PDA

View Full Version : Investigation of Experience Points (chevrons)



therother
11-12-2004, 01:06
Placeholder

zhuge
12-11-2004, 07:37
Did some XP tests today (yes, I know I've got too much time on my hands). ~D

My Greek side had the following units: 1 General, 1 Mil.Cav, 1 Peltast, 2 Armored Hoplites (1 with 1XP and 1 with 2XP), 2 Archers
The enemy rebels had 1 General, 2 Velites, 2 Peasants
The odds were 20:9 in my favor. I thought it would be a nice battle to start tests since archers can easily avoid casualties.

Below is the raw data for kills for the same battle above fought 5 times:
All kills are made on non-routing units.



Archer1
91 Peasants, 28 Velites XP1
69 Peasants XP1
59 Peasants XP0
65 Peasants XP0
3 Peasants, 66 Velites XP1

Archer2
25 Peasants, 49 Velites XP1
51 Velites XP0
63 Velites XP0
6 Peasants, 68 Velites XP1
68 Peasants XP0


Conclusion: It takes exactly 69 kills to level up an archer unit from XP0 to XP1. Quality of kills do not seem to matter.


I can upload the saved file if anyone is interested in helping to test. If there are enough interested parties, we might want to split this post off and make a separate thread titled say "XP and kills".

EDIT: It occurs to me that 69 is a rather odd number and I'm not sure if there are other factors modifying this so I will qualify the above results as being played on VH/VH and on normal unit size (40 men for an archer unit).

zhuge
12-11-2004, 15:26
A few more observations:
1)The level up kill requirement may not be linear.
If it was linear, 138 kills would get you 2XP. Tested it and it only gave 1XP. In fact, I made 173 kills later on with 145 kills on non-routers (102 Peasants and 43 Velites) and 28 kills on routers (7 Peasants and 21 Velites) and still received only 1XP.

2)Killing routers appears to net less value in terms of getting the next XP level
Killed off 46 non-routers (17 Peasants, 20 Velites and 9 General's Bodyguard (rebel)) and then an additional 25 routers (15 Peasants and 10 Velites) for a total of 71. Would have gotten the next XP level if routers were counted towards a total of 69 just like non-routers but apparently it isn't so because I received no XP. However killing routers also cannot be of no value, because killing 67 non-routers and 10 routers does give 1XP, which mean that those 10 routers are at least equivalent to 2 non-routers (since we've established earlier that we need 69 non-router kills for XP1).

Would be nice if someone could help test out archers on Normal campaign and battle difficulty. Just want to rule out some possible complicating factors.

EDIT: I vaguely recall that in Shogun, you got 1XP if every man in a unit made 2 kills (but I think that was only for the first level) and some players mentioned that you seemed to get less XP for killing routers as well. If someone has done more extensive analysis in Shogun/Medieval it would be nice to paste the findings here as I suspect the pattern may be grossly similar.

zhuge
12-12-2004, 06:07
I'm getting awfully confused. ~:confused:
There may be some randomness involved in deciding whether a unit gets the next XP level. I've tried to tie in all factors I can think of and standardize them: involvement in a previous battle, number of other friendly units in the battle, whether the enemy general was killed by the unit under study (KillGen).

Open a new game, choose Julii, VH/VH, regular unit size and you'll have my current setup. Use Flavius to hit Segesta on turn 1 (he has 1 Archer unit, 2 Hastatis and 1 Triarii). Brought them all and fought Segesta (1 Warband and 1 Barb Peasant).



Warband Peasant Total KillGen XP
49 41 90 N 0
54 39 93 N 0
54 41 95 Y 0
56 39 95 N 0
59 36 95 N 0
59 41 100 N 0 ??

57 41 98 N 1 ??
56 47 103 N 1
58 45 103 N 1
58 46 104 Y 1
58 48 106 Y 1
60 46 106 Y 1


As you can see there is a seemingly unexplainable difference for the 2 datasets which have ?? at the end.


I then tried taking 2 units only (Flavius and Archer) and refought the battles:



Warband Peasant Total KillGen XP
49 34 83 N 0
57 36 93 N 0
57 38 95 N 0
56 44 100 N 1


Having 100 kills didn't seem to get 1XP if you brought 5 units but it does when you bring 2??
It occurs to me that I might have recorded the result wrongly for a dataset or two (though I sincerely believe I haven't). There is no question about routing as all kills are made in the town square which gives infinite morale.

Would like to have some help here if anyone's interested. If there is indeed some random factor then there will be little point in continuing to probe in this area. :help:
Comments and criticism are very much welcome.

P/S - Not sure if it matters but more than 69 kills are required here for the next level probably because the Archers under Flavius already start with XP1 and levelling to XP2 may need more kills.

Simetrical
12-13-2004, 00:05
I suspect it's hopeless. There have been ample anecdotal reports of units not participating in the battle and still gaining experience. I should think that there are lots of factors, quite possibly including some random ones. I doubt there's any way to properly isolate them.

-Simetrical

zhuge
12-13-2004, 09:14
I suspect it's hopeless. There have been ample anecdotal reports of units not participating in the battle and still gaining experience. I should think that there are lots of factors, quite possibly including some random ones. I doubt there's any way to properly isolate them.

-Simetrical

Yes, I've read of a few rather weird reports on the official forums recently. Still I haven't experienced having a unit go up by 1XP if it didn't fight or for that matter didn't make any kills in the battle so I'm not willing at this stage to make conclusions on anecdotal evidences.

Rather hard to test for though. We might need a huge amount of data to really arrive at any conclusion. As mentioned before, I don't think I can do this alone anymore. But if other members want to chip in every now and then we might be able to build a small collection of findings and hopefully from then we can piece together something.

