Battling in the android app

[DeletedUser27062]

I started playing Elvenar only 8 weeks ago (am starting chapter 4) and don't generally play fighting/battle games but I'm keen to learn. Elvenar doesn't offer much of a learning curve when it comes to fighting and I managed to kill off my entire army before figuring out the painfully slow regeneration rate. I was left with no choice but to negotiate which effectively destroyed my game play. Each encounter would cost me an entire days production of goods. That city still hasn't recovered fully. I now have 4 user profiles and 17 cities, the majority of which are purely for training myself to fight. I don't want to keep building cities to figure this fighting thing so I have some questions that I hope someone can help with.

1) I mostly play Elvenar on android and I know that there is no option to fight manually but I've been having a really tough time trying to figure out what the android app's AI 'cares' about when fighting battles in provinces. For example, I read that Orc Strategists do better in Planks provinces so should I take that info into consideration when forming my team? How much of the web based info (initiative, start position, damage etc) is used in the android version?

2) I often use Elven Architect battle simulator to test out potential teams so as to avoid costly losses especially if I'm pushing the limits of my capabilities. A few days ago I simmed the following:
Enemy: Squadsize 162, troops -Mist Walker x 2, Orc Strategist x 1.
Me: Squadsize 90, troops - Sword Dancer II x 2, Golem x 3.
The simulator predicted a victory and gave it a score of 811 out of 1000 which is a very good score.
I played it out in the app and was completely destroyed. Such losses hurt.... a lot. I know the squadsize disparity is a problem but I thought given I was putting up 5 units to their 3 that the numbers were doable ( enemy troops 486 to my 450). How do I know when to battle if I don't know what the variables are? Why is the android app making me lose something I probably would have won (even auto-fighting) if I'd played the web based version?

Any advice/help would be much appreciated.

 

[ajqtrz]

The problem with the simulator is that it doesn't take, and can't take terrain into account. If it shows you with a massive win, but you lose anyway, it's probably that some of your troops were blocked from getting into the fight. This is especially bad with light melee as they move so slow and and weak to begin with. Against ranged with only one exit from behind a long vertical row of hedges makes them extremely weak. So it was probably terrain -- which you can't know before you pick your troops and start the fight (isn't that realistic?).

AJ

 

[samidodamage]

@ajqtrz posted while I was typing the following, so what he said, lol!
I never played another fighting/battle game either before this one.
Though I now fight almost 100% of the time, others may be better able to explain this outcome. The biggest issue I see with fighting on the app is the inability to see the terrain. I think the AI is the same for app and browser as the actual battle takes place on the server (I could be wrong about that!). In the above fight, my guess is there were objects on the battlefield your Sword Dancers had to navigate around, keeping them from reaching enemy units before they were severely damaged. Mist Walkers have the highest initiative of any unit, player or enemy. With a move range of 4 and a strike range of 4, they're almost guaranteed to land a hit on one of your units before you can make a move. So, they are just picking away at the Sword Dancers as they bunch up and try to get around the barricades. The Orc Deserter (not an Orc Strat, that's a player unit that comes from the training grounds and is specialized against light range) is specialized against light melee and likely got to your already damaged Sword Dancers before they even reached the Mist Walkers. I think that range is more important than almost anything else in battles in this game. I'd bet that fight would have been won if you had fielded 5 Golems because of their movement range of 3 and strike range of 3. They'd have gone after the Mist Walkers first, then ganged up on the Orc Deserter.

 

[DeletedUser27062]

The problem with the simulator is that it doesn't take, and can't take terrain into account. If it shows you with a massive win, but you lose anyway, it's probably that some of your troops were blocked from getting into the fight. This is especially bad with light melee as they move so slow and and weak to begin with. Against ranged with only one exit from behind a long vertical row of hedges makes them extremely weak. So it was probably terrain -- which you can't know before you pick your troops and start the fight (isn't that realistic?).

AJ

Firstly, thanks for your reply, I appreciate it. Yep, I know the simulator doesn't/can't account for terrain but I'm not convinced the app's AI does either - well, maybe just some aspects of it. I'm sure it doesn't account for obstacles because it would be hugely unfair to expect people to battle blind which would be the case if there were obstacles we couldn't see nor account when strategising. However, I have a sneaking suspicion the AI does consider the 'flavour' of the terrain; that is to say that certain units perform better in certain provinces (which we're told is the case for PC gaming). The reason I suspect this is because I remember putting my darling Treant up against a single unit type where he'd always smashed them without taking any damage yet in a different province but the same squadsize/unit type he took a few smacks, nothing major though. The only variable that was different was the province.

