There was a thread that raised questions about cavalry melee results but was I think never resolved. The question was why Prussian Landwehr can sometimes easily defeat elite French cavalry units. I think I may have found the answer and I would be pleased if others were to check it out. The simple explanation is this - melee combat is partly resolved using a factor called "stab time" i.e. how many seconds between sword blows or thrusts that a combatant can deliver. This value can be improved (made shorter) or deteriorated by two attributes - experience which can make it up to 4 seconds shorter (or 15 seconds longer) and "edged weapons proficiency" which can make it up to 8 seconds better but never longer. The standard stab time for infantry versus infantry is 20 seconds so the best infantry can stab every 8 seconds whilst the worse are actually 35 seconds so elite infantry can be 4 times better than poor infantry.
The issue I have found is that for cavalry versus cavalry the standard stab time has been set as 40 seconds. Prussian landwehr cavalry units are given experience of 4 - this gives them a -1 attribute for stab time whereas one of Milhaud's units might have experience 7 which gives them a -4 rating. Obviously these two adjustments to a time of 40 seconds are relatively insignificant. The edged weapon ratings are also relatively small compared to the 40 being -6 for Milhaud's men and -2 for the Landwehr. Thus the total difference in ratings become 37secs versus 30. Thus battles between them are relatively even and other factors than experience or edged weapons will determine the outcome.
The stab time for cavalry versus infantry is 5 seconds and in this case the quality ratings hardly have an impact as even the best cavalry cannot get below 2 seconds so their attribute bonuses are largely wasted.
I have experimented with changing the standard stab time to 14 seconds for cavalry versus both infantry and cavalry. This means that the most elite French units who get a -8 second attribute boost would actually go to a stab rate of every 2 seconds (the minimum possible) whilst the Landwehr cavalry would go to 11 seconds. thus the elite would be just over 5 times better which is probably the right sort of score and the same sort of relativity for infantry.
If I am correct that this is an error perhaps Reb could put a revised version of the Statetables file in his toolbar mod so everyone can get it for their system without having to add another mod.
I have added a mod containing the Statetable changed version with Cavalry Stab time set to 14 seconds.
I have since realised that I did not need to differentiate so much between cavalry unit qualities when the cavalry is attacking infantry so my new version of the mod will have all cavalry rated good or better equally effective versus infantry. This mod can be used with standard game scenarios. In my experimental scenarios in another thread I have a different version of unit attributes where fatigue reduces the effectiveness of cavalry v infantry in that variant fatigue will have less impact on higher quality cavalry units so quality becomes a relevant factor again.
Regards
Mike
Cavalry fault found?
Cavalry fault found?
- Attachments
-
- CavalryMeleeModv2.zip
- (2.85 KiB) Downloaded 4 times
Last edited by mcaryf on Wed Nov 14, 2018 12:48 am, edited 1 time in total.
Reason: Revised version of mod attached
Reason: Revised version of mod attached
Re: Cavalry fault found?
Mike,
Considering that the problem is only happening for certain people and not others, I'd suggest not changing anything.
Some possible problems:
There is no absolute guarantee that the statetables.csv file published in the SDK is the same file used by the game. There is no guarantee that everyone is using the same files, and the problem might not be the statetables.csv file.
Problem reports never contain everything that might be causing the problem.
When dealing with statistics, there are many perceptual errors. Someone might report something that only happens once as happening every time. Adverse results are over reported and I have never seen a case where someone "changed sides" and reported the same adverse result working in his favor. An example of this is someone who claimed a random number generator was broken because it rolled too many doubles with dice. All my questions about how often doubles should be expected or if only his opponent was rolling too many doubles were ignored. Adverse results are always more noticeable than favorable ones.
Always remember the first rule: The random number generator is perverse and hates you. Any result that CAN happen, WILL happen. You can run 1000 tests and never see a result outside normal limits, but see it three times in a row the first time you run the code.
In the case of the skewed cavalry results, someone was doing something differently. Find out what that was and you can possibly find a way to fix it. In most cases the fix is "Don't do that".
Hook
Considering that the problem is only happening for certain people and not others, I'd suggest not changing anything.
Some possible problems:
There is no absolute guarantee that the statetables.csv file published in the SDK is the same file used by the game. There is no guarantee that everyone is using the same files, and the problem might not be the statetables.csv file.
