Experience factor - far too powerful?

[Emx77]

In Sulla vs Marius scenario I'm regularly seeing battles where Optimates are victorious against all odds. I have played almost all previous games on AGE engine and I have "feeling" about approximate force ratios which is needed to win a battle. To my surprise what I have learned before doesn't apply here. Too often I lost battles where I was certain in victory. Of course, I'm aware that whole array of different factors are affecting battle outcome but still I lost battles where I had most of these factors on my side. I started to wonder what counts most in a AJE battle? How to explain results where defending army with larger power (PWR) regulary lost from a force which has smaller PWR?

After some of testing it seems that unit experience is far most important factor. I would dare to say it is too much powerful and needs to be nerfed down. To prove what I'm talking about I made a simple test. I throwed two different legions in battle and recorded battle outcomes. Here are the results:

TEST 1: POP vs OPT legions in meeting engagement

Figure 1: Basic test setting



Basic setting for this test consist of two legions from opposite factions marching same distance (2 days) into same province (clear terrain, fair weather) with attack posture (see figure 1). They have no commanders (15% command penalty) and they have exactly the same organizational structure with same number of men and horses. Populares (POP) have just a one point advantage in NM over Optimates (OPT): 90 vs 89.

Figure 2: Stat difference between two legions



Difference comes primarily from origin (ITA for POP faction) and experience. POP legion has 10 cohorts with one star experience. OPT legion has 7 cohorts with one star experience and 3 cohorts with two star experience. This is reflected in slightly higher discipline, assault and cohesion values for OPT cohort elements (figure 2). Judging by numbers this differences are very small. That's why POP legion has 153 PWR (Combat Efficiency of the Force) while OPT legion has 166 PWR. Again, nothing dramatic.

However, difference in battle between these two units is HUGE.

Figure 3: Performance difference between two legions (meeting engagement)



As you may see from figure 3, only slight stat difference is leading to a major performance difference during battle. And please note, OPT legion in this test is relatively unexperienced compared to other Sulla's legions.


TEST 2: POP legion is defending while OPT legion is attacking

I wondered what would happen if I put POP legion on defense? Result of this test is presented at figure 4:

Figure 4: Performance difference between two legions (POP legion defending)



It's clear that defending posture is not enough to compensate disadvantages. Optimates lost every single battle and retreated.


TEST 3: POP legion has one-star commander (3-2-2) and it's on defense while OPT legion is attacking

Finally, in order to boost POP legion even more I appointed (3-2-2) commander which eliminated 15% command penalty and raised legion PWR to 233 (compared to OPT 166 PWR). Now, everything is in favor of POP legion. It has more Combat Efficient Force (PWR), it has commander and it is defending. Where would you bet your money this time? On POP legion? Wrong!!!

Figure 5: Performance difference between two legions (POP legion defending, has commander and more PWR)



Although losses are more balanced now, Populares again lost every battle and retreated (expect one occasion where OPT AI decided to retreat before battle and battle didn't occur).
---

From above tests it's clear that we have two potential problems:

1.) Experience seems to be too powerful. As experience influence discipline, assault and cohesion, even slight difference leads to a huge casualties difference in a battle. I'm not sure how much is this realistic?
2.) Game have problem informing a player about this effects. According to manual, PWR is numerical representation of the relative power of the Force. These tests shows why it is misleading element in decision making process. That's why you will continue to see posts where players are asking about weird battle results.

Any other thoughts?

