The Role of Chance -- Why balance at 100k may not be enough.

Posts : 483 Join date : 2013-04-27

This post got long. TLDR in Blue.

RuneSlayer wrote:

Realf Lantow wrote:: Elven HI cannot face any other HI in PVP battle.

I have to disagree...In fact, Elves HI has around 0.5%-1% higher victory rate in 100.000 battles (simulation with both starting stats and capped stats) against ALL enemy races. Check your stats and gear up your HIs with items which will provide a better stat spread.

That is very "theoretically" balanced, and thanks for sharing that average Runeslayer.

That said, there are many posts from nearly every race claiming they have troubles. I know you will probably always get some of these complaints in an online game.

That is human nature and its not going away. Fair enough. I'm sure that is part of it, but I also want to make sure that Randomness is not contributing to a lot of these problems either. This is something only the devs can check.

When we talk about a simulation of over 100k battles, we can safely say that almost any amount of chance will get washed out. This is great for finding the theoretical balance point, but it does not necessarily mean that single PvP matches will feel right.

For example -- Coin flips will hit about 50% heads after 100k flips too, but no one thinks that if they flip a coin 10 times that they will have exactly 5 heads. People expecting to get somewhere in 4-6 heads will be very disappointed quite often because sometimes you will get 3 heads, or 8 heads. Very rarely you will even get 10heads.

To continue the analogy -- In BC, a single match is closer to the order of magnitude of "10 coin flips" rather than 100k coin flips. I think many PvPers are also expecting that they will get 4-6 heads almost all the time and are often seeing 3 heads or 8 heads and are wondering what is going on.

Obviously there is not as much chance in BattleCon as a coin flip, but... While it is good to know that units are balanced at the level of 100k simulated battles, the variation in unit winrates due to chance in a single match is important to players too.

I know this game isnt Warhammer or WoW or any other game - but there is something that all those successful games have in common -- They have all fine-tuned the amount of chance that feels "right" in their game at the correct level (1 arena battle, 1 tabletop game, etc).

I'm suggesting that it would be worth investigating the amount of chance that feels right at the level of a single PvP match in BattleCon. Furthermore - the amount of chance in the combat mechanics is very closely related to how skill-based you want the game to be. More skill :: Less Chance.

The good thing is - the devs probably already have the required data from the 100k simulations they ran.

===
--How can we determine the influence of chance in a single PvP match? --

We need to get a statistical distribution of how much the "average PvP match" varies due to chance. So, we want to know the following to examine the role chance plays:

In an "average" PvP match, how many 1unit vs 1unit battles occur?

I know this number will vary quite a bit, but we dont need a precise number. We just need to know the correct order of magnitude at which to examine chance. We are looking for a rough number that will allow us to relate a single PvP match to the large 100k data sets that the devs already have.

Lets call this number the Average Number of Battles in One Match (ANBOM). For the sake of this post, lets say that the ANBOM = 20.
If you randomly picked any 20 (ANBOM) 1v1 fights from the simulated 100k pool and found the average winrate of those 20 (ANBOM) fights, how does it compare to the theoretical near 50% winrate?
If that was repeated 100 times and the averages of ANBOM number of fights were plotted, we should get a normal curve showing the "Winrates of the Average PvP match."

That is the magic curve that tells us a lot.

We know that the center of that normal curve would be very close to 50% because the 100k simulation's average winrate was just about 50%. Thats great, but we need to look at the shape of the bell curve to gain insight on the role of chance.

If that curve looks more like a steep peak, then there is very little randomness/chance affecting the game mechanics and StdDev values would be very small. That means BattleCon is closer to a game of Chess.

If that curve looks like a wide hill, then there is a lot of randomness/chance in the game and StdDev numbers would be rather high. That means BattleCon is closer to flipping coins.

So the key number to pay attention to is the value of StdDev when looking at a large number of samplings that are correctly sized to represent 1 PvP match.

Obviously we want something in between Chess and Coin Flips, but I think everyone will agree -- if we want a game that is more skill-based, then it needs to be a lot closer to a game of Chess than a series of coin flips.

Posts : 782 Join date : 2013-07-19

Hello everybody, this is my first post in the forum (I finally decided to make an account).

This is an interesting topic, and I thought I might add my two cents. Keep in mind that this game is more complex than coin flips. Theoretically, if you say a battle is 20 engagements, and then simply flip 20 coins to see how close to 50% you get, you won't really get the full idea. The most important coin flips will be the earliest ones, because based on this kind of game logic, the more coin flips you win compared to the enemy, the more likely you will continue to win them (do to having the unit or strategy advantage). Therefore the real chance for a skill based instead of chance based match may actually not be as high as you think...

