Beta4 Bonus System Question

[LichKing]

But I don't think we even agree on what units are priced right

Definitely not

About dragons I meant under priced, not overpriced, sorry. 2000-2100 would be fine for me.
The 1 turn units aren't over priced, if there's a problem with negative stack maybe you solve that.

[KGB]

LichKing,

Not all of it can be solved by Negative stack stuff. Some of the cost issues are now related to balance changes.

Examples:

Lt Infantry 10/10 - 14 moves, 100 gold (where 10/10 represents combat when attacking/defending)
Hv Infantry 10/15 - 12 moves, 200 gold

Something isn't right there as it costs 100 more gold for less movement and the same ability when attacking? Yes you get a benefit on defense but if half the time you attack and half the time you defend then that benefit is only worth 50% more, not 100% more so 150 gold would be right for Hv Infantry.

Eventually at some point in a future beta, every ability/skill/strength/bonus/hit point/movement point etc is going to have to go into a giant formula to determine the cost value of a unit so that everything is priced right. When DLR was undergoing re-balancing this is what was done to avoid units being useless/under priced/over priced etc.

Example:

The Mammoth now has 3 hits. How valuable is that extra hit? Do you have any idea what that equates to?

Well thanks to DLR (where everything was all worked out), we know that a 30 strength 3 hit unit is exactly equal to a 40 strength 2 hit unit. Put another way, in straight combat with no bonus, a Mammoth is equal to a Dragon (the winning percentage turns out to be exactly 50%. If they both bless the Mammoth winning % moves to 52%). Currently the Mammoth is a 3 turn unit costing 1300 gold. It should probably cost quite a bit more given it's really a 40 strength unit with 2 hits that can be produced in 3 turns making is a VASTLY better deal than the Green Dragon (which takes 5 turns and only gives +2 extra moves).

KGB

[LichKing]

Well thanks to DLR (where everything was all worked out), we know that a 30 strength 3 hit unit is exactly equal to a 40 strength 2 hit unit. Put another way, in straight combat with no bonus, a Mammoth is equal to a Dragon (the winning percentage turns out to be exactly 50%. If they both bless the Mammoth winning % moves to 52%). Currently the Mammoth is a 3 turn unit costing 1300 gold. It should probably cost quite a bit more given it's really a 40 strength unit with 2 hits that can be produced in 3 turns making is a VASTLY better deal than the Green Dragon (which takes 5 turns and only gives +2 extra moves).

I considered units with 3 HP more or less 3/2 as strong as units with 2, now you say 4/3, but it's the same. Incidentally the green dragon has also 3 HP, so it's also quite strong. (Kraken now has also 3 HP..)

About 1-turn units: I think the current balance is almost perfect. There is a reason to buy all of them and in similar proportions. LiInf is as effective in attack as HeInf, but 5 points in defense mean a lot. HeInf are the cheapest unit that can have 15 in defense, so it's the best town defender. Pikemen are also good, but cost almost double and are slow, so you buy for open terrain (bridges etc), dwarves for hills and short range attack (and they cost a lot of upkeep, so you buy them with caution), orcs for swamp and elves for forest (elves are the strongest right now, maybe). So, even if HeInf costs double as LiInf and has double upkeep, in reality is still worth the money, above all if you're not sparing.

I don't think you have to see only the math, you have to see how they play in reality. I think the balance for them is thousand times better than in beta3, I wouldn't go back in this.

[KGB]

LichKing,

I considered units with 3 HP more or less 3/2 as strong as units with 2, now you say 4/3, but it's the same.

It's all based on strength. So a 6/3 = 8/2. But a 4/3 < 6/2 (45% instead of 50/50). Thus the need to construct a cost formula between strength/hits.

So, even if HeInf costs double as LiInf and has double upkeep, in reality is still worth the money, above all if you're not sparing.

A unit that costs 2X as much upkeep and is 2x as much to buy is not worth the money unless it is 2x as good (which it's not close to being). You *believe* it is worth it but the math doesn't support that belief.

I don't think you have to see only the math, you have to see how they play in reality. I think the balance for them is thousand times better than in beta3, I wouldn't go back in this.

Actually the math 100% defines how the units play. That's why in Beta3 all anyone made was Hv Infantry because the math supported it (cost vs benefit). In Beta3 all anyone made was Lt Calv (they were 3 strength) for the same reason.

And I agree the balance is better in Beta4 than Beta3 for the 1 turn units. It's not 100% there yet, but it's moving toward it. My point is that they still need adjustments and eventually creating a large spreadsheet will the way to go. Simply because if you decide to make a change / add a new unit you instantly have the new cost for the unit instead of guessing and potentially unbalancing things.