[yellow ribbon]

i think "PWR" is a varibale not used from the battle engine. it represents visuable efficiency, but this expresses only th stats of the leader applied to a attached unit/experience, not more.
i dont think there is any "PWR times xyz dice roll" improving rate of fire or assault value etc, but how long they make their stand and how effective they are during that time

10 cohesion points can have already a huge impact (also on PWR). you can figure that if you see the huge impact of legions off.(def. fire in round of a battle) remember the NM has 5% steps, meaning a victorious empire might have 4 cohesion points more than other legions etc. even this can be decisive if the discipline is just 2 levels lower.
to compensate this, you would need 30 points higher national moral. ( if 100 NM, you should have a basis of 85 cohesion for a normal unit)

discipline is a cornerstone, 15% higher bonuses from the traits can be decisive already, if real Romans meet Romans

of course your prior experience cannot hold in this game. few units have any fire ability, thus the loss of cohesion, before assaulting phase is the most important.
if one want to compare it, than with highly entrenched troops in a fort, that you try to attack in Pride of Nations early stages.


since experience is lost when the legion turns into an experienced model, the stats changed therefor only a few notches, and experience even is lost if elements are TRAINED, the stats are more important to explain the differences.

i cannot say more that this point had to be discussed and communicated to us a betas either, at least to me. i know only early DB for the units, but for my testing, cohesion loss in the first round and moral check/rooting of while elements is a screw...

another point why your hypothesis is at least incomplete, is the fact that even simple Garrisons with few legion elements might kill 2000.even 4000 barbarians WITH experience if the Roman garr. is entrenched

[yellow ribbon]

the message if a fight is won/lost on the header of the battle report, is announced to be corrected in the next patch, you should judge and test then

[Ebbingford]

Discipline is key.......

[Emx77]

@ Ebbingford. If discipline is a key, do you find it is acceptable that 1 point of discipline increase converts Roman legionnaire into unstoppable Jedi warrior?

@yellow ribbon, I will start from the end of your post. Your example of Roman Garrisons killing experienced Barbarians is not right example. You are simply comparing apples and oranges. If we want to draw reasonable conclusions we should at least compare units of the same type. Regarding this, I noticed that more experienced Roman legions have better stats. This imply that higher stats are influenced by experience (see figure below).



Second, we should forget about traits. In my test Populares commander doesn't have any traits. Optimates legion doesn't have leader at all. Also, if you want to put NM into pot, please note that in above test Populares have 1 NM over Optimates (see figure 2 in opening post). Again, two opposite legions have same numbers of men, horses and same OOB.

So, after we control all different factors there are primarily differences in discipline, assault and cohesion. On paper these differences are not big:

OPT vs POP
Discipline: 10.3 vs 9
Assault: 11.3 vs 10
Cohesion: 90 vs 80

From such stats (everything else being equal) who would expect that first unit, on average, will inflict almost 90% higher casualties to opponent in meeting engagement? Even when the second unit is on defense it will suffer on average 64% more casualties. As I already said, this is wrong on at least two levels:

1) It's weired that a minor difference in unit stats (15%) causes such a huge difference in casualties (90%). This is becoming even more obvious as gap in stat increases. That's why you have death-star-like Sulla's legions in first scenario. It doesn't make sense that defending unit under commander with defense rating of 2 (according to Narwhal that is 20% defense bonus) is not capable to defeat almost same leaderless attacking unit (with 15% attacking penalty).

2) If we want to keep such a model (for whatever reason: balance, history...) it would be good at least to inform player about this. It will save players from a lot of frustration. That's why it would be advisable to include such factors in PWR approximations (if they are not already included). I know, prior gaming experience cannot hold in this game, but having 1000 PWR stack regularly beaten by 300 PWR stack is nonsense which I didn't experience in other Ageod games. Remember, Combat Efficiency of the Force (PWR) is a numerical representation of the relative power of the Force. What is the point of PWR indicator if it is totally misleading?

