Spring RTS Engine

I was surprised when i watched this list once again:
http://modinfo.adune.nl/index.php?act=i ... s&MOD=ba66

It shows on the top exactly the units which are commonly used in most of the games, for example seaplane bombers, they are fucking ownage, same with adv bombers. gorg isnt so good though. And bantha should be higher, but still its near the top... and those boats should be lower rank maybe...

And there are some other units which shouldnt be anywhere near the top, but its pretty accurate if you wanna know what units you should use in some mod you are a newbie in.

I remember using that list in some mod which i never played, and it was worth it

Having the decoys at the top makes me think it doesn't take special damages into account.

TBH I'm not impressed by the ranking system (as itself, that is; as a website it is really neat and I use it a lot). It's like a horoscope: interesting but not ultimately helpful.

On the empirical side, here's some useful units that are nowhere near the top (of the 184 units listed):

Stealth fighters (#161 and #167)
Jeffy (#158)
Missile trucks (#148, #149)
Scout hovers (#132, #140)
Flash (#92)

On the theoretical side, the calculation is roughly equivalent to plopping down one of each unit next to each other and seeing how much damage they inflict, which is not very realistic--if you had a dozen Flash, would you send them at a Reaper one at a time? I wrote a bit about this myself, although even this is very approximate: http://trac.caspring.org/wiki/BasicEfficiencyModels

Finally, on the engine side, it doesn't calculate the DPS for bombers properly (since they do not use manualBombSettings), which is why they are on top. The calculated DPS for bombers is only for the 1 s or so that the bombers are actually dropping bombs, which doesn't include the time they spend circling around and reloading for each attack run.

#92 [armflash - Flash - Fast Assault Tank M/E: 109/914] armflash

simple math isn't enough to really show what units are useful and which aren't

If flash is at 92 your post is BS trademark

+ yeh with a crapton of special damages its hardly accurate

lurker wrote:Having the decoys at the top makes me think it doesn't take special damages into account.

I cant use the special damages, if i used, i would need to compare every unit to each other, make a fucking big list about every unit difference against every unit, which wouldnt really tell you any simple answer since all you would see is "which unit is best against which unit".

You would need to go through that big list to know how to play, though, that would be better in manner of learning, but much slower to read 184*184 = 33856 possibilities

NOiZE wrote:#92 [armflash - Flash - Fast Assault Tank M/E: 109/914] armflash

Evil4Zerggin wrote:TBH I'm not impressed by the ranking system (as itself, that is; as a website it is really neat and I use it a lot). It's like a horoscope: interesting but not ultimately helpful.

Yeah, me either, as i said i was surprised it works even that accurate with such simple math...

Evil4Zerggin wrote:On the empirical side, here's some useful units that are nowhere near the top (of the 184 units listed):

Stealth fighters (#161 and #167)
Jeffy (#158)
Missile trucks (#148, #149)
Scout hovers (#132, #140)
Flash (#92)

Those arent usually the game ending units though (which was the point of the list). But you are right, swarm is always better, though, why should it change ranks, since you can swarm all of the units?

Evil4Zerggin wrote:On the theoretical side, the calculation is roughly equivalent to plopping down one of each unit next to each other and seeing how much damage they inflict, which is not very realistic--if you had a dozen Flash, would you send them at a Reaper one at a time? I wrote a bit about this myself, although even this is very approximate: http://trac.caspring.org/wiki/BasicEfficiencyModels

Swarm is always better than sending one by one, everyone should know that... Shouldnt change the way how you calculate best units though.

Evil4Zerggin wrote:Finally, on the engine side, it doesn't calculate the DPS for bombers properly (since they do not use manualBombSettings), which is why they are on top. The calculated DPS for bombers is only for the 1 s or so that the bombers are actually dropping bombs, which doesn't include the time they spend circling around and reloading for each attack run.

Hmm, didnt know that, but those bombers are that good actually, should be on top anyways

Otherside wrote:If flash is at 92 your post is BS trademark

+ yeh with a crapton of special damages its hardly accurate

As i said before, i cant add special damages in calculations.

--

The list just gives a hint about best units, isnt really accurate, and will never be accuarate, since its impossible, because of the high amount of different possibilities.

trademark: try to group units by factory in that list.

TradeMark wrote:

lurker wrote:Having the decoys at the top makes me think it doesn't take special damages into account.

I cant use the special damages, if i used, i would need to compare every unit to each other, make a fucking big list about every unit difference against every unit, which wouldnt really tell you any simple answer since all you would see is "which unit is best against which unit".

You would need to go through that big list to know how to play, though, that would be better in manner of learning, but much slower to read 184*184 = 33856 possibilities

Average against unit categories it's allowed to hit, average damage against each unit, average damage against any unitit can hit, flat-out average among each damage type, there's no reason for you to add another dimension.


And swarms aren't always better; they have much more of an effect on some units than others.

lurker wrote:Average against unit categories it's allowed to hit, average damage against each unit, average damage against any unitit can hit, flat-out average among each damage type, there's no reason for you to add another dimension.