There is another way to approach this problem and that is to directly decipher the saved files and get the exact XP for each unit. The XP is brought forward from battle to battle so it must be saved somewhere in the saved file. RomeSage (a modding tool) can currently alter campaign, and battle difficulty, change unit size, amongst several other functions. If we can figure out how and where XP is stored then I believe we'll have made an important step to cracking this puzzle.

HopAlongBunny
12-13-2004, 18:23
Here to confuse the issue.

I have, on more than one occaision, had a unit with no kills gain exp (always to a unit with no chevrons) Last night I had a cataphract archer gain "0" exp after running over 201 routers...I thought they were worth something in bulk ~:cool:

Kinda miss the log files from MTW; was nice to look at every now and again and see what happened.

Tamur
12-13-2004, 20:29
Wasn't there mention of the game tracking the experience level of individual soldiers, so that troop combinations would come out correctly? I suppose its existance doesn't imply that there's a history logged to go along with it.

Simetrical
12-14-2004, 04:01
There is another way to approach this problem and that is to directly decipher the saved files and get the exact XP for each unit. The XP is brought forward from battle to battle so it must be saved somewhere in the saved file. RomeSage (a modding tool) can currently alter campaign, and battle difficulty, change unit size, amongst several other functions. If we can figure out how and where XP is stored then I believe we'll have made an important step to cracking this puzzle.
Definitely. My guess would be that XP is stored as a noninteger value from 0 to 9, or some equivalent. If a soldier* gained, say, .1 XP from simply being in a battle, plus an additional .05 for every enemy killed, that would explain things nicely. We could quantify things very precisely if we had precise values for unit experience, since we could save, check the values, start up a battle under reasonably controlled conditions, and then recheck the values afterwards to find the difference. The other possibility would be a percentage chance of gaining a full experience point for various events, but that would be hell to figure out, so let's all hope that it's the gradient possibility. Now all we need is to hack the save files . . .

*I think XP is stored by soldier, not by unit. For one thing, that would explain units' actually losing experience—if the unit happened to lose its more experienced soldiers, then its XP would go down. The alternative would be the unattractive possibility that units can actually lose XP from routing or something, but that explanation falls short of explaining my second piece of evidence: the merged-XP phenomenon.

For example, take a unit of 40 soldiers with one bronze chevron (Unit 1) and a unit of the same type with only 10 soldiers, but one silver chevron (Unit 2). Drag Unit 1 onto Unit 2 and see what values you get. According to the XP-by-unit theory, the XP should either be averaged and both units should have exactly the same XP, or the units should keep their original XP (so Unit 1 will still have 1 and 2 will still have 4). This isn't the case. The two units often have different XP values, both from each other and from their starting values. This is true, I propose, because soldiers from Unit 1 go over to Unit 2 either on the basis of experience (highest first or lowest first), or on a random basis, and so the averages will end up different.

Furthermore, I believe that soldiers in battle quite possibly have stats according to their own experience, and not according to the group's experience. Many people have mentioned (unscientifically, but coming from multiple independent subjective analyses) that units seem more likely to gain experience when they suffer casualties. I propose that this is because the green soldiers are more likely to die off, thus raising the average XP of the unit.

Finally, it's worth mentioning that official MTW materials apparently said that XP in that game was stored on a per-soldier basis.

-Simetrical

Kraxis
12-14-2004, 12:16
I can't believe that CA would have left the individual tracking.
That means every single soldier gets tracked as per his experience (and other factors of course). Thus in battle a unit will advance in XP when more than half the unit has XP1 and the rest XP0. So obviously it will take a considerable time before the unit reaches XP2 as they only get to XP at 69 kills (which in turn is about half the men). And if CA has kept the double system as previously (for every level up they need to kill double over as many as they did before), then getting to XP2 in a battle is night impossible.

And it was XP1 for one kill in STW and MTW. Then 2 for 2, then 4 for 3, then 8 for 4 and so on. That is why knights could rake up very high XP. But remember that some units only counted half or a quarter of normal ones, such as Peasants.

Leodegar
12-14-2004, 17:44
unfortunetaly i have no awnsers, but just another point to think about... if you retrain units, the whole retrained unit has the same xp-level than the remaining soldiers before retraining (no matter how few they were!). isn't that a point against the assumption that xp is tracked for each individual soldier?

Kraxis
12-15-2004, 11:45
unfortunetaly i have no awnsers, but just another point to think about... if you retrain units, the whole retrained unit has the same xp-level than the remaining soldiers before retraining (no matter how few they were!). isn't that a point against the assumption that xp is tracked for each individual soldier?

No not directly. It just means that the reinforcements are faulty. I mean what are the chances that a green recruit is as experienced as a grizzled veteran?
Ther are many odd instances in RTW that is hard to explain from an older TW point of view. I think that in this case the devs madea small oversight and gave the recruits the unit experience (it too has to be tracked).

zhuge
12-15-2004, 14:09
I can't believe that CA would have left the individual tracking.
That means every single soldier gets tracked as per his experience (and other factors of course). Thus in battle a unit will advance in XP when more than half the unit has XP1 and the rest XP0. So obviously it will take a considerable time before the unit reaches XP2 as they only get to XP at 69 kills (which in turn is about half the men). And if CA has kept the double system as previously (for every level up they need to kill double over as many as they did before), then getting to XP2 in a battle is night impossible.

And it was XP1 for one kill in STW and MTW. Then 2 for 2, then 4 for 3, then 8 for 4 and so on. That is why knights could rake up very high XP. But remember that some units only counted half or a quarter of normal ones, such as Peasants.