[yellow ribbon]

@yellow ribbon, I will start from the end of your post. Your example of Roman Garrisons killing experienced Barbarians is not right example. You are simply comparing apples and oranges.



no, i dont. i play it for th 5. month. Experience cannot compensate the higher stats of troops.

actually the higher experience level by set up, not the few you can gain with most troops during a medium sized campaign, even has no impact on the overall battle results, nor cohesion etc.. i.e. you find highly experienced troop just operating with 93 cohesion, thus not gaining a bonus from it as one would expect it like from the manual

the most important difference comes from basis-model and experienced-model, with different stats

its discipline and loss of cohesion the first round, not experience.
if you loose a lot of national moral and thereby cohesion, the losses are not progressively higher. thus the stats have the higher impact, thus most probably discipline (working together, as mentioned, with the high loss of cohesion in round one of battles, especially for non-romans)


also you imply a "steady state analysis" which is simply not given. there are dozens of variables based on the stats multiplied by the number of elements multiplied with all the ´lil pictures in the battles screen, dice rolls.
the way you want to build the hypothesis around a inappropriate core...

[yellow ribbon]

Second, we should forget about traits. In my test Populares commander doesn't have any traits. Optimates legion doesn't have leader at all.

wrong, in a larger battle the deviation coming from tests with traits from commanders and with out, by otherwise equal stats/bonus of commanders, has even impacts on a scale of nearly 7000 men in the battle result, nearly 1/5 lower losses if the commanders is offering discipline bonus, as an example, not more nor less

y´our major problem seems to be, the incapability to see that this scales you compare ARE NO NOMINAL SCALES. you have even to stop to count weighted average levels, each stat is tested for each element.
that, compared with frontage what is also missing in any of your tables, as far i see, make stats like discipline even more important.

than you need to list every battle result, not only men lost.

elements on the run, elements destroyed etc. the engine doesnt know mixed-up values. as one can see i.e. from the famous POWs higher than losses discussion.

[Cfant]

Thanks. You tested, what I just argumentad by feeling in some threads. I absolutly agree - it IS a problem. It is even more dramatic, if you transfer your results to the scenario itself. The Optimates incredible strength leads to easy victories - and therefore to a rising moral (the second very important aspect imho).

My "test" was simple and not very technically: Playing POP, using "best" leaders, huge numbers, defensive posture - loosing again and again. Playing OPT, marching trough the campaign without loosing even one battle. Before playing Sulla, I believed moral to be the key. But as POP I still lost, even with 20-25 higher moral. Now I think: starting with experienced legions is massivly overpowered, because the single battle results you showed will multiply with the rising moral.

The problem I see is not only, that many players will be frustrated, but simply that the scenario (as the Sertorius scenario) is quite unbalanced. Even more: In a strategy game, the player should use many factors to achieve victory. Here, moral and experience seem to be much more important than other single factors (terrain, general, posture, mix of troops), resulting in a steamroller Sulla (Sertorius). I also see the problem, that the game does not very good explain what happens in the battle. (Maybe one of the veterans could make a beginners-section like Narwahl how to read the battle-report?)
In an other thread I was told, that battle results are still in work for a patch. So maybe Sertorius and Sulla can be beaten after the patch

@yellow ribbon: That's all a talk about the technic of the engine. But what about the scenarios? Would you call Sertorius and Marius vs. Sulla a balanced scenario? Would it be fun in a Pbem between players of similar knowledge of the game?

[yellow ribbon]

I know, prior gaming experience cannot hold in this game, but having 1000 PWR stack regularly beaten by 300 PWR stack is nonsense which I didn't experience in other Ageod games.

here comes another problem, as said, PWR is not any fixed variable, you CANNOT compare PWR your side with PWR enemies side at all, nor does a higher PWR level that the stack is better / on the winning side.
just drop this nominal equations.

its in games to make the player able to judge what possible size/impact of troops he is facing. the fact that you have far more barbarians in game, with PWR 50/60 as units, but barely able to fight an Auxiliar with PWR 30/40 should jump to ones eye.

if it would be PWR == PWR or even close to it, Romans would never been able to win such an empire. Most troops did run and were not killed. so it is in game.
Especially as frontage makes it impossible to play 40.000 mowing down 9000

[yellow ribbon]

@yellow ribbon: That's all a talk about the technic of the engine. But what about the scenarios? Would you call Sertorius and Marius vs. Sulla a balanced scenario? Would it be fun in a Pbem between players of similar knowledge of the game?

