But tbh outside of team games we basically never see arb vs halbs.
Bec it doesn’t makes any sense.
And the change wouldn’t make HC more viable…
1 Like
We do see Arbs vs Halbs, as they are both very common, and form a neat triangle with Hussars and Cavaliers.
dis sound like thebest solution for it
1 Like
Sounds a bit traditionalist to me.
There’s no tech like arson to emphasize it.
2 Likes
I threw the following scenario into my Calculation Sheet:
Opponent has teched into halb and producest them with 80 vills. You have the option to tech into Arbs, HC or Elite Skirm to counter. As it is only a addition to your current cavalry force you don’t want to spent your whole eco to that addition. So I chose for the beginning about 60 vils are a reasonable amount of vills invested into that switch.
For the Arbs and Skirms I chose to count all upgrades except the attack ones, as they are very likely to have already for sake of your defenses. Still the investment into Arbs/Skirms is way higher than into HC. In case of the arb the tech-in makes about 31 % of their cost whilst the hand cannon tech ratio is almost half of that (16%).
With this I got a equilibrium of numbers with 56 Halbs: 34 arbs, 27 HC, 42 skirms.
Then I threw these results into https://aoe-combatsim.com/
With medium hit and run.
Result:
Arbs: 30 Arbs left - 4 lost
HC: 15 HC left - 12 lost
Skirms: 25 Halbs left - 31 killed
Now I checked how berbers xbows would do:
Berbers xbows: 30 xbows left - 7 lost
Meaning, in that reasonable sit Berbers would do better to tech into xbows than into HC.
Now I got Curious and threw Teutons xBow, lacking bracer and thumb ring, into the ring:
Teutons xbows: 20 xbows left - 22 lost
One of the worst archer civs in the game - and it almost makes sense to tech into xbows for teutons, almost as much sense as teching into HC. I think this is the reasoning behind the bad state of HC. Even in the Sits that seem to be suited for them, they can’t even outperform the archer line lacking several of the key upgrades. No wonder they have that bad state.
Quick note to skirms: They seem to underperform, but ofc they make more sense in the bigger context, as you will ofthen face halbs with ranged support. Then you will also invest more into the skirm switch and have a bigger mass. Then halbs have problems against them. This test only showed investing less to counter halbs and skirms can’t really shine in that situation. With equal investments skirms easily win (and ofc cost no gold).
PS: If you want I can calculate the Unique counters + champions also, to see how they perform in this situation.
Edit: I found a mistake in the calc of the halberdiers, there I assumed falsely only 60 workers to push them out. As it only applies to the halbs I will leave it there. I don’t think it changes the assertion qualitatively. Just keep in mind that the assumed halb numbers are lower than they would be if that situation occurs in a real game. There it would be about 75, not 56.
Exactly which techs are you buying? I can’t make your total costs match up.
Also it’s extremely questionable a player would preemptively get more than fletching and ballistics for defenses and maybe bodkin. You really only need the range preemptively for forward castles. Chemistry and Bracer are especially questionable given the high base attack of castles.
Also even ignoring these your Teuton examples shows players should not go xbow. They had to invest 40% more gold and lost 65% more gold in the exchange for the units alone.
2 Likes
I calculated with a dept-pay-off over time.
So the techs aren’t paid immediately rather indirect bec I let the units cost more.
And I only took away the archer attack techs, not chemistry, thumb ring or balistics.
If you try to calc with like only a mass of x vs 75 halbs in one single engagement you can never get serious results, as the tech-in costs in the lategame only make sense if there are several engagements.
My approach is quite complex, if you want to understand what I’m doing there read the link I posted above.
This doesn’t make sense in this context. You can use intertemporal discounting using a discount function (e.g. exponential discounting) to make costs at different times comparable but you can’t arbitrarily extend the time horizon of the payment. All upgrades and resources have to be paid for before the fight begins and the production time is very short, around 3 minutes for 6 ranges. Over this time period you’re only going to be talking about discounting things at most 20%.
You can amortize the fixed costs of the techs obviously over the expected number of units, but that amortization has to be discounted as well. I.e. instead of dividing by total number of units produced you divide by the time-discounted number of units. But that future production should be explicitly indicated, otherwise no one can reproduce the numbers being used.
I use a pay-off over time model. It never divides by the amount of units produced or any value corresponding to it.
The Techs (and all investments) are paid of over time.