Thanks for the info and correction Kraxis. :bow:
I've not been playing S:TW and M:TW for ages and have gotten slipshod over details. :embarassed:
I will try to recontinue testing some time later as I am pretty tied up at the moment. Would very much appreciate it if someone could pick up testing and continue to post data. I was wondering if switching over to arcade mode might enable easier testing as we then get unlimited ammo (which would make it easier to get really high kills for archer units). Of course we need to make sure that leveling/XP follows the same general scheme, when in arcade mode.

therother
12-15-2004, 22:00
It occurs to me that it might be best to do at least some of the testing against a human opponent in multiplayer. That way, you can better control the situation.

Also, instead of using arcade mode, which might introduce unwanted side effects, would it not be better to mod the number of arrows?

See this thread for GodsPetMonkey's excellent editors, which should help if you don't feel comfortable modding the files by hand.

Simetrical
12-16-2004, 03:28
If you want to give your archers unlimited arrows, you can just do a simple toggle. Open up preferences.txt, in your base RTW folder (usually named "Rome - Total War"), look for UNLIMITED_AMMO, and set it to TRUE.

-Simetrical

Epistolary Richard
12-16-2004, 17:35
I have, on more than one occaision, had a unit with no kills gain exp.

I had the same thing happen just now. Medium-sized siege, a few hundred kills, the one unit of mine that did get experience was the unit of barbarian horsemen that didn't move from their original starting location: XP1 0 casualties 0 kills!

mfberg
12-16-2004, 21:37
On a medium level game I have had Hastati gain 1 experience while only breaking a wall down.

mfberg

Maltz
12-16-2004, 22:50
Recently in my Carthagian campaign I purposely build up my starting slinger's EXP. After countless battles now they are experience 8 soldiers. With some blacksmith upgrade - man they shoot bullets. In just one volley they can kill 30-40 militia hoplites at town center.

Besides that here are a few of my observations of this exp. issue:

== Sometimes experience shrink one level after I hit "start the battle", but get back to original level after the battle is over. I heard this is because the way battle handles exp. is a little different, so the number flips & flops. ~:)

== Experiences are stored with invidividual soldier's. There are 2 examples:

1. When I merge a few soldiers of "no weapon upgrade" unit with an almost full "silver weapon" unit, they brcome bronze weapon. If experience is counted as a unit, then everybody should have silver weapon.

2. Sometimes part of my general's bodyguard die, and the entire bodyguard unit actually lost experience. If experience is counted as a unit, then there is no way for them to lose experience from killing enemies.

== Common sense: experiences are harder to get in higher levels.

***

I wonder the information of exp. is actually stored as "kill number" of each soldier. Then, the program will covert this kill number into proper exp. This means that you can be very close to a threshold for the next level, and not worrying about losing all "partial experiences" after this battle is done. This makes achieving very high experience possible. ~D

***

Let's assume experiences do write on individual soldiers' forehead. Then it will explain the weird thing mentioned in Zhuge's experiment. To simplify the case I now assume it is purely 100 "exp1" archers vs. 300 peasants.

Let's also make up a promotion rule:
Exp Total number of kills required
1: 1 (kill)
2: 1+2
3: 1+2+4
4: 1+2+4+8
5: 1+2+4+8+16

Case One:

When arrows run out, every archer gets 2 kill. From the rule we know they already start with experience 1, so 2 kill is just enough for them to gain 1 more experience, from 1 to 2.

Total number killed = 200.

Case Two:

When arrow runs out, 21 archers had a better line of sight and each of them killed 6, and the rest 79 got poor sight and each of them killed 1.

(Total number killed = 205, higher than case one).

That 21 good archers, from the chart, you can see they were great and they just had enough kills to gain... wow 2 experience each, from 1 to 3. But the rest 79 didn't qualify for lv 2 yet.

So, when it comes to calculating the average experience, you get (3*21 + 1*79) /100 = 1.42.

If the program rounds up the experience point, then they still show an "experience 1" despite some of the guys were experience 3 already.

Hopefully this explaination helps.

~;)

zhuge
12-17-2004, 07:06
Maltz's hypothesis makes a lot of sense and would indeed explain the discrepancies in the data without inclusion of a random factor. However, to really confirm this, we'll still need to know where the data for kills is stored in the saved file. I agree that it is most probably stored in a kills for each soldier format, so we might be looking for a series of 40 integers for a regular unit.

I'm no good at hex editing/deciphering saves though and will have to leave that stuff to more talented people. Anyone?

Simetrical
12-19-2004, 01:16
I don't think it'll be a simple kill number. My guess is that it will be just a number, and straight kills will add to it—as will other things, such as merely being in battles. You have to account for units gaining XP without scoring kills.

-Simetrical

Kraxis
12-21-2004, 14:43
== Sometimes experience shrink one level after I hit "start the battle", but get back to original level after the battle is over. I heard this is because the way battle handles exp. is a little different, so the number flips & flops. ~:)

If your general is an expert attacker and is on the defensive, the troops will indeed not be as good in the battle as they might seem on the strategic map) or perhaps the other way around depending on how you move about with your general.

Otherwise a great post. It basically says what I consider the engines does, averages out the total experience level.
But I think that some units are worth more experience than others. For instance I would expect a Spartans to be worth more than a Peasant, and an elephant to be worth more than an Eastern Infantryman.
I remember that one of the devs mentioned that killing (not capturing) a king in MTW was equal to killing about 50 normal enemies. That was pretty accurate as most units that killed a king often rose one level right away. I can't beleive they have changed that.

Sinner
01-05-2005, 15:06
From what I recall of editing savegame files for both STW & MTW, the game did indeed record the number of kills per soldier.