1.) its a military game, its a strategy game...

2.) With experience in playing AIE

- i can win with Pompey, many claim it impossible
- I can force Opt to make peace with me as PONTIC in MS87, as well as i can drive the AI out of Spain, as well as i can win a moral based suddendeath victory as Populares over the Optimates
- For Sertorius i cannot say much, but that in prior versions it was hard to win. As some points like frontage were changed, it became just a question of destroying Romans piecemeal
my thought is that his legions in the Ser. scenario became to powerful thereby, which wasnt given before.


Depending who is actually playing which side in which forum and how intensive he is playing the game, the complains are quite different, the range is enormous.

[lukasberger]

Depending who is actually playing which side in which forum and how intensive he is playing the game, the complains are quite different, the range is enormous.

Which I'd venture to say means that the engine is actually working well. After all that's surely one of the goals, to make the game challenging to players who are playing at many different levels.

[Pocus]

Emx77 send me your test bed files to pocus@ageod.net please!

Very interesting research, you get a numbercruncher cookie for that

[Emx77]

no, i dont. i play it for th 5. month. Experience cannot compensate the higher stats of troops.

It all depends on starting stats of a unit. Of course, if you have barbarian unit with low starting stats it will be very hard to pump it up through experience during gameplay. That's why we need to use units of the same type for such a tests. In such a way we are keeping many other factors under control. Otherwise we are talking about apples and oranges. Also, If we are going to have serious discussion about this I would like you to give us numbers rather then anecdotal evidence.

the most important difference comes from basis-model and experienced-model, with different stats

I agree. In above test we have same basis-model so all difference on a battlefield comes from experience (or to be more precise: from experience influenced stats).

also you imply a "steady state analysis" which is simply not given. there are dozens of variables based on the stats multiplied by the number of elements multiplied with all the ´lil pictures in the battles screen, dice rolls.
the way you want to build the hypothesis around a inappropriate core...

Yes, I know. There is a lot of variables in a calculation. But, by having two opposite units with identical organic structure - fighting the same battle over and over again, in the same terrain and weather conditions - we are keeping a lot of these other variables under control. I repeated each test five times and result is the same. Sorry, you can't blame it on a dice rolls. You miss my main point: slight difference in discipline leads to a huge difference on a battlefield. That's produce a lot of counter intuitive results.

in a larger battle the deviation coming from tests with traits from commanders and with out, by otherwise equal stats/bonus of commanders, has even impacts on a scale of nearly 7000 men in the battle result, nearly 1/5 lower losses if the commanders is offering discipline bonus, as an example, not more nor less

Clearly "leader-trait factor" is not causing what we are seeing in above tests. In first two tests there are no leaders and in last test leader is on a losing side.

y´our major problem seems to be, the incapability to see that this scales you compare ARE NO NOMINAL SCALES. you have even to stop to count weighted average levels, each stat is tested for each element.
that, compared with frontage what is also missing in any of your tables, as far i see, make stats like discipline even more important.

Frontage, why you want to put frontage into this? In above tests we have clear terrain, fair weather and ONLY one unit per side. Frontage is not a issue here.

I'm afraid your major problem is you can't see the forest for the trees.

PWR is in games to make the player able to judge what possible size/impact of troops he is facing. the fact that you have far more barbarians in game, with PWR 50/60 as units, but barely able to fight an Auxiliar with PWR 30/40 should jump to ones eye

According to manual PWR is a numerical representation of the relative power of the Force. It doesn't represent only the size of the force but also quality. You may see this from above example. Two legions identical by numbers and structure have different PWR value because slight differences in stats (166 vs 153). When you add a 3-2-2 leader to a relatively weaker legion, it's PWR jumps from 153 to 230. It was always like this in previous AGEOD games. I've never have an issue with lower PWR force beating a larger PWR force. Variation and uncertainty is welcomed and you were able to find reasonable explanation for defeat (terrain, weather, posture...). But here we have a lack of uncertainty as force with lower PWR almost always beat a larger PWR regardless of anything else.