And as I don’t know when the game ends I need to assume it’s infinite.
lol
Okay. Let’s start from the top:
- He’s counting all resources equal
- He’s assuming the ranged attack upgrade are already going to be paid for (the game example I gave him that he hates so much didn’t get the ranged attack upgrades at all, therefore he’s literally pulling this expectation out of thin air) and thus don’t count towards the Arb upgrade.
Those two factors are enough to make this test absolutely hilarious. Nobody looking at this with even a basic understanding of the game will contend this test makes any sense. In Imperial, so gold should be at it’s maximum value and yet he evaluates it even on gold. That way, he can make the HC look much more expensive than the Arbalest, since it’s got a much higher overall cost even though it barely costs more gold.
I’ve already explained this before. In order to make up the overall gold cost spent on upgrading Arbalest with buying Hand Cannons, after all upgrades are taken into account, You’d need to buy 180 HC which is where the value lies. So the
Is an absolute joke. Next?
He wants to tech these to counter halbs. As in, these weren’t part of your master plan until 80 Halbs just showed up. Meaning you need to run your blacksmith through six techs, your archery range through three techs, your University through two techs, and this massive transition time is not being accounted for at all in his cost evaluation even though, for like the sixth time, this is what I pointed out as the purpose for Hand cannons. To be a fast transition.
Hand cannons need Chemistry, and then they’re practically fully upgraded by the time the first is out. The only way you could come close to that speed (at like, four times the overall cost, mind you) is if you just assume the Cav player decided to get all his archer attack techs earlier. As I mentioned, the example video I showed, an excellent example of HC making a massive difference in a spot they needed to work out, the Cav player did not tech attack upgrades and therefore he’s still spitballing. There’s a chance this has been done if he wanted to use them for a surprise transition. There’s no guarantee. The fact he still insists on just assuming it’s happened in an attempt to strengthen his argument is getting a bit absurd.
1 Like
But that doesn’t accurately represent reality. The main problem here with how you are doing it is your model cannot differentiate between a short production of units (e.g. a quick counter) and a full on tech switch followed by significant production. You are discounting the techs identically in both cases despite the lifetime use of one being specifically longer than the other. This may work for GAAP or IFRS but that doesn’t reflect economic reality in AoE2
The game has no loans. All techs must be paid up front. For estimating the cost of units you include the amortized cost of the techs. E.g. if you make 20 units instantly but want to calculate the cost of 10 of them, you divide all fixed costs by 2 and add the variable cost of those 10 units.
If they aren’t produced instantly you will attempt to spread the present value of the tech over the units depending on when they are made. E.g. a tech which benefits 2 units, one now and one in 10 minutes will attribute 3/4 of its cost to the first unit and 1/4 to the second unit (at 6.9% interest). This is done to better reflect the sunk cost of fixed costs. E.g. fletching on arbs is insignificant if it was researched a long time ago for archers.
Doing it any other way will give misleading present values because the game has no negative resources or loans. It only has time preference. These are distinct concepts, and economic models with one but not both have very different implications.
Nope. I don’t count them equal. Read it first please. I also made gold actually make worth about 20 % than the gather rates suggest, to account for it expending. Don’t make false claims. It’s described in the calc sheet…
Yes And I will also assume for the champs upgrades that attack upgrades have already been paid for. It’s accounting for real gameplay. If you don’t see that, you seem to have no real understanding about what scenario we are speaking about.
Yes your answer is a joke. A bad joke. Next?
Nope, impossible. HC won’t work as a fast transition either. You need to spot it early as any other transition you want to make. HC transition isn’t fast or cheap either. Sorry.
They need the archer armour upgrades, too. Even against pure halbs they will save you a lot of HC in the process. Seems you never had experience but make just false claim after false claim.
HC without archer armor are complete trash.
When I read your posts like this, it comes clearer and clearer to me that you just hate HC addition to the game. ANd that’s ok. But please stop making false claims here to fool other people our work on figuring out why they are in that bad state would be wrong.
It’s really annoying to see somebody just trying to destroy a serious conversation trying to figure out what is happening there with stupid killer arguments. I put a lot of work into this, and your argumends don’t defy it. You just make false claims.
That’s exactly how I actually do it. If you have read my thread about that method, there is a small passage where I explain how I calculate the pay-off of the techs. It’s a time-delay of the units.
I just wanted to give you a chance to figure out that I do it the exact way you want it. But if you are too lazy to read it, I can’t help it.
The way I calc is I calc how long it takes to tech into these units. And then I recalculate how much they would have been cost at that time when you started to tech into them.
So the “higher cost” from the tech-in is actually relied to the time you spent to tech into the units. It’s quite smart to make it like this, because this way the time delay is included in all later investments into that unit too, regardless of the amount you spend into them.