I also vaguely remember a comment somewhere that in RTW less experience was gained for killing routers, for example, than in close combat. If the combat stats at the end of a battle displayed the number of kills with no distinction between types, but the internal record keeping did differentiate for experience purposes, then that would explain the apparent randomness factor when examining the combat stats compared to experience changes.

eg. trooper A got 5 kills, of which 3 were in close combat and 2 routers, while trooper B killed 6, all routers. It might appear that B would earn more experience from the kill totals, but if routers were worth only .5 'kill points' each, then A would have 4 kills as far as experience was calculated and B only 3 kills. Extend that to unit level and we have units gaining the same experience level with differing numbers of kills listed in the stats.

I have a similar nagging recollection that missile kills were also worth less than melee kills. Unfortunately I can't recall on which board I read this, let alone what thread or the reliability of the poster.

Kraxis
01-07-2005, 03:29
Now I'm certain that men are considered as individuals.
I had three units of Militia Hoplites who after a hard fight had been depleted quite a bit. One of them had 1 XP, the others 0. When I reinforced the 1 XP unit suddenly the one I had taken the men from had 1 XP as well, so I took the last one and reinforced that one. Now the second unit lost its XP but the third gained one... So obviously some men in each unit had leveled up, but not all, and not enough in two of them to get an average of 1 XP.

Lord Hammerschmidt
01-22-2005, 16:55
Assuming that experience is done on a man by man basis, here is a potential way to test how many kills each man needs to gain a level of experience. This would require some modding, which is beyond my current knowledge, but perhaps someone would be interested in trying this.

What you need is a 1 man archer unit, and some easy to kill fodder. Tell him to shoot at the unit in question, and you can keep a tally of how many men he kills, and watch to see if he has gained experience (as it seems to be added to the unit card in real-time during battles). This should at least allow you to get a rough idea of where the cut-offs are.

For melee, it's the same concept, except that you would have to mod the target unit to have an attack of 0, so that your poor one man unit wasn't wiped out. Not sure if all of this was possible, but it's a way to answer some of the puzzle.

EDIT: Just realized that this wouldn't account for the experience that seems to be gained just from being at a battle, but it's still something.

zhuge
01-22-2005, 23:55
What you need is a 1 man archer unit, and some easy to kill fodder.

Very interesting idea. However everytime I tried modding the unit number to 1, I got kicked back to desktop immediately. Tried with a 5 men unit and still couldn't play. 10 men is fine though and so is 9 and 6.
So basically I had to resort to testing with the minimum: a 6-man archer unit.

The battle is the same one for Julii as detailed above with the same specs. So the Archers here start at XP1.

A few results:


Kills XP
51 3
30 2
18 2
13 2


Doesn't seem to take many kills to levelup units with fewer men. Killing 13 gives you a level which supports the idea that there is a kill requirement/man to progress to the next level and the unit XP is averaged as a whole.

Is there any good hex editor that allows comparison of 2 files side by side?
We could perhaps take a normal game with 40 men units, make a save then make another saved game with only roman archers units at 6 men. Comparing the 2 saved files should then bring us closer to identifying where the XP info is kept.

Lord Hammerschmidt
01-23-2005, 04:45
Well, I'm not the hex editor sort of guy, but glad to see the idea helped.


On a related note, when you retrain a partial strength unit, the unit retains its previous experience. I.E., a unit with two chevrons will still have those two chevrons, even if you retrain it from a single soldier. If we assume that unit experience works as Maltz suggested (average of each man's experience), it would seem to follow that the men being added during retraining would be assigned the unit's average experience value. Right now, I cannot think of a way to test this through in game data gathering, but if we could find where the experience values are stored, this theory should be confirmable.

therother
01-23-2005, 04:56
Yeah, there certainly seems to be a minimum number of men per unit. AFAICT, a 1-man unit, unlike previous TW games, is not possible.

If you are thinking of exploring saved games, I'd suggest modding as basic a campaign as you can. The best template I've see is Duke John's modding_campaign_template, which is available here (https://forums.totalwar.org/vb/showpost.php?p=651604&postcount=2), with instructions for use. Mod in the units you want, and then start it up as a provincial campaign. Hopefully that will make it easier to find any changes. Plus it'll help you set a structured, repeatable set of tests.

There are various good free hex editors/file comparison tools. For various reasons, I've always liked XVI32 (http://www.chmaas.handshake.de/delphi/freeware/xvi32/xvi32.htm), but you should be able to find quite a few scattered around the web. Download.com (www.download.com) would be as good a place as any to start.

PS A warm welcome to the Org, and especially the Ludus Magna forum, Lord Hammerschmidt! ~:wave: I hope you enjoy your time here. :medievalcheers:

zhuge
01-23-2005, 16:44
Noticed a post dealing with the same topic and thought I'd link it:
http://p223.ezboard.com/fshoguntotalwarfrm7.showMessage?topicID=19572.topic

The message turnover rate at the official forums is pretty high though so it might no longer be available after a few weeks.
Not sure if what Dutch (.com moderator) says is official info from the devs or based on his own views/tests but here it is:

"Experience and kills are both tracked on a per-soldier basis and kept across campaigns, while the experience shown for the unit is the soldier average. This accounts for most of the apparent random-ness - for example when a unit fights a battle more than once different soldiers may do the killing, so that sometimes they level up and affect the unit average, and other times not. It also accounts for any exp changes you may see when merging or partially merging units on the campaign map - there is often a bias towards having more experienced soldiers at the front of a melee unit (first in the unit's soldier list), which can cause some substantial changes in unit exp when they are transferred out.

The number of kills required to gain exp is not a linear progression, and it is possible for units to gain a global one-kills-worth bonus if it has been involved in a battle where the sides are fairly evenly matched (according to the strength calculation). Also, kills on soldiers which are routing have a much smaller chance of counting as a 'kill' for experience purposes."