[Emx77]

Another test. Everything is same as in TEST 2 (see first post) expect that Rome fell to Populares a turn before. As a result we now have a really huge national morale (NM) gap between OPT and POP (49 points difference!). This lowers OPT legion cohesion which is now 76 (compared to 93 for POP faction legion).

TEST 4: POP legion is defending while OPT legion is attacking (NM gap)



Although a losses are more similar now, even this is not enough to compensate small differences in discipline and assault values (Populares lost every battle).

[Emx77]

Emx77 send me your test bed files to pocus@ageod.net please!

Very interesting research, you get a numbercruncher cookie for that

Mail sent.

[Narwhal]

I believe this is a very compelling and conclusive test, and you might have found out one of the main reason for unbalanced combats.

I believe another reason is terrain bonus / malus not being adapted to the focus on melee combat, but I will post something about this on the beta forum ASAP . In any case, between legion and on clear terrain this second reason has no impact.

Two small notes, before a longer answer :

- Italian legions are "better" than non-Italian legion (bonus in base cohesion for instance). I believe you mixed them in your second test.

- There is a "bug" with the non-linearity of the NM impact on cohesion. I mentioned it on the beta forum. So 1 difference in NM can have a significant impact on max cohesion.

[Emx77]

- Italian legions are "better" than non-Italian legion (bonus in base cohesion for instance). I believe you mixed them in your second test.

I have used Legio II for Populares faction in all tests. It has IT designation in upper right corner (figure 1). In first three tests each cohort of this legion has 80/80 cohesion. After Populares took Rome it cohesion changed to 93/93.

[yellow ribbon]

I'm afraid your major problem is you can't see the forest for the trees.

told you something about that in a PM, exchange of minds doesnt appear for normative bias. the impact is there for sure. from my data from End May till today, i can simply prove you wrong while calculating with far more than experience.

You dont like it, not my problem

[Kensai]

What if discipline and experience are INDEED the main battle winners in AJE? Reading about Sulla's actions in the Mithridatic war and his march to Rome, he had indeed some formidable and battle-experienced troops, veterans of the previous years' Social War.

It doesn't make sense that defending unit under commander with defense rating of 2 (according to Narwhal that is 20% defense bonus) is not capable to defeat almost same leaderless attacking unit (with 15% attacking penalty).

Emx77,

Unless there is something changed in the engine 2 points of Defensive ability add up 5% x2 so the total is 10%, not 20%. Regarding the leaderless attack penalty of 15%, are we sure it applies in fighting defensively as well as offensively? By mistake or bug it might be like a disactivated General, meaning applied only to movement (as per PON tooltip description).

All things considered, it might as well be a problem with overpowered experience as clashes in this ancient war title happen at close range (or better: assault range!) where discipline gives serious imbalances. Or the losing side always gets its troops hunted down and slaughtered. Indeed, a 15% difference in stats should NOT give consistently 90% more casualties for the losing side. At least not always.

[yellow ribbon]

[quote="Kensai"]


wrote you something via PM about this topic

[Kensai]

Thank you for the technical reply. There is only one thing I don't understand... He DID try to test similar units (legions) over neutral open terrain giving all givable bonuses to the weak side (defensive, with a leader, with much NM, etc)... and still the experienced Optimates won.

Now, personally I don't care about where this experience comes from (setup or later campaign victories of the scenario) as long as (1) it is historically accurate meaning that indeed Sulla had the most experienced legions. Sources seem to show this was the case. But most importantly, I would love (2) the difference in experience have more balanced results... so you either give MORE experience to Sulla's legions for the same results: 90% casualties. Or you lower the damage done to match the smallest experience difference.