2 Likes
Look I just don’t see how that’s valid.
With 60 villagers consider the following:
- 0 float → wait 20s for chemistry → wait 45s for 9 ranges → wait 35s for archer armors → 20s (chemistry finishing) → 12s (not enough res for 9 ranges) → start all 9 ranges on HC. 77s later finished. Total time: 209s
- 0 float → wait 45s for 9 ranges → wait for 20s for resources for thumb ring → wait 35s for resources for archer armor → wait 33s for resources for archer upgrades → wait 20s for chemistry → wait 15s for ballistics → 20s (not enough res for 9 ranges) → 9 range archers → 81s later finished. Total time: 269s
Build time for the archery ranges is neglected here for simplicity. But if included both would go up by around 10s.
At least one of a few things is problematic:
- You are allowing units to come out at different times, which can only be compared by properly discounting
- You are ignoring the fixed costs of the archery ranges
- You are assuming there is a large enough resource bank at the beginning which can be “borrowed” from and paid back.
Regardless of what you are doing as far as I can tell your model has no built-in way to prevent this kind of stuff from being identified and fixed. This is what I am getting at.
Now if this is just a modelling assumption and you aren’t actually discounting units that’s fine. But it sounded like you were discounting units, which is why I don’t understand what youre doing.
FYI I have a degree in economics so you don’t have to sugar coat any of the math if that’s whats holding you back. Talkin about integrals, differential equations, discounting whatever if it helps
This I fix with basically waiting till all upgrades are done.
Nope, they are also included in my calculation. The ranges make about 6 % of the cost.
Nope, As I said Above I calculate as how long it takes to get the res for the upgrades.
My approach can’t be perfect. None can. And I’m aware of the mistakes I make. But you actually didn’t found them yet. Biggest mistake: I don’t acount for previously done investions and not used ressources. Because you can’t implement this.
But my approach is way better than everything else i’ve seen so far. Til yet I could debunk all the claims how wrong it would be. What is very funny, cause I know there are still a lot of uncertainties etc in it.
There is no perfect calculation for that. But I try to get as near as possible. If you have something relevant I would be glad to include it into my approach.
If you read carefully what I wrote you should get it. If I would need to tell you every little detail I did, it would be as long as this thread here. But if you have a specific question I can answer it, no problem.
I dont understand this. So you allow a batch of units to finish at different times? How do you compare them if one comes out earlier than the other? Are you discounting or not?
Because I assume it’s not worth to make an engagement with them before all upgrades are in. SO I wait until all upgrades are done, basically.
EG alls units made before that engagement “lose” worth because they need to wait for that engagement.
What is the raw/nominal value of villager seconds used for the arbalest and HC examples in your model?
You mean the basic cost to train a single unit?
It’s 271.4 for HC and 171.1 for Arb.
Note that a lot of this comes from HC costing food which in this model is quite an expensive ressource, as it isn’t only gathered comp. slow, it also costs wood preemtively to be gathered.
Ok well those numbers aren’t exactly far off from many approximations so I’ll skip to a more direct question.
The standard way to do this is to have a production function of time, say p(t). At time t=X, you spend Y. You then multiply this by a discount function e^-rt, say d(t). So the present value (PV) of any particular function is just p_object(t)*d(t). Integrate to get total value over whatever interval expressed at the point in time of t=0.
This isn’t hard to explain so I’m inclined to think you are doing something very different. For example in the hand cannon/arb example you’d be constrained at any one point in time X by integral(vil(t)*t,dt,0,X) villager seconds where vills changes depending on building ranges or not. Assuming your production function involving your technologies and ranges and satisfies this constraint, you just discount it by multiplying. These production functions are pretty easy to express in a spreadsheet or data structure like an array or python list. You can also pretty easily make families of them and put probability weightings on them.
But this also means you’d never decrease the benefit of units just because they haven’t fought yet. Their cost increases if you choose a point forward in time, but that’s different.
So are you doing this or something different?
You mean, like that example of real game play at the absolute highest levels of competition that tells you the exact opposite of what you are considering “real gameplay” because it doesn’t support your position?
Unfortunately, your expectation of “real gameplay” doesn’t line up with observable reality. As such, you keep making absurd claims, like asserting that you’ll be getting the ranged attack upgrades every game as civs that don’t get Arbalest because it’s just bad if you don’t.
I will remind the conversation, the video posted in this post has a HC vs mayans game 3, and in that game, this top 10 player in the world did not get archer attack upgrades as Persians.