If i take average, still the decoys would be on top of the list :p
I think i have tried average in that list and results were fail. Ill try it again later if i was wrong.

lurker wrote:And swarms aren't always better; they have much more of an effect on some units than others.

Hmm, yeah, if you wanna kill krog, its not good to send swarm, since they will be blown up fast at once... though, its possible to kill krog by 20000 fleas. And swarm of AK's against levelers, will make AK's just explode at once :D

imbaczek wrote:trademark: try to group units by factory in that list.

You mean i would put every unit with value 2000-3000 in same group, and units with value 3000-4000 in same, etc? Doesnt really help anything, but makes Noize happier since flash rank would go higher


--

Instead of watching the rankings I would read that list like: flash is slightly worser than gator, but better than AK. etc.

I could sort the list by tech level though... or make 3 separate lists depending on tech level.

trademark: no, i just wanted to have this comparison per factory, not per whole mod so i can see which unit is "the best" in a single lab.

I think the best units are whatever a situation demands. With all the different playstyles you can never know until you're in game. All the flashes in the world won't help against a comnap and even the coolest bomber rush gets pwned by a fighter shield. Math may help to figure out the basics, but from there on it all depends on what happens in a game.

it doesn't consider speed, firing range, blast radius, energy cost, laser-falloff, etc. All you get is damage. This isn't Magic: The Gathering - there are more stats then offense and armor. It also doesn't account for

Although the fact that the LLT isn't very high up there explains why porcing in BA is such total fail.

This is about as useful as SA.

imbaczek wrote:trademark: no, i just wanted to have this comparison per factory, not per whole mod so i can see which unit is "the best" in a single lab.

oh lab... i thought you meant factor lol.

yeah i thought the same... i could try that later.

Pxtl wrote:it doesn't consider speed, firing range, blast radius, energy cost, laser-falloff, etc. All you get is damage. This isn't Magic: The Gathering - there are more stats then offense and armor. It also doesn't account for

yea, since its hard to calculate those.. sometimes speed doesnt matter so much etc... also the turn rate is important, if your units changes direction and turn rate is nearly zero, you are screwed then... acceleration is important too, idk how to add all those in one formula which would rank them...

Get the SA balancing formulas. :D

KDR_11k wrote:This is about as bullshit as SA.

fixed

trying to use simple maths to determine the best unit ignoring most factors is epic fail

TradeMark wrote:

Evil4Zerggin wrote:On the empirical side, here's some useful units that are nowhere near the top (of the 184 units listed):

Stealth fighters (#161 and #167)
Jeffy (#158)
Missile trucks (#148, #149)
Scout hovers (#132, #140)
Flash (#92)

Those arent usually the game ending units though (which was the point of the list). But you are right, swarm is always better, though, why should it change ranks, since you can swarm all of the units

Game-ending isn't necessarily the best or most-commonly used. If the game lasts two hours, then this list isn't too bad for the endgame--at this point, the battle is more about how much power you can pack into each square elmo of the map and less about how much things cost. If the game lasts ten minutes, though, the list is useless.

Swarming should change ranks because it affects how units perform against each other. If you send Flash one at a time against a Reaper, it might take ten to destroy it, and you might say that the Reaper is more cost-efficient than the Flash. On the other hand, if you send five Flash at the same time against the Reaper, you might destroy it, and say that the Flash is more cost-efficient than the Reaper.

The reason the list works better in late game is due mainly to two things: packing and splash damage. In the early game, people may only be able to afford a dozen Flash, so they can spread them out to avoid splash damage and allow all of them to fire at the same time. However, as economy grows, people can afford more or heavier units. With a hundred Flash, only a few can get in range at a time, and they cannot spread out to avoid splash damage. Thus, one Bulldog will not do well against eight Flash, but ten Bulldogs will against eighty Flash. This is why games progress from Gator/Flash -> Raider (not so good for its role though)/Stumpy -> Reaper/Bulldog -> T3.

If you want an earlier-game calculation, I would suggest something like sqrt(DPS * health) / cost, which corresponds to an ideal swarm (this tends to favor small units). In contrast, the current calculation is DPS * health / cost. Or you can try something in between, like (DPS * health)^0.7 / cost. Obviously none of these take anything into account besides DPS, health, and cost.

Put another way, the list uses Lanchester's Linear Law, but many situations are better modeled by Lanchester's Square Law. (http://en.wikipedia.org/wiki/Lanchester%27s_laws)

A note on bombs: without manualBombSettings, the reloadTime determines how many bombs the bomber drops (provided it is below 0.5 s); the bomber actually takes 5 s to reload its bombs, while the attack run takes ~14 s. I'm not saying bombers aren't good, just that they aren't for the reason the list thinks they are.

You calculation is way too simplified...add speed,area damage,range energy cost and build time.
Than you will get much better results,trust me.

KDR_11k wrote:This is about as useful as SA.

Fact:SA is the most balanced OTA mod.