KGB

[LichKing]

A unit that costs 2X as much upkeep and is 2x as much to buy is not worth the money unless it is 2x as good (which it's not close to being). You *believe* it is worth it but the math doesn't support that belief.
KGB

You don't consider one aspect. At turn 1 you have around, say 500-1000 gold with total income of 100. 2 gold of upkeep are near to nothing in comparison. If you get HeInf instead of LiInf, at turn 2-3 and on you'll have 25-50% more strength in defense. A crow has a chance against LiInf (wall 5), against HeInf no. You have to pay for this advantage, and 100 gold is not much. LiInf is better to attack, being more convenient in this sense. You buy a unit once, the advantage is cumulative in time. For the same reason I think strong units like dragons/devils/archons should cost more.

If you want to do it with the math, you can't say 2x cost=2x strength and 2x upkeep. Upkeep counts little, unless is very high like dwarves and light cavalry. 2x strength means to me at least 4x cost. Moreover, when you want to make stronger stacks, you look for the strongest units possible, since max is 8. Compare 8 LiInf attacking 8 HeInf. That 50% means 100% victory.

If you want to do it mathematically, you should have a battle calculator, consider the average winning chances on all kinds of terrain, when attacking and when defending, behind walls (considering a 1 turn unit gets 4 times the wall bonus a 4 turns unit gets), and all this repeated for each unit against all kinds of units (anti-air is situational), and how would you calculate swarm? And movement points? And the ability to fly? And all special abilities (negate, cancel terrain, move bonus, etc)? And stack bonus? It depends on the number of the units it is applied to, how do you calculate that? On average of 4 units? And in towns? It's senseless, all this would be arbitrary, not mathematical. Even if you did that, and it would take ages, it would still be wrong.

Consider also: 2 turns units are 4 times stronger than 1 turn units? no, but they cost double and more. Light cavalry costs 300, heavy cavalry 725, in 4 turns you have 18x4 vs 35x2, almost the same. So should they cost the same? But when you're on open with a hero that could be attacked by dragons or who knows what, you don't want light cavalry, you want heavy, because you can't pack more than 8 units in a stack.

It all depends on particular situations, and with mathematic you couldn't solve exactly the problem, as one should pretend from math.

You really want to get near the ideal cost? then you should ask (maybe good) players what they would pay for each unit, after they have sufficient confidence. I don't have it yet, for example, but I can tell you that I always plunder spiders and would plunder Griffins on sight (never saw one, lately), and I love Minotaurs instead.

Actually the math 100% defines how the units play. That's why in Beta3 all anyone made was Hv Infantry because the math supported it (cost vs benefit). In Beta3 all anyone made was Lt Calv (they were 3 strength) for the same reason.

In beta3 1-turn units were almost as strong as 2-turns units. And personally I bought different kinds of armies, because I liked it more and I even won some games (ok not many) just because other players had only infantry, and I had less but more varied units.

[KGB]

LichKing,

You don't consider one aspect. At turn 1 you have around, say 500-1000 gold with total income of 100. 2 gold of upkeep are near to nothing in comparison. If you get HeInf instead of LiInf, at turn 2-3 and on you'll have 25-50% more strength in defense. A crow has a chance against LiInf (wall 5), against HeInf no. You have to pay for this advantage, and 100 gold is not much.

It's a terrible deal. Imagine you repeat this for 10 cities. Player A buys 10 Hv Infantry. Player B buys 9 Lt Infantry. At this time he has 1100 extra gold to spend and buys a Pegasi for his 10th city. Now, which player would you rather be?

Plus you handicap yourself with slower expansion due to 12 moves instead of 14 and the Hv Inf does not help with city capture, only defense which you may *never* need. Paying for things you may not need is not a good idea. You should never buy Hv Inf, only make those you find in cities.

For the same reason I think strong units like dragons/devils/archons should cost more.

Agreed. 2000 for the Devil and 2400 for the Dragon and 2200 for the Grand Archon would be about right. You can't make it too expensive or players won't be able to afford one.

If you want to do it with the math, you can't say 2x cost=2x strength and 2x upkeep. Upkeep counts little, unless is very high like dwarves and light cavalry. 2x strength means to me at least 4x cost. Moreover, when you want to make stronger stacks, you look for the strongest units possible, since max is 8.

Agreed. The cost formula is *not* linear. Never claimed it was.

Compare 8 LiInf attacking 8 HeInf. That 50% means 100% victory.