The figures quoted by erk55555 appear to be quite different from mine though... (80-man Hastati unit needs 400? kills to advance to XP1)... perhaps there are other factors involved which we haven't included.

Simetrical
01-23-2005, 20:34
A minute's Googling got me Hex Workshop (http://www.download.com/3000-2352-10004918.html?part=52461%2520&subj=dlpage&tag%2520=button) as a free download for comparing hex files. I would take any campaign, save the very start, mod any one 0-XP starting unit in the campaign to have 1 XP, start up and save, and repeat for 2 XP and maybe 3 XP. Then compare all the files and post the results. You should hopefully see one difference between all the files, and we can work from there. The problem is, of course, all sorts of random factors—I can just imagine three hundred provinces' random annual farming quality wrecking the whole thing. But someone should try it and see.

-Simetrical

Lord Hammerschmidt
02-09-2005, 20:40
Seems that we were on the right track here. Here's a quote from the official CA Q&A thread at the top of the page.

What exactly makes a unit gain experience? Killing peasants gives less than killing elite units, but what exactly determines this?

Actually, a kill is always considered a kill for xp bonusses, except when the unit is routing (in which case it drops to a 20% chance of counting as a kill). Kills and xp are tracked on a per-soldier basis, and what is displayed is the unit average rounded to the nearest integer. There is a non-linear relationship between kills and xp which resembles a Fibonacci series. Then units which have been engaged in combat in a battle which started with a close to even strength balance may be awarded an extra kill per soldier at the end of combat.

Seems that all that's left unanswered is the actual sequence of numbers of kills that is required to gain a level.

Simetrical
02-10-2005, 04:40
For those who don't know, by the way, the Fibonacci series goes 0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, etc.: you add the previous two numbers to get the next one. What JeromeGrasdyke probably meant is that it would take x experience to reach the first level, y to reach the second, x + y to reach the third, x + 2y to reach the fourth, 2x + 3y to reach the third, etc.

-Simetrical

QwertyMIDX
02-14-2005, 11:25
The question about units not killing a single enemy and still gaining experience remains...perhaps there is carry over from previous battles and some sort of base amount given for taking part in a battle?

Epistolary Richard
02-14-2005, 12:35
I think we have to conclude that.

Before we can properly do the kill-numbers we need to know what is the 'participation' XP (aka the 'well, thanks for coming anyway' XP).

The only way to test it, though, would be starting up a campaign and trying to include a unit in as many battles as possible without it killing anyone until it finally levels up.

roguebolo
04-09-2005, 12:42
Jerome stated that sometimes at the end of the battle if the two sides were approximately equal they were sometimes awarded an experience bonus in addition to that accumulated by their kills.

ZZR Puig
04-11-2005, 12:55
I'll post a few examples of my experience, these are not controlled condition tests, but I think that may be helpful.

The first one is related to the units merging. Happened to me sometimes that I merged two units (about 2/3 of full strength) with EXP 1 and as a result get one unit with EXP 1 and the other with EXP 2. So seems clear that in a unit may coexist soldiers with different EXP level. So when you transfer preferentially some of them with high or low experience the final EXP of the whole unit may change.

Another interesting event happened on a battle. I mistakenly left my general unattended and was charged by cavalry and infantry. I managed to fight back and finally rout the enemy units, but the size of my general unit was reduced to 1. I used my general to chase the enemy units and made a good amount of kills, so during the battle he reached gold chevrons from scratch or almost (perhaps he had 1 bronze chevron at the beginning). When the battle ended a few of the casualties taken by the general escort recovered (3 or 4 if I'm not mistaken) and I then noticed that the resulting general escort unit had only increased their experience to level 3, not EXP 9 as my general already had.

I also think that the kills needed to raise experience may depend on the unit that is inflicting the casualties. If we agree that the kills needed to gain EXP are taken into account as kills per men of your unit, then smaller units, like cavalry, arcanii, etc. should increase their experience faster than bigger infantry units with the same number of kills. A few days ago I fought a battle where the enemy sallied forth. I made good use of my unit of mercenary elephants (36 men in large unit size) and killed 458 men. My elephants had EXP 1 or 2 at most and they didn't gain any additional experience in
that battle. Probably many of that men where routing, as enemy units don't stand fighting for too long when charged by elephants, but even taking that into account I think that this doesn't fit to my previous experience. My cavalry units (54 men each) would surely increase their experience at level 1 or 2 with so many kills, even with routed units. So perhaps there is something odd here.

As has already been said, I also noticed sometimes that the EXP of a unit is decreased when entering a battle, so there is some kind of differentiated treatment of EXP inside and outside battles.

I hope this can be of some use.

A.Saturnus
04-12-2005, 19:22
I tested your observation about elephants in a custom battle.
A unit of Carthaginian armoured elephants (36) killed 63 peasants in ranged and melee. The game halted when the peasants routed so the number of routers killed should be 0 or close to. The experience level of the elephants was 0 and they did not gain a chevron.
In another test, when the elephants killed 155 peasants, they did earn one.

ZZR Puig
04-13-2005, 11:16
I'm not sure which is the usual number of kills per men to reach the first XP level, but 155/36 = 4.3 seems to me that is much more than the usual. Perhaps someone could check, I'll try do do it myself if I have enough time.

k@mi
04-15-2005, 05:01
just something from the game itself that I think many of us noticed, but I would like to mention for those who might not have.

When a unit fights in melee combat, the first units are the ones that get the kills... the down side is that they also are the first to die.
so what you have a is a sort of seesaw
your guys kills enemy, other enemy kills your guys, repeat
i found this very amuzing when my Hastati where fighting in the enemy town square.
one of my guys would kill an enemy, and then another enemy would kill my guy who probably just gained 1 XP
also, this pertains to the units avg exp.
if only the front guys fight, then only they get the kills

so if you have a front guy who has just killed four enemis and then dies, well you just lost 2 XP.