I can't even find any detailed information about battling in the app vs PC which surprises me. I thought Inno might have at least told us what things we need to consider when going to battle other than simply pointing out which units are weak/strong against another. I know the app was never intended as a full game experience hence only having autofight but I know there MUST be a set of criteria they use to determine wins & losses.

Re: not knowing the terrain before picking troops - it seems Inno does a few sneaky things like that including allowing enemies to field more than 5 units and fielding troops way beyond the players game level. If I don't have access to Mist Walker (or something similar) then I shouldn't be expected to battle her IMO. All battles should be winnable so long as I haven't massively over-scouted.

@ajqtrz posted while I was typing the following, so what he said, lol!
I never played another fighting/battle game either before this one.
Though I now fight almost 100% of the time, others may be better able to explain this outcome. The biggest issue I see with fighting on the app is the inability to see the terrain. I think the AI is the same for app and browser as the actual battle takes place on the server (I could be wrong about that!). In the above fight, my guess is there were objects on the battlefield your Sword Dancers had to navigate around, keeping them from reaching enemy units before they were severely damaged. Mist Walkers have the highest initiative of any unit, player or enemy. With a move range of 4 and a strike range of 4, they're almost guaranteed to land a hit on one of your units before you can make a move. So, they are just picking away at the Sword Dancers as they bunch up and try to get around the barricades. The Orc Deserter (not an Orc Strat, that's a player unit that comes from the training grounds and is specialized against light range) is specialized against light melee and likely got to your already damaged Sword Dancers before they even reached the Mist Walkers. I think that range is more important than almost anything else in battles in this game. I'd bet that fight would have been won if you had fielded 5 Golems because of their movement range of 3 and strike range of 3. They'd have gone after the Mist Walkers first, then ganged up on the Orc Deserter.

I could literally listen to you and ajqtrz talk about battle strategy all day! I can't wait until I have that knowledge and experience! I feel like the kid sitting in the corner chewing crayons whilst the grownups talk Very impressed!

I have developed a simple tool for helping me determine my team and it works 99% of the time. I almost always win (and I'm battling 90%, negotiating 10%). I imagine it won't work very well once I get into serious battling (I'm only starting ch.4) and it's only really applicable for use in app battles as it's a blunt tool. But the fact that it does work as well as it does says to me that the app's AI must be more of a simple numbers game rather than the more complex PC version.

Incidentally, you may have figured out why things worked out differently in the simulation - I might have selected the wrong team member (orc deserter vs strat) as I'm still learning their names (and still call HR the 'sparkle boys' because of their icon lol).

 

[ajqtrz]

I'm thinking that while you may be right about the AI knowing the terrain, it isn't necessary for it to do so and still get the results you got. If you think about it, if the AI also doesn't know the terrain then the chances of you getting "bad" terrain for the troops you pick is about even with the chances of the AI getting bad terrain for the troops it picks. So X times out of Y battles you'd get clobbered unexpectedly but, the same ratio would be applicable to the AI. In other words, you'd be clobbered one out of X number of times but the AI would get clobbered at about the same rate. The thing is, you notice when you get unexpectedly clobbered, but you probably don't notice as much when the AI gets unexpectedly clobbered. It would be interesting to do an experiment (and tedious to do so as well) where you ran the simulator, noted the results and catalogued them. You'd have some definition of "clobbered unexpectedly" and would count the number of times you got and the AI got it. I'll bet, over time, they would be about the same. However, if you got it more than the AI you could conclude one explanation could be the AI had an advantage of knowing the terrain. Technically though, the AI should get it a bit more than you since you pick in response to it and thus have the more informed position. It pick blind to what you can do, and you pick knowing what it has done.

All in all it would be an interesting, though tedious, experiment.

AJ

 

[Deleted User - 4646370]

I'm sure it doesn't account for obstacles because it would be hugely unfair to expect people to battle blind which would be the case if there were obstacles we couldn't see nor account when strategising.

I think it does. And yes, it's unfair. I think the fight is actually simulated step by step on an actual battleground withnbstacles. I came to this conclusion after noticing a corremation between the time taken to determine the result of the fight and the result itself (easy wins generally takes shorter to compute), which can be explained by number of rounds.

