Combat Results vs Predictions

[Zerstorer]

Can the combat results be more consistant to ensure the predicted combat results are realized "unit wide" (instead of just per individual strenght point)?

While a certain amount of randomness is required/necessary, sometimes over-whelming odds that should always result in "victory" do not. On several occassions a unit with a low percentage odds of a kill (ie; 14-17% etc) achieved 80-90% kills and another unit in the same combat with high odds (80-85% etc) only had 20% kills. This occurs for both the AI and the player a few time each scenario. For example, I've seen 2 strenght towed AT hold out in the open against a 10 strenght Grenadier and take no loses (while inflicting 2)... the odds of a kill where only 17% for the AT and 80% for the Infantry so the expected losses where the exact opposite. Since victory conditions are turn based, a couple of combats like this can have catastrophic effect on the scenario. Unless this was by design, I believe there may be a problem with some combat result calculations or its end application.

I realize this is just a game, but such wildly varying results tends to undermine the strategic elements of the game. An attacker to defender combat with a 5-1 ratio (or more) should always be a win (of some degree). Die roll results which allow the defender to occassionally win against an overwhelming attack may seem to be exciting, it detracts from the strategic & historical elements of the game.

Can the combat results table (or its application) be modified so that when the game says "80% chance of a kill"... that there will be indeed a 80% kill average for EACH UNIT (+/- 20% for example)... not just per individually attack factor? If a 10 strenght unit is given an 80% chance of kill (per strenght pt) the end result should be at least 6 kills (example) - never no kills. Since combat die rolls only appear to reference individual attack factors and not the entire unit, it appears that a 90% kill could theorically mean no kills at all for the entire unit.

My appologies for the rather lenghty rambling as I only wished to address a game mechanic that I believe will improve the game. The game runs flawlessly and is a major improvement over PG, so this post was not intended to deminish that accomplishment at all.

[Rudankort]

I do agree that sometimes combat results can get way off. I remember attacking that polish cavalry in the Poland scen with german inf, and german inf was fully destroyed while cavalry took 1 or 2 damage. Oops.

So yes, this aspect may need some more attention.

Random number generators are tricky. The generator I use now is very close to perfect uniformly distributed generator on a big number of samples, but it does generate long sequences of similar (similar low, or similar high) numbers from time to time. I'll think what I can do about it.

PS. I think I did not say it yet, but it's nice to see you on this forum.

[SeaMonkey]

Just cap the randomness at a plus or minus 2 applied to the predicted results. This works well for SC which has the same unit strength scheme.

[Zerstorer]

Thanks for giving this issue the attention this fine game deserves, you have done a remarkable job correcting so many issues the original game had. I also apprieciate the time and dedication you have show to the forum and continued support for the game.

And thanks so much for the welcome, it's been a while since I worked with the PG series and it was only recently that I ran across your PGF - such an improvement itself. Best to you and continued success with PC.

[BriteLite]

Hey I remember you Zerstorer![:)] Welcome.

For those who don't know back in the days before Rudankort did PGForever, Zerstorer was one of a handful of modders who worked on improving PG. His equiptment file was considered by many to be the best, including me.[&o]

[Josh]

ORIGINAL: BriteLite

Hey I remember you Zerstorer![:)] Welcome.

For those who don't know back in the days before Rudankort did PGForever, Zerstorer was one of a handful of modders who worked on improving PG. His equipment file was considered by many to be the best, including me.[&o]

In that case: welcome to the forum, we owe you. [&o]

[VPaulus]

Welcome.

[Zerstorer]

Thanks BriteLite, that was quite a compliment. And thanks to each of you for the welcome. This forum and it's members contributions are another way PC is such a improvement to the PG series. And finally, one last thank you to Rudankort! PC is really very nicely done. A near seamless blending of many improvements which has greatly enhanced the old classic.

[fuelli]

Just cap the randomness at a plus or minus 2 applied to the predicted results. This works well for SC which has the same unit strength scheme.

I don´t think that the randomness should be capped. Just the probability of results that differ significantly from the prediction needs to be kept at a reasonable level. I have not played enough yet to say that there is something wrong with the probability up to now.

[Razz1]

I have been petitioning to reduce the randomness for some time.

If you play the Allies you can really see how it hurts.

[Dragoon.]