I hope you see where im going with this.


a mini experiment that might be worth trying is to string your guys out in a very thin line and try to flank the enemy, my guess is that if your superior, then more of your guys get kills, there by raising you avg XP.

Hastati vs pesants
and make the hastati as thin as possible

MajorFreak
04-18-2005, 05:43
but the size of my general unit was reduced to 1.ignoring the fact that general's unit size is different than retraining, the best way to go about testing is provided by ZZR. Grab a friend to play multiplayer, have a general unit on one side with maxxed out armour/weapons and on the other have a bunch of archers with maxxed weapons, plus the rest of the slots full of weak peasants.

turn the general unit so that the general is farthest from the archers, pelt away till only the general is left and then send him off against LOOSE formation, HOLDING pattern peasants (who have lots of flanking friends so they rout later than sooner)then you simply count the kills and the chevrons as they pop upit's a shame you cannot modify the general's strategic command rating, nor his traits like "horribly scarred"

Potemkin
05-16-2005, 20:48
FYI: each man in a unit uses 4 bytes in a saved file

Tigranes
05-16-2005, 21:48
I don't know if this will help, but starting at nothing, a group of my peltasts gained 4 chevrons, with 869 kills.

magnum
05-16-2005, 22:43
A question specifically regarding elephants. As elephants themselves are mounts, and the mounts themselves can kill, do these kills count towards exp/chevron gain or does only the kills obtained by the riders count? With horse units its much easier as the horses themselves dont (at least I don't think they do) obtain kills. Could this be the cause of the seemingly high number of kills that are required to gain exp with elephant units?

Uriel
06-10-2005, 11:52
I'm wondering about Wardogs. Do kills from dogs count or is it just the handlers? Perhaps you can get experienced dogs and this experience is lost if the dogs are killed?

TheViciousChicken
06-24-2005, 16:27
I would certainly think so. I used a lot of wardogs against barbarians in my last julii campaign. Usually unleashing 4 wardog units into a town after breaking the gate.

Those dogs easily killed 600 to 800 enemies (warbands are unarmored light infantry, therefore easy to kill for wardogs) but since usually the all the dogs died no experience was ever gained by the unit.

By the end of my northern campagin some of this wardog units should have maybe 2000 to 3000 kills but still zero experience since most togs were lost in each battle. Must be tallied by the dogs than, not the handlers.

Uriel
10-25-2005, 15:23
Just wondering about retraining with 1.3. Does this repopulate the units with inexperienced troops or at the average experience of the unit?

Any observations?

Kraxis
10-25-2005, 20:29
Just wondering about retraining with 1.3. Does this repopulate the units with inexperienced troops or at the average experience of the unit?

Any observations?
The same as always... You still get 'free' experienced soldiers. So those 3-5 super experienced troops are certainly worth to keep clear of combining and send them home to be retrained.

Willie McBride
11-04-2005, 02:35
Yes- if by "free," you mean pay extra for. So far as I can tell, there's a significant increase in cost to replenish a highly-experienced unit.

hellblazer
11-04-2005, 07:37
Yes- if by "free," you mean pay extra for. So far as I can tell, there's a significant increase in cost to replenish a highly-experienced unit.

You just contradicted yourself. You said that you don't pay extra to retrain experienced troops and then said that there IS a significant cost increase to retrain experienced units.

Willie McBride
11-04-2005, 08:07
There is no contradiction- not if you define "free" as 'pay extra (cash) for.' These replacements are still "free" in that they start with the same high level of expertise without having to go kill several hundred folk apiece- presumably, they learn to kill from the core veterans who have seen the elephant.
This is not so outlandish as it might seem- after all, units retain their experience over hundreds of years in the game, although no individual would be in the ranks for than a couple decades. Rather, experience and esprit de corps must be handed from each generation of grizzled veterans to the green boys who come to replace them.

Trithemius
11-05-2005, 05:15
In BI it seems that sometimes after retraining experienced units I get a drop in experience. Perhaps the way in which units retrain has been altered for BI?

I also noted that I could retrain units trained elsewhere to benefit from the automatic experience increases granted by structures (i.e. I could retrain experience 1 Bosphorans in a town with a top-level Barracks and a top-level Kolaksy shrine and end up with experience 4 Bosphorans). As far as I know this is different from RTW.

Willie McBride
11-07-2005, 23:47
I have noticed in the most utd RTW (without BI) that I can retrain units with no experience to units with experience in a town with the appropriate temples. However, this does seem to require the appropriate bldg types as well, eg can't increase merc experience, and need high-level barracks to retrain high-level infantry.

Artuk
11-17-2005, 05:26
Hi,

I was reading through this about individuals in a unit having the experience and I have observed something when I combine units in BI that seems to contradict this. I have observed that when I combine a portion of a unit with chevron or two into another unit but have left overs, the left overs can go up in experience..

In example, I had a 2 chevron and a 3 chevron unit both damaged while playing as the saxons. I dragged the 2 chevron onto the three to fill it out. There were 7 men left and their experience actually went up leaving me with a pair of three chevron units.

At the time I figured that what happened was the unit unit experience pool is divided amongst the men remaining in it.. so when I transferred men out the amount of experience per man was increased.

This is anecdotal, and I didnt think much of it until I read this. I will try to reproduce it next time I play.

Regards,

Artuk

magnum
12-28-2005, 16:24
When combining units it is possible for a unit (either the one gaining or the one losing troops) to go up in level. It is also possible for them to decrease in level. Since RTW tracks exp on a per man basis, the chevrons displayed are simply an average of the exp of all the men in the unit. When two units are combined, RTW simply takes the needed number of men off the front of the giving unit and places them at the end of the recieving unit.