Problem reports never contain everything that might be causing the problem.
When dealing with statistics, there are many perceptual errors. Someone might report something that only happens once as happening every time. Adverse results are over reported and I have never seen a case where someone "changed sides" and reported the same adverse result working in his favor. An example of this is someone who claimed a random number generator was broken because it rolled too many doubles with dice. All my questions about how often doubles should be expected or if only his opponent was rolling too many doubles were ignored. Adverse results are always more noticeable than favorable ones.
Always remember the first rule: The random number generator is perverse and hates you. Any result that CAN happen, WILL happen. You can run 1000 tests and never see a result outside normal limits, but see it three times in a row the first time you run the code.
In the case of the skewed cavalry results, someone was doing something differently. Find out what that was and you can possibly find a way to fix it. In most cases the fix is "Don't do that".
Hook
Re: Cavalry fault found?
One thing I noticed in going over the Cavalry thread:
People having the problem say they are using version 1.01 of the game. If I understand correctly, version 1.01 was known to have the cavalry problem and it was fixed in version 1.02.
I have the Steam version of the game, purchased 2/28/18 as the Collector's Edition. I can't find anywhere that says what version I have in any of the code file metadata, but the readme file says version 1.02.
How do people know they are using version 1.01? How can I check my game?
If the Steam version of the game is not updating automatically (it IS possible), then try the "verify integrity of game files" function. Right click the name in the Steam client, select "Properties". Click the "Local files" tab and click "Verify integrity of game files."
It is not impossible that the Steam version of the game is different from the retail version. I have another game that is thought to have that problem. Luckily I have the developer's retail version rather than the Steam version of that game.
Hook
People having the problem say they are using version 1.01 of the game. If I understand correctly, version 1.01 was known to have the cavalry problem and it was fixed in version 1.02.
I have the Steam version of the game, purchased 2/28/18 as the Collector's Edition. I can't find anywhere that says what version I have in any of the code file metadata, but the readme file says version 1.02.
How do people know they are using version 1.01? How can I check my game?
If the Steam version of the game is not updating automatically (it IS possible), then try the "verify integrity of game files" function. Right click the name in the Steam client, select "Properties". Click the "Local files" tab and click "Verify integrity of game files."
It is not impossible that the Steam version of the game is different from the retail version. I have another game that is thought to have that problem. Luckily I have the developer's retail version rather than the Steam version of that game.
Hook
Re: Cavalry fault found?
Hi Hook
Thank you for replying even if I think you are somewhat pessimistic about the integrity of the released version.
I entirely accept that there needs to be an element of randomness in the outcome of combats within a game but that randomness has to be within a context. One of the strengths of SOW Waterloo is that it attempts to differentiate units both by levels of experience and particular attributes. Those units with more experience and superior attributes should win more often. Thus the attributes for Imperial Guard infantry fighting in melee against Prussian militia are such that they should win 4 times out of five, the randomness comes in by a player not knowing when that fifth time might be.
My experience with cavalry combat is that there is just too much randomness and not a sufficient bias in favor of the better trained and better equipped unit. In consequence I went looking in the various game files that influence combat results and I think I have found some faulty settings in the Statetables files that give for example an advantage of 33/40 versus 37/40 between Imperial Guard Cavalry and Landwehr. Now 40 seconds is quite long for a melee combat anyway so it is quite probable that the combat would be resolved by a single first hit or at most 2 which both sides might make given that differentiation.
It is a perfectly simple mod to make just requiring a minor adjustment to one or possibly two values in entries to be found at the end of the Statetables file. If there is general consensus that the original values were some sort of mistake what would be the objection to changing them? I have not yet looked at the KS mod's values in this file but I will do to see if they picked up the same incongruity.
Regards
Mike
Thank you for replying even if I think you are somewhat pessimistic about the integrity of the released version.
I entirely accept that there needs to be an element of randomness in the outcome of combats within a game but that randomness has to be within a context. One of the strengths of SOW Waterloo is that it attempts to differentiate units both by levels of experience and particular attributes. Those units with more experience and superior attributes should win more often. Thus the attributes for Imperial Guard infantry fighting in melee against Prussian militia are such that they should win 4 times out of five, the randomness comes in by a player not knowing when that fifth time might be.
