's favor.
Of course, You can't prove much without documentation and data so I think I will start doing just that.
Rollback Post to RevisionRollBack
To post a comment, please login or register a new account.
I always had my doubts about the "enemies will randomly hit" in Battlegrounds, because i observed that certain minions like Baron Rivendare, Soul Juggler or Mechano-Tank are either the very first target after taunt is killed, either they are avoided as hell until they boosted enough to stand a chance/win the game.
I also tried to avoid the bias and put everything on pure luck, but i also see a lot of other people complaining about the same behaviour. Since i'm not a guy that takes such obvious patterns as lucky coincidences, i started using my skills to create a simple bot that reads, parses and analyses HS logs. The idea was simple: how many rounds my key minions are killed as a primary target when there is no taunt on the table, and how many times the key minions luckily survive.
The results are pretty interesting and either i am the unluckiest man alive, either this game is scripted as fuck.
Out of almost 600 games, which generated 7745 rounds, i had Rivendere in 278 games and 1946 rounds. 76% (!!) of these rounds, my Rivendere was the very first minion to be sniped when the taunt was gone (or not at all on the table).
Soul Juggler. I played 167 games against Soul Juggler, more precisely 948 rounds. Here are the % of my hits on it:
7 cards on table - 1% direct hit
6 cards on the table - 1% direct hit
5 cards on the table - 2% direct hit
4 cards on the table - 9% direct hit (!)
3 cards on the table - 27% direct hit
2 cards on the table - 60% direct hit
How it is possible that Rivendare is sniped at a rate of 76% with 7/6 cards on the table, while the Soul Juggler is avoided 81% of the time with just 4 cards on the table? The very very similar stats we get with Mechano-Tank.
Yes, i know, it's a bias. I also know "Why would Blizzard script the game? What is the point?" rhetoric. However, keep in mind that Battlegrounds is nothing more than a slot machine mode, where the chances of hitting one of the 3-4 meta builds are actually very low in a game of 8 players, especially if you apply the same math that it's used to explain the visible patterns of key minions sniping. Therefore, the game engine compensates by providing very rare hits/sequences, generating spectacular lucky moments almost every game if not every round in some situations.
Behind "is a luck game" you can easily hide scripted behaviour. It's too easy and the benefits for Blizzard are way too huge to avoid this. Battlegrounds is literally a dopamine machine that survives only because you can achieve super mambo jambo combos, but if you'll step back and analyse the ods of this happening so often, you'll understand that's not random at all.
Peace.
Now this is certainly an interesting study you did. I'm always reluctant with calling out things, when random is included, but you got some pretty radical numbers with a lot of observations. I like that you are still pointing out biases too. What if you did a statistic analysis? I'm not an expert on this but maybe a chi square would do? Just to see what the actual odds are for getting such a result. Let's not forget how many games are played and you might have done what's equivalent to rolling a double six and isolated that. Still, great that you tried this and for now you have an indicator of a scripting of odds.
Assuming that this script is a thing:
Is this rigging? Sounds like this script favors specific builds rather than people, perhaps for balance purposes even.
Do I experience it myself? Not really. There's always that game where baron lives forever and that where he's killed first and of course I feel like the game hates me ;)
This has certainly some more merit than the usual "its all rigged"-posts but as YourPersonalGuru pointed out a proper statistical analysis of the data would be helpful.
But that aside i wanted to point out that baron dying earlier than other minions could also be because of its usually lower health compared to a lot of other minions. Baron rarely has more than its standard 7 health while most minions in the later stages tend to have much more.
Are these 948 rounds with no taunts on the board? Or are these the total rounds, but the stats are only from the rounds with no taunt. If so, how many cases were used for each percentage?
It‘s quite a nice analysis but I think it‘s the partially also the usual problem with statistics and too much data, which can‘t always be used
for example: when you say „they sniped it“, do you mean they killed it or does it also account for when it‘s hit…. because then your answer for the difference could be, that minions on tier 3/ a board with max tier 3 minions have a much bigger problem killing a 5 health minion (juggler) in one hit than tier 5 minions/tier 5 board have with killing baron in one hit… also there is then a chance to be zapp‘ed (which is in general quite difficult for jugglers) especially nowadays wherw zapp isn‘t the first minion often to play around unfrogged rats/rat packs
I think it would be easier to record say 100 rounds of Rivendare being in combat and analyze the footage, that way you can get a better idea of what may be causing any oddities.
But I win with Rivendare builds regularly. That woiuld be impossible if he was snped first 76% of the time. Beasts and Mechs rely on him heavily..
I can't say too much on the topic, i'm not as regular a player - but i do spend a LOT of time when i do play being annoyed that my minions flat refuse to take out the soul juggler, like, ever :P
Add in zaps chance of attacking baron late game and it adds in other variables. I often pocket one for people with barons. Also I would argue soul juggling is more mid build, it's pretty useless after that so again available minions and consideration of baron being t5 and soul t3 etc.
Definitely interesting and proper approach! Mind sharing the script/bot? Then we could collect more data from others here and look deeper into it.
This is interesting. You got the bot's output recorded (i.e. the actual parsed data to be analyzed regarding number of rounds)? First time I'm seeing anything of this ilk actually backed up by data.
I presume all of those baron rounds were without taunts or Zap, and we're talking about baron being the first attacked, rather than just first to die? Just making sure we're all on the same page here!