The only time if truely makes a difference is when you are planning on retraining a unit. Since replacements are of the avg experience of a unit, combining two units to possibly 'gain' a chevron allows you to get new recruits that are 1 chevron higher in experience. (Of course things go go the other way and your unit 'loses' a chevron of experience.)

Stoneface
01-23-2006, 01:58
I did some experiments.

Setup:
Custom Battle
Berserkers (gold weapons and armor) against peasants (no upgrades) defending a townsquare.
Fist slot was a token general, second slot the berserkers.
I only did two valid tries for zero and one chevron so the numbers should be taken with care.

Results:
The berserkers needed to kill 47 peasants on the first run and 45 on the second run to gain the first chevron. On the second run they lost one man.

To go from one to two chevrons they needed to kill 74 peasants on both runs.

They needed to gain a little less than two kills to get the first chevron and a little more than three for the second.

Conclusion:
I think CA uses "Gimli's rule for counting slain enemies" i.e. everything that isn't routing counts as one. Also War and Armored Elephants may count for three. "Experiments" with berserkers against a town full peasants also support this.

A unit needs about twice as many kills to gain its fist chevron as it has men. This is in line with my game experience.

I think CA counts kills for each soldier individually. The way some units gain experience while merging supports this.

Things I do not (yet?) know:
Is the average experience rounded to the nearest or rounded down.
The exact number of kills a soldier needs to gain a certain chevron. It may be only one if the displayed experience if rounded down and up to three if rounded to the nearest.

Further experiments required maybe with more armor and hitpoints for the berserkers and more moral for the peasants and outside town.

Edit 1:

Upgrades my berserkers and increased their armor, defense and hitpoints by ten. Also gave peasants a morale of twenty. More of this later.

Berserkers against Chosen Swordman. Battle setup as above.

Berserkers needed 46 and 49 kills to get the first chevron. This pretty much confirms the "each dead enemy counts as one" rule. Cavalery and chariots still have to be tested.

Soon to come:
An estimation of kills per man required to gain the first five chevrons by 24 berserkers and 601 peasants

The effect of distribution. Comparing Missile and Phalanx gaining experience by slaughtering helpless peasants.

Edit 2:

After a long time I came back to some chevron resarch. I did five runs of the 24 berserkers against 601 peasants.

Results:
1st chevron at 55/46/55/50/57 average 2.2 kills per man
2nd chevron at 134/127/134/138/133 average 5.6 kills per man
3rd chevron at 206/217/206/215/230 average 9.0 kills per man
4th chevron at 314/310/314/307/308 average 12.9 kills per man
1st chevron at 428/438/428/416/411 average 17.7 kills per man
I didn't go beyond five chevrons, because of the enormous number of peasants required.

A note on distribution: I do not have any good data yet, but slingers seem to need less kills per man in comparision to phalanx. Phalanx a bit more than the berserkers, slingers about 1.9 kills per man. Playing with the numbers revealed no hint of an possible algorithem so I believe CA counting method is a pretty obscure one.

Any comments an further research welcome.

magnum
02-16-2006, 22:36
Actually the difference you are seeing can be explained by the difference in the two types of units and how they kill. Remember, what is being displayed is the average level of the individuals in that unit. So some points to remember as you look at your data:

1) All slingers can attack (and thus get a kill) at the same time and so can advance at roughly the same rate as the others in the unit.

2) A Phalanx unit however only the front rank(s) can attack and get kills. The further back you go the less likely they are to get a kill.

The experience series goes roughly like 1, 1, 2, 3, 5. So for example take 2 40 man units, one slingers, one phalanx. The slingers kill 40 men, 1 each. The displayed exp is 1 (rounded down). Phalanx killed 40 men also, the front rank of ten men killed 4 each, while the other 30 men killed none. So 10 men have level 3, while 30 have level 0. Average displayed (rounded down) is 0.

Now the thing to remember is that is just the displayed information. Its used for nothing other than letting us the player have an idea what level of performance we can expect from that unit. The game doesn't care about that, its using the actual experience level of each individual soldier. In the above example, the front 10 phalanx troops are exp level 3 and since they're the only ones actually fighting they're not to bad.

Both the experience needed to level (1, 1, 2, 3, 5) and the rounding down are examples only. I don't know of anyone who knows exactly how many kills it takes to go up in each level. And as far as rounding goes, again no one knows for sure and in my opinion the game actually uses different rounding methods for the displayed exp level at different points in the game.

What I'm trying to show is how it could easily appear that some types of units require more kills to level than others when in reality (err, in game) it's just a matter of how those kills are spread amongst the unit members.

econ21
02-17-2006, 22:25
In the above example, the front 10 phalanx troops are exp level 3 and since they're the only ones actually fighting they're not to bad.


Interesting, but if it's always the same guys at the front that fight, doesn't that make the units experience level very fragile? The guys at the front are almost always the ones who will die, so does that not mean the unit will find it very hard to get to a high experience level as the vets are gradually culled? Loses for the slingers will be much less troublesome.

Stoneface
02-20-2006, 19:33
Actually the difference you are seeing can be explained by the difference in the two types of units and how they kill. Remember, what is being displayed is the average level of the individuals in that unit. So some points to remember as you look at your data:

1) All slingers can attack (and thus get a kill) at the same time and so can advance at roughly the same rate as the others in the unit.

2) A Phalanx unit however only the front rank(s) can attack and get kills. The further back you go the less likely they are to get a kill.


You found the reason for my test setup. Good



The experience series goes roughly like 1, 1, 2, 3, 5.


Sorry, I am quite sure this is not even remotely in the right area. Please quote your sources so we can find the reason why they contradict with my research.