 

[samidodamage]

Here's my 'wall of text' for the day, lol!

I think the fight is actually simulated step by step on an actual battleground withnbstacles

I think you're right. After all, on browser you can start a fight on manual, switch to auto once on the battlefield, watch the fight and see how the AI handles things. That was the single biggest help to me in learning combat.
I also think AJ is right, the AI makes equally bone-headed moves for both sides, lol! That's why manually fighting battles decreases losses; I can make moves the AI won't.
For me, viewing the battlefield is about more than just terrain. It also helps me with troop placement, even in the absence of obstacles. I do better when I can see the battlefield at figuring out the best way to line up my troops. It's really time consuming, though, so I generally just place them opposite the unit I'm putting them in to handle. That isn't as much help with auto, though.

AI must be more of a simple numbers game

The AI does not target units based on which other units they are best against. The AI targets the opposing unit within range that has the highest initiative. It seems to also take the movement and strike range of the opposing unit into consideration, so for example the long strike-range cannoneer may target a long strike-range Blossom Mage instead of an Archer even though the mage has a slightly lower initiative than the archer. All of these unit characteristics are assigned number values, so I think the AI is a simple numbers game even if I have no idea what formulas are used.
Here's an example of how that might work:
I put in a Mage to handle a heavy melee (ex: enemy Orc General); if there's an enemy light melee (ex: dog), the AI will first send my Mage to strike at the dog that is specialized against Mages (because the dog has higher initiative and longer movement range than the heavy melee), but the Mage will get mauled when the dog gets its turn. Surviving to strike the Orc General is unlikely, lol!
Also, the AI seems to always advance on the opposing units even when waiting out a turn or moving away from an opposing unit and/or behind a barrier would be the better strategy. The AI will also sometimes ignore the fact that a unit could have been moved to a space where they could strike the desired unit without moving into range of a different opposing unit that should be avoided and move the unit into range anyway. Again, you can take advantage of these bone-headed moves when fighting manually.

certain units perform better in certain provinces (which we're told is the case for PC gaming)

The guides that talk about this (player-generated; Inno has never explained any of this stuff afaik) remain valid for map province battles. Recent changes to the tourney structure means this no longer holds true for tourney provinces. Tourney provinces now always have 5 enemy slots. Certain enemy unit types have a higher probability of appearing in a specific tourney province, but it is now possible to face all 5 unit types in one battle in the tourneys. The units that are more likely to appear in each tourney province type are:
Marble: Heavy Melee
Steel: Mage
Planks: Light Range
Crystal: Mage & Heavy Ranged
Scroll: Light Ranged & Heavy Ranged
Silk: Light Melee & Heavy Melee
Elixir: Light Range, Light Melee & Heavy Melee
Magic Dust: Light Melee, Mage & Heavy Ranged
Gems: Light Melee, Heavy Ranged & Heavy Melee

 

[DeletedUser27062]

I think it does. And yes, it's unfair. I think the fight is actually simulated step by step on an actual battleground withnbstacles. I came to this conclusion after noticing a corremation between the time taken to determine the result of the fight and the result itself (easy wins generally takes shorter to compute), which can be explained by number of rounds.

That's a really good point; I've noticed that too.

 

[DeletedUser27062]

I'm thinking that while you may be right about the AI knowing the terrain, it isn't necessary for it to do so and still get the results you got. If you think about it, if the AI also doesn't know the terrain then the chances of you getting "bad" terrain for the troops you pick is about even with the chances of the AI getting bad terrain for the troops it picks. So X times out of Y battles you'd get clobbered unexpectedly but, the same ratio would be applicable to the AI. In other words, you'd be clobbered one out of X number of times but the AI would get clobbered at about the same rate. The thing is, you notice when you get unexpectedly clobbered, but you probably don't notice as much when the AI gets unexpectedly clobbered. It would be interesting to do an experiment (and tedious to do so as well) where you ran the simulator, noted the results and catalogued them. You'd have some definition of "clobbered unexpectedly" and would count the number of times you got and the AI got it. I'll bet, over time, they would be about the same. However, if you got it more than the AI you could conclude one explanation could be the AI had an advantage of knowing the terrain. Technically though, the AI should get it a bit more than you since you pick in response to it and thus have the more informed position. It pick blind to what you can do, and you pick knowing what it has done.

All in all it would be an interesting, though tedious, experiment.