I am not sure which is the best course of action, it all depends on historical accuracy and better in game balancing. I fully understand that perhaps the engine cannot perfectly simulate ancient warfare at the moment given that most battles happen at close/assault range anyway. I am pretty sure that it will be fixed soon, though.

Just do not over-correct it, please, I am soon starting a PBEM as Sulla!

[yellow ribbon]

if you loose the TQ (discipline test) assaults/firing are stopped. now imagine what happens if the other side has full high cohesion and some iron glove.

Dont have time to put it in a fine excel, but Exp. come from the setup mainly, represents the historical difference.
Everyone cries out loud. 80% chance developers will make it more gamy

situation, based from the last beta version i tested few days before release:

values for

[color="#FFFF00"]Sulla best legion / fresh Populares legion / Sert. best legion / Caesars best legion[/color]

Experience: 5 star / fresh / 3 star / 5 star
Cohesion 90 / 80 / 85 / 90
NM (bonus/malus) 90 / 88 / 105 / 100

off. fire 5 / 5 / 5 / 5 [color="#FF0000"] NO CHANGE[/color]
def. fire 5 / 5 / 5 / 5 [color="#FF0000"] NO CHANGE[/color]


[color="#FF0000"]Initiative 8 / 6 / 7 / 8 slight changes from huge experience difference, but could also be legion model based differences[/color]

range 1 / 1 / 1 / 1

rate of fire 2 / 1 / 1 / 2

protection 2 / 2 / 2 / 2

[color="#FF0000"]discipline 12 / 7 / 10 / 12[/color] huge differences, but 4 point differences may come from models based difference

Assault 15 / 9 / 12 / 15 can be compensated if more troops are in the stack IF TQ test did not fail and frontage allows it, at least three point difference can come from models based differences

ranged damage 1/5 // 1/5 // 1/5 // 1/5

assault damage 6/35 // 5/30 // 5/30 // 6/35

[yellow ribbon]

Now remember what i stated, if you add one single commander and increase discipline with it for the stack of, i recall 15%, the LOSSES for a military weak opponent like Pompey can already be reduced by 20% in a clinical test environment,

- 100 turns
- always using the same forcepool
- always playing the same strategy

Again, as you see above the impact of Exp. is there, the magnitude is low (unless all the tooltips and elements panel doesnt work) and it increases just problems which are in game BEFORE experience does even count.
If i loose a lot of cohesion and then have a quite higher chance to run away from the battle, well...

but remember, i used the BEST LEGIONS on the map, for the comparing table above, most are much weaker and the NM bonus/malus is up to you, events, events, and another event, like taking Rome etc

Sulla shouldnt have 5 star legions (Socii and Pontics are weak enough), Sertorius should have more penalties from weather and frontage in Spain. he can do the same as a good player as Caesar, he can even steamroll in the Extremadura and the nasty, dry hills around...
For Caesar, well more then 10 year training and a much smaller force for the first year...

[Narwhal]

No time to answer in detail (last day at work), but there might be an issue experienced units are different from "fresh unit", not only in the number of stars, but also in the model. This together can lead to an overwhelming advantage (better "model" and then experience bonus). This is just an hypothesis and I need to check this, but as I have no AJE computer handy right now...

[Kensai]

Very enlightening, yellow ribbon. Now if what you state is accurate then a simple (?) engine update (and obvious balancing tweaks in every scenario) will put the AGE engine in perfect condition to simulate and abstract ancient warfare clashes that happen mostly at close range. I say simple fix because I presume even a straightforward change to the TQ roll (making it succeed in +50% of the cases) will mean that most poor units will have a chance to fight at close range rather than freezing while taking hits.

That said, please everyone read Sulla's Social War campaign. His legions had indeed gained huge experience, thus Sulla starting our scenario as a darn tank might as well be historically accurate.