Allow me to elaborate with a real battle situation. Let's say that you are PvPing and you have 3 equal units vs 3 equal enemy units. Now, you've each got one flank on your opponent and all units engaged equally. But your enemy gets some unlucky rolls and kills 4 units with his flanking unit, while you quickly kill 10 units on lucky rolls with yours, causing the enemy to flee. Not only did the balance go off for a short time, but it's now become practically impossible for the other player to win, especially if their fleeing units don't turn around quickly. You send your unit(s) that caused the enemy to flee and flank another one of your enemies units, quickly causing him to flee as well and saving your flanked unit from fleeing. You quickly mop up the leftover units and suffer only mild casaulties, despite all things being equal at the beginning of the engagement, all because of luck at the very beginning. I don't know the exact programming or formulas of the game but from what I can tell a situation like this is not all that uncommon. In the end the real number of rolls that really mattered for the battle was the amount it took to get the first unit to flee, (a number which is unknown to me), and the chances of each roll [not as simple as heads or tails] depend on the unit stats and formulas. Anyways, the reason why such few rolls at the beginning having such a big effect on the battle could be a problem with the game's design is because it makes it feel like the game is the one playing and not us. We just watch helplessly and hope that our hero gets that hit in, or that our HI dodges the arrows. Of course we place the units to the best of our ability, but after that it's up to our stats and the random mechanics to decide our fate.

It's also important to note that some fluctuations within the game are at a far greater randomness than a coin flip... take the archers and artillery for example which do several hits each shot. It's possible for either every hit to land or every hit to miss, and then it's FURTHER possible for each hit to do more than one damage. This means one shot could kill nothing, or it could kill several or even more units. Don't get me wrong, the possibilities will line up near the middle with 100k shots... but say you get some bad shots in a row in a match, and your units with the bad shots have become nearly useless despite having high capabilities. This randomness is part of why I hated the Dark Legion HI before the hp boost/stat reduction, as sometimes I would lose several of my HI in no time at all just due to bad luck (not because of superior enemy troops). It's less random now that the hp is 2, so I have to give a big thanks for that.

With the above points in mind, I think it would be good to be more liberal about using randomness within a battle (especially a PvP battle). I do agree that the game would not be as fun if it had no random variability like a chess game, but as it is now I think it has gone a bit too far and matches are based too much on luck, especially with so much at stake when considering PvP matches: not only having to pay to heal yourself for what could have been the computer's mistake, but losing the entire reward for the battle as well. I believe it is a somewhat big reason why PvP is not played as often as it should. The risk vs reward of PvP is another topic altogether (sort of) but it's also something that may be worth reevaluating, as for now it seems rather unfavorable.

That all said, it's not a critical issue and it's something you should wait to look into until after you've dealt with the more important things like guild wars and improved PvP queue. It doesn't destroy the experience for me, I only would enjoy it more with around 1/2 the randomness it currently has. And I know digging into almost every battle formula in the game would be super fun.... =)

PS: I know it sounds complainy and that it seems like the forum is one big complaint basket, but don't lose hope developers. People complain when they enjoy something because they want to refine it or because it frustrates them when something is so close to being just right. I enjoy the game a lot, and I respect you guys for the work you are always putting in. In all honesty, I think that if people aren't complaining you are probably not doing your job right. If we didn't think you guys cared about the game or about the audience, we would probably complain less because we would think "Ah they wouldn't listen anyways", and in the end the game would suffer due to negligence. This is why I respect you guys.

Thanks for a fun game and for your dedicated effort into making the game better,

Bobba

PPS: Geeez that was longer than I intended.

Edit: Something I said was misleading... hopefully made it better.

Posts : 1043 Join date : 2013-07-09 Age : 41

Well Bobba for your first post I think it's more than all my posts combined Smile

but yeah you make some great points that it does seem like that when that unit flees that is game over.

Posts : 3124 Join date : 2012-11-13

Great posts...

I'm personally a vivid strategy fan, having played for many many years games like Warhammer Fantasy, Warmaster, Flames of War and other games.

As I have played more than a thousand games, most of them being Warhammer and Warmaster, I can say with certainty that wherever there is dice there is randomness. I've lost countless battles because the dice were not "friendly" with me. I am talking about cases where I had to have 2 successful rolls (4+) in 20 dice and failed miserably, while my opponent rolled 2 dice (4+) and had 2 successes. Take 10 people who have played these games and they will all tell you that there have to be changes to "balance things out". All of them having a different race.. Smile

Hegorn, first thing that needs to be verified and confirmed is the fact that when a unit goes head to head with another will have equal chances of gaining victory (starting stats AND capped stats). This is very important. Having confirmed that, then you start looking into mechanics to verify and confirm that special abilities or other mechanics are not distorting the balance. So far...we are happy with the results. What we are not happy about is...

PVP matchmaking... THIS is the mother of all complaints...and we can understand that.. A player creates an army with a Rookie unit fully equipped with Godlikes along with 2 other rookie units total worth of AP around 450. The player enters Lobby. A new player brings 6 units, with an "OK" setup of items, again worth 450 AP. Is that match up correct? Short answer..No
Please look here if you haven't already: http://www.battleconforum.com/t638-what-if We believe that this fail safe, along with the unit level reqs for items is going to provide some "relief" to the PVP matchmaking problem.