My experience with cavalry combat is that there is just too much randomness and not a sufficient bias in favor of the better trained and better equipped unit. In consequence I went looking in the various game files that influence combat results and I think I have found some faulty settings in the Statetables files that give for example an advantage of 33/40 versus 37/40 between Imperial Guard Cavalry and Landwehr. Now 40 seconds is quite long for a melee combat anyway so it is quite probable that the combat would be resolved by a single first hit or at most 2 which both sides might make given that differentiation.
It is a perfectly simple mod to make just requiring a minor adjustment to one or possibly two values in entries to be found at the end of the Statetables file. If there is general consensus that the original values were some sort of mistake what would be the objection to changing them? I have not yet looked at the KS mod's values in this file but I will do to see if they picked up the same incongruity.
Regards
Mike
Re: Cavalry fault found?
Just looked at the KS Mod and they have already picked up on this Statetable anomaly. They have different values not the same as mine but in a similar ballpark. I would have to examine other factors they might also have changed before fully understanding why they chose the values they did.
Anyway I will carry on testing my change. These are what the KS Mod has for cavalry versus infantry and then cavalry versus cavalry
Melee hit chance/1000 ................ Stab time in seconds
Cav V Infantry: 198 ................ 6
Cav v Cav 100 ................... 10
The standard game has
Cav v infantry 200 ................. 5
Cav v Cav 50 .................... 40
My plan was to change both Stab times to 14 as the maximum benefit for high quality units reduces stab time interval by 12 secs
Regards
Mike
Anyway I will carry on testing my change. These are what the KS Mod has for cavalry versus infantry and then cavalry versus cavalry
Melee hit chance/1000 ................ Stab time in seconds
Cav V Infantry: 198 ................ 6
Cav v Cav 100 ................... 10
The standard game has
Cav v infantry 200 ................. 5
Cav v Cav 50 .................... 40
My plan was to change both Stab times to 14 as the maximum benefit for high quality units reduces stab time interval by 12 secs
Regards
Mike
Last edited by mcaryf on Fri May 18, 2018 5:31 am, edited 1 time in total.
Re: Cavalry fault found?
I can't comment much on the cavalry problem because I don't have it. It seems to be limited to people who are somehow running version 1.01 rather than 1.02. At this point the only reason to change the values is because you can as the original problem has already been fixed.
The game is like a multi-dimensional spreadsheet array where changing one value will change others, and some values will change half the values in the entire spreadsheet. This is an area where angels fear to tread, and I will not change any value that I can't predict the results. Then I'll test the heck out of it. I have backed out changes that didn't perform as I expected and most of the time no one ever hears about them.
By test, I mean 5 or more of the same battle with ONLY the one change you are testing. Then do the same with a couple more battles. Pretty much standard beta testing procedure.
Like I said, I do not have the cavalry problem and I have no wish to alter the values for my game. I'm not saying not to make them, or not to publish them, but they have no place in the standard game. Or in this case, in a mod that is considered standard, like the Grog toolbar. The KS mod is a whole separate game and I have no problems with what they've done.
Hook
The game is like a multi-dimensional spreadsheet array where changing one value will change others, and some values will change half the values in the entire spreadsheet. This is an area where angels fear to tread, and I will not change any value that I can't predict the results. Then I'll test the heck out of it. I have backed out changes that didn't perform as I expected and most of the time no one ever hears about them.
By test, I mean 5 or more of the same battle with ONLY the one change you are testing. Then do the same with a couple more battles. Pretty much standard beta testing procedure.
Like I said, I do not have the cavalry problem and I have no wish to alter the values for my game. I'm not saying not to make them, or not to publish them, but they have no place in the standard game. Or in this case, in a mod that is considered standard, like the Grog toolbar. The KS mod is a whole separate game and I have no problems with what they've done.
Hook
Re: Cavalry fault found?
If you select the multiplayer tab it will tell you what version you have, mine still says 1.01 & your right the readme file says version 1.02.One thing I noticed in going over the Cavalry thread:
People having the problem say they are using version 1.01 of the game. If I understand correctly, version 1.01 was known to have the cavalry problem and it was fixed in version 1.02.