[yellow ribbon]

No time to answer in detail (last day at work), but there might be an issue experienced units are different from "fresh unit", not only in the number of stars, but also in the number of stars. This together can lead to an overwhelming advantage (better "model" and then experience bonus). This is just an hypothesis and I need to check this, but as I have no AJE computer handy right now...

i have the DB 21a, i know most of the differences, thats why i simply compared what you have in turn ONE and what your opponent, here the most complained Populares, can build to substitute or replace troops/to increase the army.

As said, its from the last long-range-beta test i did the weekend before preoder-version went online and there might be changes. if so, tell me, i correct it.

based on my observation during testing i assume:

Initiative of republican legions may differ for models WITHOUT impact from experience: 5-7

Discipline of republican legions may differ for models WITHOUT impact from experience: 7-10

Assault of republican legions may differ for models WITHOU impact from experience: 8-10

cohesion in the setup files do differ from 75 to 90 plus/minus change from NM (10 NM == 5% change of cohesion) WITHOUT any experience involved

[Emx77]

What if discipline and experience are INDEED the main battle winners in AJE? Reading about Sulla's actions in the Mithridatic war and his march to Rome, he had indeed some formidable and battle-experienced troops, veterans of the previous years' Social War.

I agree. I have nothing against discipline and experience to be main battle winners in AJE. My complain is more about impact which 1 point difference in discipline has on battle outcome. Tests show that this slight difference is hard to compensate with anything else. Leaders (maybe most crucial factor on battlefield in ancient times), posture, cohesion, NM... seem to have much lesser impact. Simply, discipline effect is overrated.

Unless there is something changed in the engine 2 points of Defensive ability add up 5% x2 so the total is 10%, not 20%.

Narwhal in his AAR wrote ..."circa 10% by “point”, compared to 5% for other games, so good leaders can be decisive". I don't know how he came to this but I'm sure he will explain it.

Regarding the leaderless attack penalty of 15%, are we sure it applies in fighting defensively as well as offensively? By mistake or bug it might be like a disactivated General, meaning applied only to movement (as per PON tooltip description).

I really can't tell. But it has more sense this penalty to negatively impact attacking rather then defending side. In tests attacking side always win.

All things considered, it might as well be a problem with overpowered experience as clashes in this ancient war title happen at close range (or better: assault range!) where discipline gives serious imbalances. Or the losing side always gets its troops hunted down and slaughtered. Indeed, a 15% difference in stats should NOT give consistently 90% more casualties for the losing side. At least not always.

That's what I'm talking about. My test is simple battle where two almost identical legions fight each other. You may assume what effect will be in larger battles where you have bigger differences.

Now, personally I don't care about where this experience comes from (setup or later campaign victories of the scenario) as long as (1) it is historically accurate meaning that indeed Sulla had the most experienced legions. Sources seem to show this was the case. But most importantly, I would love (2) the difference in experience have more balanced results... so you either give MORE experience to Sulla's legions for the same results: 90% casualties. Or you lower the damage done to match the smallest experience difference.

Bravo, you nailed it on the head.

[Ace]

I have a feeling, the designers were looking to model ancient warfare in which it could be possible for a single disciplined roman legion to destroy a force of 100.000 barbarians (Boudica's army for example). When you compare Roman army to Celtic armies of the time, discipline was a key difference. I dont think they expected, it would give so much imbalance in Roman vs Roman clashes.

I agree there should be great difference in 1 star vs no star experience. This was close hand fighting, and troops which never saw action were likely to break. That is why Romans historically developed three battle lines, with green troops fighting in first line, and veteran troops behind them. This way, they had nowhere to run.

[Pocus]

Tests indicate this is a combination of factors, and there is no 'magical number' in particular, that will transform an average unit into an über one, just by upping a variable with a +1.