Bobba, thank you for your kind words. We are mostly driven by the love of community and though it can sometimes become very frustrating reading "angry" comments to the extent that they become "offensive", we understand that it is the passion that leads to such "explosions". We have great plans for this game and for this community. I wish I could share them with you....The people who will follow us will witness a huge transformation of BC in the future and will have the chance to play in other "worlds" as well... Smile

In regards to your example, I have to admit that this is something that we want to have in our game. It is part of the fun, watching your units decimating the enemy..Shouting at them to hold the lines...to not break as you have already sent another unit to hit your enemy's flanks...just...a...bit....longer...HOLD....! A charge that decimated 5 troops of the enemy (good rolls)...An artillery shot that went crazy killing a row of troops! (great rolls)...A giant unleashing his special ability "Crush" and sending to death 6 troops at once... These are the moments that really stand out...

Having said all that, and as I mentioned above, the PVP matchmaking is what has created frustration...and it is understandable...as currently it's not working as it should...It is mainly the reason why we delayed GW, as competition is very important and we want to provide a "fair ground" for everyone... Let us see how the new system will be..We have high hopes for it.

The Unyielding Posts : 802 Join date : 2013-05-31

Wow... Such a perfect thread, such great posts... Such numbers, examples, length of text <3

I don't have anything to add to it, so just gonna express my joy upon reading this, even if a bit offtopic Smile

4b, please write more bounce

Hegorn, thanks for starting this and great long post as ever Wink

Rune, really, so enjoyable! (Well my earlier suggestion "you can post less with more info" applies here, it's sooo nice to see such answer, that I could trade at least some of *my* questions for such posts. I can reask later or bother others, just...just write such things more)

Posts : 483 Join date : 2013-04-27

Heh. Thanks for the replies guys.

Bobba -- good points. I agree that the early rolls can snowball the battle. That amplifies the effects of chance and maybe we shouldnt be looking at measuring the total number of 1v1 battles in a single PvP match. It might be better to look at some fraction of that number.

I just wanted to start the conversation on this topic - especially since I believe more balanced items will highlight the issue of how much RNG affects combat outcomes.

How much reduction is needed? I personally think halving the amount of chance might be a good start, but it is really hard to say - partly because it is somewhat subjective.

I'm a numbers guy, so thats why I was curious about the Std Dev numbers. Comparing it to coin flips as a baseline for what is "completely random" still has its uses as a negative control. In the end though, it will come down to playtesting it and giving the devs good feedback.

Also, as a side note I think you meant "less liberal" in your post?

Quote :: With the above points in mind, I think it would be good to be more liberal about using randomness within a battle (especially a PvP battle)

Posts : 483 Join date : 2013-04-27

Rune -- I definitely appreciate that your 100k simulations found good ~50% winrates for units at base stats and capped stats. That is a huge first step to getting fun balance in Battlecon. That really cannot be understated.

Some games have a lot of other aspects to it that make it fun - like collectable/paintable/customizable tabletop games. Some of those games may be able to afford having more randomness because a lot of the fun is outside of actually doing battle.

Looking further at that though - I think the root cause of the complaints about item imbalances were because people did not feel like they had enough control over their battles - i.e. the outcomes of battles were based way more on items than on player skill. I do think the item limitations will help a lot with balance. I'll post my thoughts on the PvP tiers in that thread too.

I understand you may not want to turn the game into a Chess battle - I dont either, but there are many other things that would increase the skill cap in a fun way by increasing the amount of micro that players can do in battle. I'm advocating that player controlled actions should be the strongest actor on the outcome of battles.

Things like:

Controlling when abilities are used.
Being able to attack the ground with circle-area style ranged attacks. I wouldnt even mind if this was the only method of attacking at range.
Controlling stat development of units.
Fog of War adds a lot of options.
Flanking angles that matter at the degree level instead of on a per-side basis. Maybe show us unit hit/collision boxes to help with this (as an option)?
Controllable Unit Formations.
Controllable Unit Stances.
Movement speed and/or unit mass affecting charge bonuses and formations bonuses/penalties.

These could all increase the skill cap and push player actions as the dominating factor in wins - not chance, not items, and not necessarily in a chess-like way.

Btw, I dont think this is an impending issue that should delay GW - Just something for players to keep an eye on as they play under the new changes.

Posts : 782 Join date : 2013-07-19

Hegorn wrote:: Also, as a side note I think you meant "less liberal" in your post?

Hmm less liberal, more liberal, they could both work but to avoid confusion lets just say I meant "There should be less randomness in battle."

RuneSlayer wrote:: In regards to your example, I have to admit that this is something that we want to have in our game.

Fair enough, although I don't agree. I am always much more satisfied when something like that happens (whether in my favor or against it) because of difference of tactics (at least predominantly). I do agree that when PvP has been balanced it will be more fun for everyone, but I still think unlucky people could be turned away from PvP because they lose a lot of their first matches due to bad rolls and not poor strategy.

Hegorn wrote:: Controlling stat development of units.

Yes please! Very Happy

Fyrr wrote:: 4b, please write more

That's the plan, Fyrr. Smile