AJ

You don't know how many times I've wished I could harvest the data from my battles in the app so I could look for patterns better and do experiments (data geek here). You're right of course - humans are hardwired to be loss averse so we notice and care about losses more than wins I'm fascinated by, and keenly aware of, companies use of AI to manipulate consumers and I'm guessing Inno is no different. I'm not saying they're some evil corporation out to fleece players, simply that their goal is a financial one and smart business strategies require intelligence. Some games are just straight up engineered to get you to spend money. Games like Bubblewitch where the coloured bubble you need is deliberately withheld once you're very close to passing a level. I record gameplay on that to track the AI's strategy and it was shocking. Spend money once and the AI remembers what your break point is and continues to push you to it. Inno uses a lot of common psychological strategies but I haven't found it to be excessive or at the cost of the gaming experience (well, except the golden keys thing which was quite heavily manipulated).

I like skill based gaming where advancement is earned, not bought. I love strategy and problem solving which is why I like Elvenar - it brings together creativity and strategising quite well. Given this it makes me happy to know there's that element of fairness where the AI gets an unexpected wallop just as much as I do.

Technically though, the AI should get it a bit more than you since you pick in response to it and thus have the more informed position. It pick blind to what you can do, and you pick knowing what it has done.

Unless the AI knows the terrain in advance which would give it that edge when selecting troops. I'm sure those who play manually a lot will have figured out whether this is the case or not. Either way, Samidodamage gave me hope when she pointed out there are things she can do that the AI can't. I love the thought of outsmarting the AI

I might run the game through bluestacks so I can record gameplay. At least then I'd have footage I could review rather that relying on my less than reliable memory or skewed perspective lol

 

[DeletedUser27062]

Here's my 'wall of text' for the day, lol!

I think you're right. After all, on browser you can start a fight on manual, switch to auto once on the battlefield, watch the fight and see how the AI handles things. That was the single biggest help to me in learning combat.
I also think AJ is right, the AI makes equally bone-headed moves for both sides, lol! That's why manually fighting battles decreases losses; I can make moves the AI won't.
For me, viewing the battlefield is about more than just terrain. It also helps me with troop placement, even in the absence of obstacles. I do better when I can see the battlefield at figuring out the best way to line up my troops. It's really time consuming, though, so I generally just place them opposite the unit I'm putting them in to handle. That isn't as much help with auto, though.

The AI does not target units based on which other units they are best against. The AI targets the opposing unit within range that has the highest initiative. It seems to also take the movement and strike range of the opposing unit into consideration, so for example the long strike-range cannoneer may target a long strike-range Blossom Mage instead of an Archer even though the mage has a slightly lower initiative than the archer. All of these unit characteristics are assigned number values, so I think the AI is a simple numbers game even if I have no idea what formulas are used.
Here's an example of how that might work:
I put in a Mage to handle a heavy melee (ex: enemy Orc General); if there's an enemy light melee (ex: dog), the AI will first send my Mage to strike at the dog that is specialized against Mages (because the dog has higher initiative and longer movement range than the heavy melee), but the Mage will get mauled when the dog gets its turn. Surviving to strike the Orc General is unlikely, lol!
Also, the AI seems to always advance on the opposing units even when waiting out a turn or moving away from an opposing unit and/or behind a barrier would be the better strategy. The AI will also sometimes ignore the fact that a unit could have been moved to a space where they could strike the desired unit without moving into range of a different opposing unit that should be avoided and move the unit into range anyway. Again, you can take advantage of these bone-headed moves when fighting manually.

The guides that talk about this (player-generated; Inno has never explained any of this stuff afaik) remain valid for map province battles. Recent changes to the tourney structure means this no longer holds true for tourney provinces. Tourney provinces now always have 5 enemy slots. Certain enemy unit types have a higher probability of appearing in a specific tourney province, but it is now possible to face all 5 unit types in one battle in the tourneys. The units that are more likely to appear in each tourney province type are:
Marble: Heavy Melee
Steel: Mage
Planks: Light Range
Crystal: Mage & Heavy Ranged
Scroll: Light Ranged & Heavy Ranged
Silk: Light Melee & Heavy Melee
Elixir: Light Range, Light Melee & Heavy Melee
Magic Dust: Light Melee, Mage & Heavy Ranged
Gems: Light Melee, Heavy Ranged & Heavy Melee

This is extremely helpful! Thank you.