A potential culprit could have been the first shock modifier, a multiplier to damage that is a novelty of AJE, to account for initial 'barbarians' charges, in the case of Roman, this is not the main factor, as this value is 110, so a mere +10% bonus (still, it adds up a bit to the injury).

Another culprit could have been the cbtTQModifierForAssault variable, which is an additional penalty to whoever has the less discipline in battle. Here the value is 100%, so that's not that.

So we are back to simple maths. That advantages add up geometrically and not arithmetically. i.e that (and this is what happen in real life battle) a sum of advantages produce very very big effects. Now I'm not saying that the axe did not swung too far, don't mistake me. There is room for changes, probably, but it must be done with care and details, or the axe will swing too much in the opposite direction.

You can learn all that by inspecting the detailed combat log by the way.

So here it goes. The ranged damages in these battles are not important, they account for at most 5% of the total damages done (except in very specific situations like tons of archers etc.). Consider now the basic formula for assault: your chance to assault or counter-assault (because the defender can reply) is Assault Value X Discipline* X correcting factor (set to 0.5 in AJE).

*: Discipline is often abreviated TQ, as in Troop Quality.

Basically this gives a 5-star legion (model 187 OPT) a 82% probability to assault and a 0-star legion (model 186, POP) 45% chance, because

OPT: Assault 15, TQ 11, x0.5 = 82%
POP: Assault 10, TQ 9, x0.5 = 45%

If you check others combination, with a few stars, then you get a nice spread between these 2 extremes.

In addition to that, before any assault, you must pass a TQ check under a 10-sided dice. If you fail, you don't assault or counter-assault.

So in extreme cases, you get almost a 2 to 1 probability in favor of the best troop. Now consider the damages done, they are very high. A cohort has 10 hits, and the damages done are 6 hits. If a cohort is unlucky enough to get hit twice, it is gone, if not, it is still severely weakened, enough sometime that an hasty retreat will kill it anyway.

On the contrary, defensive terrains may be not enough defensive, for example Hill only give a +1 TQ bonus. Perhaps this is not enough.

As for command penalty, 15% is not a 15% reduction in efficiency. This is the chance that you lose one rate of fire (that you don't care about) and have half your initiative (this is more problematic!). Well this is problematic if your enemy manage to hit you, because otherwise you'll hit hard second, but hit hard nonetheless. So command penalty is less a problem against average or poor troops anyway, they have low hit chance, you are strong, so you retaliate anyway. Now the reverse is also an interesting thing to consider: if you are strong and have initiative, then you will trash an enemy element. Its cohesion will be so lowered that it will get an additional penalty because of that. Its chances were already rather low to hit, now they are even worse than that.

As said, combination of advantages can really make a small elite force beats a large but average one.

[yellow ribbon]

OPT: Assault 15, TQ 11, x0.5 = 82%
POP: Assault 10, TQ 9, x0.5 = 45%

If you check others combination, with a few stars, then you get a nice spread between these 2 extremes.

In addition to that, before any assault, you must pass a TQ check under a 10-sided dice. If you fail, you don't assault or counter-assault.

So in extreme cases, you get almost a 2 to 1 probability in favor of the best troop. Now consider the damages done, they are very high. A cohort has 10 hits, and the damages done are 6 hits. If a cohort is unlucky enough to get hit twice, it is gone, if not, it is still severely weakened, enough sometime that an hasty retreat will kill it anyway.

On the contrary, defensive terrains may be not enough defensive, for example Hill only give a +1 TQ bonus. Perhaps this is not enough.

a big thank you for the "uncoding" of your engine
as far i get your equation, than its close enough to what my guts told me, the magnifying.

Assault value alone could be compensated (frontage, terrain, larger numbers when combats take a couple of rounds...)
but the combination of a foe with low cohesion and the high test probability from TQ/discipline is getting dis-proportionally larger and larger

i think i know that i will be killed for modding during the winter, choked from the angry woman of the house...