Agreed. So that justifies the 200 cost because 8 beats 8? Instead Here is an example of what I mean. I would say that if a 10 strength unit costs 100 (Lt Infantry) then 50% of the time that is what the Hv Inf is (offense). So it's costs starts at (100/2) = 50. Now you have to decide what a 3 strength unit is worth (I don't know, but if we use 200 as a cost, the current price of Hv Infantry) then you need 50% of that cost (on defense) which is 100 (200/2)=100 and the total cost is 50+100=150 (about what I think the unit is actually worth). Instead if you think a 3 strength unit is worth 300, then the cost would be 300/2=150 + 50 = 200. So the question is, do you think a 3 strength unit is worth 200 or 300? Those are the things that get worked out in the formula.

If you want to do it mathematically, you should have a battle calculator, consider the average winning chances on all kinds of terrain, when attacking and when defending, behind walls (considering a 1 turn unit gets 4 times the wall bonus a 4 turns unit gets), and all this repeated for each unit against all kinds of units (anti-air is situational), and how would you calculate swarm?

I have such a calculator. Wrote one ages ago. They are easy to write. Wrote one way back in the DLR days too for the same reason.

And movement points? And the ability to fly? And all special abilities (negate, cancel terrain, move bonus, etc)? And stack bonus?

These are easier than you think to cost. Especially movement points which is fixed (ie unit always has same movement points). Stuff like terrain can be estimated (for example there is open, forest, hills, swamp, snow, desert. 6 terrains. You get the % of each terrain on maps to determine how valuable each bonus is. So if on the maps we find that there is 2X as much open as forest then an open bonus is roughly 2x as valuable).

It's senseless, all this would be arbitrary, not mathematical. Even if you did that, and it would take ages, it would still be wrong.

It's not senseless. One of the nice things about having DLR as a base of reference for Warbarons is that DLR has been around in *serious* hardcore multi-player since 1996. There has been 16 years of gaming experience and hundreds of thousands of multiplayer games played. A lot of time was invested in those first 5 years modeling everything in the game to get is balanced well. You can *easily* tell when you get something wrong because:

A) Virtually no one uses the unit
B) Virtually everyone uses the unit

If you don't get either of those situations over several thousand games you know you got it right. That's how the models developed.

It all depends on particular situations, and with mathematic you couldn't solve exactly the problem, as one should pretend from math.

As I said, you can solve it or come very close to solving it. The goal is to model the costs of everything. Then it's up to the players in a particular game, map, situation to do the best they can with each skill, bonus etc. So on maps with lots of Hills, Dwarves and Giants are things you buy/build because that situation demands it.

You really want to get near the ideal cost? then you should ask (maybe good) players what they would pay for each unit, after they have sufficient confidence. I don't have it yet, for example, but I can tell you that I always plunder spiders and would plunder Griffins on sight (never saw one, lately), and I love Minotaurs instead.

That's obviously how you estimate costs for various bonus's by starting with the players and asking what they think each bonus/skill is worth. Then you put those in a table and see if they hold true for all units of the same type (turn to produce). At the same time you create a 'base unit' for 1 turn, 2 turn, 3 turn, 4 turn, 5 turn. By base I mean lets use the Lt Infantry as the Base 1 turn unit (10 strength, no bonus, 14 moves, cost 100). So 10 strength, 14 moves costs 100. From there you can build all the other 1 turn units. So maybe each extra move >14 costs 20 gold (linear) or maybe each extra move costs successively more 10, 20, 30, 40 gold (scaling). It doesn't take long to get the costs right.

Incidentally I also plunder spiders/gryffons on sight too. Stack bonus's are about the only valuable thing I want on units that take 3+ turns or expensive 2 turn units. Inflating stacks of Lt Infantry with cheap bonus's (Crusader, Pegasi, Dragon, Medusa, Devil) is the best thing you can do other than inflating blessed crows with the same.

KGB

[LichKing]

It's a terrible deal. Imagine you repeat this for 10 cities. Player A buys 10 Hv Infantry. Player B buys 9 Lt Infantry. At this time he has 1100 extra gold to spend and buys a Pegasi for his 10th city. Now, which player would you rather be?

I didn't say that I would do this. If I did that, this would mean that HI would be *always* better than LI, this wouldn't be balance. I actually buy them in similar proportions (considered that I buy also other units), then I use LI to attack and HI to defend. That's the way I do, at least. If you lowered the price (or even worse the upkeep) of HI, I wouldn't buy LI anymore for sure. This means to me that now they're balanced. Ask yourself, I lower the price for HI, would I then buy LI anymore?

Moreover, you don't need buy a lot of HI to be better defended, 1-2 is usually enough, because attack units like LI have higher turnover (if you don't lose), so your example 10 LI vs. 10 HI is doubly misleading. On the long run, LI is more convenient, if you have Pegasi (and want to use them defensively, not a great deal imo), but at the beginning of the game to spend 100-200 gold more for a couple of HE is a good investment, imo. Often you find it, so you don't even have to buy them.