I have the Steam version of the game, purchased 2/28/18 as the Collector's Edition. I can't find anywhere that says what version I have in any of the code file metadata, but the readme file says version 1.02.
How do people know they are using version 1.01? How can I check my game?
If the Steam version of the game is not updating automatically (it IS possible), then try the "verify integrity of game files" function. Right click the name in the Steam client, select "Properties". Click the "Local files" tab and click "Verify integrity of game files."
It is not impossible that the Steam version of the game is different from the retail version. I have another game that is thought to have that problem. Luckily I have the developer's retail version rather than the Steam version of that game.
Hook
If the steam version is different from the retail version would it be still possible to play a multiplayer game?
Leicestershire
Re: Cavalry fault found?
Just looked at the KS Mod and they have already picked up on this Statetable anomaly. They have different values not the same as mine but in a similar ballpark. I would have to examine other factors they might also have changed before fully understanding why they chose the values they did.
Anyway I will carry on testing my change. These are what the KS Mod has for cavalry versus infantry and then cavalry versus cavalry
Melee hit chance/1000 ................ Stab time in seconds
Cav V Infantry: 198 ................ 6
Cav v Cav 100 ................... 10
The standard game has
Cav v infantry 200 ................. 5
Cav v Cav 50 .................... 40
My plan was to change both Stab times to 14 as the maximum benefit for high quality units reduces stab time interval by 12 secs
Regards
Mike
If you do try this can you make it as a mod for me to try. For the certain people who have this bug it's a game killer.
Leicestershire
Re: Cavalry fault found?
Hi,
I've literally just stumbled across this thread whilst checking up on some others that I am interested in, and wondered if it explains some of the strange behaviour I'm getting with the Dutch cavalry in the opening stages of Quatre Bras. They seem able to wipe the floor with Pire's Cavalry Division despite being outnumbered and up to now I've just been putting it down to my use of the KS Mod which is know to cause issues with stock scenario's.
However, I did check my game and according to the Multiplayer Menu I am still on Version 1.01, despite having the game on Auto-Update with steam.
To try and resolve this I have downloaded the 1.02 update .exe from Nordsoft direct, but when i try and run it I get a message telling me that it cannot find an installed version of the game. Presumably because it;s not checking the Steamapps folder.
Does anyone know how to a) Get steam to download the update, or b) get the Norbsoft update to recognise the steam installation?
I've literally just stumbled across this thread whilst checking up on some others that I am interested in, and wondered if it explains some of the strange behaviour I'm getting with the Dutch cavalry in the opening stages of Quatre Bras. They seem able to wipe the floor with Pire's Cavalry Division despite being outnumbered and up to now I've just been putting it down to my use of the KS Mod which is know to cause issues with stock scenario's.
However, I did check my game and according to the Multiplayer Menu I am still on Version 1.01, despite having the game on Auto-Update with steam.
To try and resolve this I have downloaded the 1.02 update .exe from Nordsoft direct, but when i try and run it I get a message telling me that it cannot find an installed version of the game. Presumably because it;s not checking the Steamapps folder.
Does anyone know how to a) Get steam to download the update, or b) get the Norbsoft update to recognise the steam installation?
Re: Cavalry fault found?
Thanks Roy. My multiplayer says 1.01 as well. If you have a readme that says 1.02, then the other files should be correct. As for different versions being able to play multiplayer together... well, some people will be using 1.01 (because they didn't get the patch) while others will be using 1.02 which is reported by the game as 1.01. Was this a problem when the patch was first released? I've seen cases where the "multiplayer version" was not necessarily the same as the "game version" when two game versions could play multiplayer together.
In any case it is worthwhile to look for a fix and you have a couple of people who could test any changes.
Best of luck.
Hook
Mike: See any possible problems here? If Didz has the problem with the KS mod, it might not be that file causing it as KS already changed those values.I've literally just stumbled across this thread whilst checking up on some others that I am interested in, and wondered if it explains some of the strange behaviour I'm getting with the Dutch cavalry in the opening stages of Quatre Bras. They seem able to wipe the floor with Pire's Cavalry Division despite being outnumbered and up to now I've just been putting it down to my use of the KS Mod which is know to cause issues with stock scenario's.
In any case it is worthwhile to look for a fix and you have a couple of people who could test any changes.
Best of luck.
Hook