So couple questions about this and the study you did:
Did you account for cards such as Zapp being played?
Is the minion stats of board accounted for throughout the rounds? As your board is typically much larger than your Baron, that means that they can sometimes tank extra hits whereas Baron is a small, weak-statted minion by the time he gets played and dies in 1 hit. To account for this, do you have statistics on how often your Baron, or other "key cards" are attacked first? Dying first is very easily skewable in that comparison.
Going back to your current metric of first to die - Baron being compared to Juggler, and to an extent mechano, is an iffy comparison. They go in different builds. Juggler survives to be a later minion alive with your analysis, sure, but if you think about how a juggler demon builds, that makes more sense that their board will be thinner, especially as voidlord is a common tool for jugglers. By the time that voidlord and all its spawned taunts die, your board is going to be down to less minions (4-5) on board by the time that Juggler can even be hit. Baron is more often played with beast or mech comps (more beast), where they don't have returning taunts. So once your first taunt is down, everything is immediately open, Baron immediately vulnerable. Mech builds are fairly similar to beasts in this way that they typically don't have returning taunts, once their initial ones are gone. Though it's thrown off more by divine shields. Mechano and Baron both in mechs are comparable, mostly, when used in mech comps, but not other comps. If you have them both on board in any of your games, you'll have to account for that, your board positioning, and cleave. As baron usually goes further back, closer to taunt, and more readily cleaved. So other things to consider with your program would be to account for cleave and how often that impacts your results as well as what comp it is being played in. If you have Baron and Juggler (or mechano) on board, which dies first more often? If you have Baron in a juggle demon comp instead of Juggler, does baron die noticeably more often?
Very interesting preliminary research and statistics though, has me curious for sure. Makes me want to see further data on this for sure! Some of those things I mentioned above I think would be particularly interesting to see accounted for in data because otherwise, currently, the data is interesting but has enough potential holes and confounding factors for me to consider it proof of non-RNG being used (yes, it is significant as of now, just not sure it's necessarily telling).
Quite interesting. At least someone can get data of some sort on this issue.
I'd have to echo what some other people have said though which is that given Baron's low stats at a point in the game when hit by pretty much anything it will die. Also, Juggler builds tend to have lots of taunts in the way since there are lots of demon taunts that spawn more demons for more Juggler hits. And yea, how many times was Baron "sniped" by Zapp?
Overall, I'd say it's a good start. However, it does seem to be biased. The bias in this case is trying to make the stats justify the conclusion without looking at (or just not publishing) all the details.
I would like to see a large amount of data collected on this. Lately I've been frustrated that my Sefin seems to always have taunt and getting attacked immediately, Once again there is the case of it being a low stat minion at that part of the game. So I am also just praying nothing hits it and I'm also emotionally invested in it by that point too which isn't good for rational thinking.
I will say this though. A while back Kripp had mentioned something about it being confirmed by Blizzard that cleave minions hit the corner minions more often than would be expected just in general. So there may be something there. Just need a reliably unbiased, and preferably open source data collection and analysis.
I am not surprised.
I have had numerous rounds where I called out exactly what would happen ahead of time and it did on boards with 7 minions.
Me: "Ok, with these 2 boards ... the only way opponent can win ..."
Me: "Is if he goes first, his minion hits my X minion, and then my Y minion hits his Z minion, and his W minion hits my V minion."
My thought was there might be a luck variable set at the beginning of the game.
Staticially dumb stuff happens at too high a rate in Battlegrounds. If this is true, do they do it to spice up the games?
Anyway, if you have a taunt minion, Battlegrounds has far less flexibility to dick you over.
This is definitely a cool study and I would love to see if there was a way the community could easily track stats like this. It does make sense that while the battles are randomized, there would be some sort of "Priority" the AI takes when attacking certain stuff, which might lead to all those moments where an important card gets sniped.
Wow Rain Man! Maybe you get another movie with Tom cruise.
I was just wondering a day or two ago about whether or not I had the energy to expend on proving that Soul Juggler is blessed during attack determination. It seemed so obvious, but I am also aware of biases and am data driven. Thank you for saving me the time and effort.
Sounds like you need more taunts.
Yeah most people worth their salt go out of their way to protect their high value low statted assets, otherwise you just accept the risks of running those builds without proper protection.
The thing here is that in the case of beasts, 1 taunt is usually enough to guarantee success cause that way you guarantee that your bird hits at least once and then with stuff like multiple (or golden) birds or wolf or by giving them reborn is usually enough for your comp to work and win games even if your Baron gets sniped immediatly after your one taunt dies.
In the case of mechs Baron can also be incredibly important but in this tribe he's usually protected by many taunts that have divine shield on top of it and some of them can even re-gain divine shield many times (deflecto and mecherel) so it's very rare for Baron to die early. The only cases I can think of being either your opponent having Zapp in which case bad luck and shouldn't be addressed in this study of sorts the OP is doing or is just straight up terrible positiong, in which case should be considered either cause it's player error.
Well, I am happy that you did this so I didn't have to because I have similar stats. I started just writing down statistics because I saw the possibilities were so far from RNG at most times you just had to laugh especially on certain cards that were targeted and this went both ways since I also noted when I was in so-called luck's favor.
Of course, You can't prove much without documentation and data so I think I will start doing just that.