So for example take 2 40 man units, one slingers, one phalanx. The slingers kill 40 men, 1 each. The displayed exp is 1 (rounded down). Phalanx killed 40 men also, the front rank of ten men killed 4 each, while the other 30 men killed none. So 10 men have level 3, while 30 have level 0. Average displayed (rounded down) is 0.


So I thought too but I am pretty sure I was wrong. The difference just isn't big enough. Assuming that only a quarter of the unit does the killing and the fourth chevron takes at least 12 kills (average 12.9 in the berserker vs peasants test) the phalynx unit would need 3 kills per man for the first chevron. The actual number is around 2.3. If you do not believe me run your own tests.



What I'm trying to show is how it could easily appear that some types of units require more kills to level than others when in reality (err, in game) it's just a matter of how those kills are spread amongst the unit members.

Yes, I am pretty sure it is. But I am also pretty sure there is a more complicated calculation going on. If the spread is uneven a different member of the unit might be awarded the kill or something like that.

Sorry I got a bit lazy in the end and was not as clear as I wanted to be.

Severous
02-20-2006, 20:42
Hi

Some interesting observations and research on this forum and in this thread.

Do you think soldiers may gain experience without killing anyone ?

The act of attacking, defending, firing ? Just being in a battle?

I had a small barbarian cavalry unit gain a level of experience without inflicting or sustaining any casulaties. It did run through an enemy unit. Not to attack it but because it was in the way. There were a few clashing sounds but no casualties.

magnum
02-21-2006, 20:36
As far as where I got 1 1 2 3 5 etc, it was stated by a dev that the kills needed for advancement was LIKE the Fibonacci series (which is 1 1 2 3 5 etc). It was re-stated earlier in this thread. Now the dev said like, so I'd take that to mean not exactly. However it would be reasonable to assume that because what the devs have said, it takes increasing numbers of kill for each level of exp.

As far as front rank earning the kills making melee units rather easy to gain/lose experience your are correct. In the battle screen if you watch a unit you can sometimes see that happen as a unit pops up and down 1 level of experience as the front rank kills or are killed. I have no idea if the location of the men in formation is static from one battle to the next nor how replacements are added / removed from a unit. So this battles front men might not be next battles front men.

Finally, in response to Severous2's query. CA has stated that in battles where the sides are roughly equal (they didn't say what represented roughly equal) that 'bonus' exp is awarded to all units that participate in the battle. So yes, a unit can gain exp even if they never engaged, simply for being on the battle field.

General4Hire
02-21-2006, 21:59
I suspect it's hopeless. There have been ample anecdotal reports of units not participating in the battle and still gaining experience. I should think that there are lots of factors, quite possibly including some random ones. I doubt there's any way to properly isolate them.

-Simetrical

Is it possible that some general's retinue includes "experience for all troops in a battle"? Maybe this is hidden at times? who knows...heh

Avicenna
02-26-2006, 18:12
Different amounts of experience for killing different units and different amounts of experience for different types of units might explain the occurence where a unit that killed 98 got 1 chevron and a unit that killed 100 got none. eg. Peasants gain exp faster than hastasi because generally they are less capable of killing.

Stoneface
02-27-2006, 13:12
The exact number of kills to gain a certain amount of experiance does vary a bit. The sizuation Tiberius discribes is a perfectly normal one. I made a quick test of super berserkers against bastarnae and egytian melee chariots. The experiance gain for killing peasants and bastarnae is similar, so neither troop quality nor hitpoints seem to count. Againt chariots the berserkers gained experiance somwhat slower than against peasants probably because the spread is more uneven. One kill is acounted per man, not per chariot.

symball
02-28-2006, 12:57
Something you might not see in custom batles which certainly opens up the possibility that there is a 'base' experience for being in a battle is that I saw that after a battle in which my reinforcements did not arrive (huge settings) there were a couple of units in the reinforcements that gained a chevron!

I am at work but I will post screenshots when I get home

http://www.imagedump.com/index.cgi?pick=setandget&tp=387463&poll_id=0&category_id=19&warned=y

Stoneface
03-02-2006, 16:03
Unfortunetly there are many things I did not see in my tests and the participation bonus is among them. I also negelected the effect of dying, which might lead to an even slower experience gain for phalanx. Autocalc is also neglected and probably a few other things as well.

Severous
03-02-2006, 21:37
A General with 3 chevrons of experience. Lost half his force. Spent several turns stationary in a town and regained his lost bodygaurd. All came in at the 3 experience level.

Byzantine Mercenary
03-04-2006, 00:31
this threads pretty long so someone else may have said this already, but im sure i heard CA say somewhere that routing troops have a 1/3 chance of counting as a kill for experience purposes, at least thats what i remember, i read it on this forum a while back...

macsen rufus
04-07-2006, 14:33
Those guys gaining experience by doing nothing have caused me great confusion! I haven't looked at it in detail, but I get the feeling that troops gaining experience without fighting have earned it *somehow*.

Cavalry held back from the action in a siege assault *could* have gained experience by not going into an unfavourable battle, and also impetuous troop types have also gained experience where they haven't been used in the battle, so maybe it represents a gaining of discipline.... Also if experience gained is related (somehow) to the starting odds of the battle, maybe this behaviour of staying out of a fight can actually be useful training. "Good decision, men, stay out of that or you'll be mincemeat" is one possible scenario. But how it works in game mechanics terms is a mystery, alas... One thing's for sure, though, it's not purely down to number of kills.

Vincent Butler
05-30-2014, 21:54
That might explain why sometimes when a unit is depleted and is retrained, it loses experience, and sometimes it does not. Maybe the less experienced troops were killed, the more experienced survived, and so the overall experience would not have been affected.