Consider also that it depends on the type of game. In small maps(like Desert/Ice Showdown or Lonely Island), or maps like Eight Kingdoms, where you meet the opponents very early, a slightly better defense can be vital.

You can *easily* tell when you get something wrong because:

A) Virtually no one uses the unit
B) Virtually everyone uses the unit

That's the point. An empirical method, based on players' experience cannot be replaced by an entirely mathematical one. In your calculator you can compute STR/terrain/swarm/anti-air and stack bonus(maybe), but you can't compute movement (e.g. slow movement for spiders, an attack unit, is a much bigger fault than for pikemen, who can go around also alone on grassland), nor terrain move bonus (it's map dependent), nor other things like Negate. If you test Devils in the battle calculator, it will have an average performance against most units, much worse than Dragons. But in reality, a player who has only dragons mostly loses against one with devils (a 4-turns unit, now). In my limited experience at least.

What I mean is that many elements are arbitrary even with a battle calculator. It could give some indications, but not much more. I mentioned spiders and griffins, I can add to the list demons, elementals, elephants, often mammoths (I'd get a green dragon instead), in 1 vs 1 ghosts too.

So the question is, do you think a 3 strength unit is worth 200 or 300? Those are the things that get worked out in the formula.

300 in my opinion.

[KGB]

LichKing,

Ask yourself, I lower the price for HI, would I then buy LI anymore?

Actually if you lowered the price I would buy HI. Right now I *never* buy HI. I only use what I find but never buy them because they bring nothing useful for their cost (I'd rather spend that extra 100 on something else)

but at the beginning of the game to spend 100-200 gold more for a couple of HE is a good investment, imo. Often you find it, so you don't even have to buy them.

Strange, that's the worst time to buy them because you are unlikely to be defending in the early turns (<10) unless you started *really* close to someone. If you are worried about crows, Elves are the better early investment since they kill crows really well, even blessed ones and their 12 strength on offense, extra moves (so they can go with a hero without slowing him down) and woods bonus is far superior to HI.

Getting them for free is another matter of course. You can build those but buying them is not smart.

That's the point. An empirical method, based on players' experience cannot be replaced by an entirely mathematical one.

Well you start with what you have/players experience. Put all that into a spreadsheet and formula and then test which units/bonus/strengths/skills etc are correctly priced. This type of thing is done *all* the time in the real world. On far more variables than Warbarons has (in my online Stanford AI class they are talking about doing this for tens of thousands of variables for complex stuff like trading algorithms, military stuff etc.) so it can be done and it is important to do (Esp if one day they want to add an AI which must do this kind of thing).

but you can't compute movement (e.g. slow movement for spiders, an attack unit, is a much bigger fault than for pikemen, who can go around also alone on grassland), nor terrain move bonus (it's map dependent), nor other things like Negate.

Movement is one of the easier things to compute/rank since it's absolute (a unit always has same movement it's not subject to bonus's). Each point is worth so much in terms of what you are willing to pay for it. You can determine how much more someone would pay for 1 more move etc to work out the value of 1 movement point for a 10,15, 20 ... strength unit. It's much easier than you actually think it is otherwise how would players know what is a good value and what isn't.

300 in my opinion.

Interesting. So you see

10 strength = 100
15 strength = 300
20 strength = ?
Can you fill in the rest. Even roughly. If they go up 200 a shot, a 25 strength unit would cost 700. There are 25 strength units that cost less than that (Giant only costs 575). So 25 strength is <525 since the Giant also gets extra moves, move bonus on hills and combat bonus on hills. So I'd stay 25 strength is worth <500 at least on a 2 turn unit.

If I asked you about how you felt about a 2 turn, 30 strength unit (14 moves, no movement bonus, no combat bonus) and it's cost what would you guess it to be? 600, 700, 800, 900, 1000, 1000+?

These are the kinds of answers the game needs to be able to add new units or make changes in future Beta's.

KGB

[magian]

The only time I build Heavy infantry, is if I expect the city to be attacked in the next few turns, and I don't have enough cash for something better. They could use a discount. I think 150 is about right, although I would still build light inf in many situations.

[KGB]

Interestingly enough, today I was attacked in a game in the hills with Dwarf+Dragon+Hero.

The total bonus for the Dwarf was 15+5(attacking)+10(hills)+30(max stack bonus after Hero+Dragon) for a total of 60! If those Dwarves were blessed they would have been 65! Despite having some Dwarves of my own + a hero, I was easily slaughtered.

Not sure this is exactly the way the bonus system should work as getting +45 seems just a bit over the top.

KGB