Well the randomness is not a problem per se, like with AT units it always have been part of discussion since the first PG and the 7 other games that followed it.
I mean AT units are by nature and design a defensive unit not meant to lead the attack, at best defending the flanks. Since the player usally is on the offensive he find limited usage in opposite to the defending AI. But back to the randomness there was always a certain spread with all games. People naturally complain when they get blown on the nose, and less if the randomness caused them to end up better than expected.
The only issue I have noted and it got already Rudankort attention is that the number generator sometimes generates long sequences of similar results.
I had one freak accidents with 48 rolls in succession all between 3 and 29! After that I was considering tempering with my high resolution timer.
Of course I wouldn't have complaind if I had 48 rolls all between 71-97. [:D]

[Mart.n]

I absolutely agree that the luck factor has to be capped. It's frustrating if you lose 4 units to a single infantry due to bad luck. I know that it evens out over the course of the game, but this favours the load/save abuse so much.

Maybe you can compare it to traditional RPGs - a critical hit deals 2x damage. That's harsh but no overkill. A critical does not automatically kill your whole party.

I would at least love an option to cap the RNG at a plus or minus 2 applied to the predicted results as mentioned above.

[willgamer]

ORIGINAL: Dragoon

I had one freak accidents with 48 rolls in succession all between 3 and 29! After that I was considering tempering with my high resolution timer.

Actually that essentially impossible!

If taken as 1 chance in 3, the odds are computed as .3 to the 48 power ~= 8 x 10 to the -26 power.

This is about 1 chance in 1,000,000,000,000,000,000,000,000,000. [X(]

Strong conclusion: If that sequence can be confirmed, the random number generator is broken. [:(]

[Dragoon.]

Seriously I was shocked when several combat results in a row where completely out of line. A strength 10 PZIV attack a strength 6 infantry units on a river hex without doing any damage but manage to lose one tank. But what made me break out in loud maniacal laughter was when a strength 10 ME110 attacked full strength french engineers in a forest hex losing 4 planes and doing no damage. I finally pressed L to see the dice rolls. All 10 below 30 and had several red numbers in the nineties.

Now I got curious. Since the game don't change seed on reload I loaded the axis autosave. Replayed the whole turn in the same order and checked the detailed combat results each time for the dice rolls and there was that ridicules spree again where all my rolls where a max 29 and AI in exchange rolled many red numbers.
Unfortunately I didn't keep the autosave to show it off however I wished a camera had save the dumb expression on my face. [;)]

[SeaMonkey]

I've played through LC on FM difficulty and I'm noticing some really outlandish combat results.  Maybe we should have a discussion on survivability of units at this level?  If we're talking about battalion sized units, then only when units are completely outnumbered and semi surrounded should they suffer a complete demise in one turn.
 
Some of the massed unit attacks seem to come off with completely fantastical results as one unit that has suffered air and artillery softening, plus soak-off attacks by ground units, will deal out extensive damage to full strength, experienced attacks from a plethora of surrounding units.
 
The whole sequence is highly suspicious of a blatant departure from reality.

[greykemp@gmail.com]

There is no true randomness in a computer generated die roll. The seed determines the results and the seed is determined by.....usually, the clock, sometimes a number picked from a table. This means that once the seed is determined the entire sequence is no longer random and OFTEN resultes in long strings of numbers within a narrow range. "Good" or "Bad" depending on whose side you are playing.

Unless someone wants to load large numbers of random number tables we either cap the results or live with it.

DigIn, another orginal PG Forum guy.

[soldier1]

Some of the massed unit attacks seem to come off with completely fantastical results as one unit that has suffered air and artillery softening, plus soak-off attacks by ground units, will deal out extensive damage to full strength, experienced attacks from a plethora of surrounding units.

The whole sequence is highly suspicious of a blatant departure from reality.

I don't see too many softened and mass attacked units dealing out fantastic amounts of damage. Unless heavily entrenched they usually go down without causing too much trouble. Of course sometimes they hold out but that's war.
Haven't seen troops on rivers do much but I have seen them in forests hit planes hard and take no casualties. There are some odd results but overall i found the original PG to be more wide ranging and unpredictable.
and if your losing four units to a single troop, then i think its more likely bad tactics than a bad generator.

[dorjun driver]

Any one who considers arithmetical methods of producing random digits is, of course, in a state of sin.

-Johnny V (1951)

On the other hand, Chapter 3 of TAOCP is still a good read.

[SeaMonkey]

Not losing units, just taking abnormally high rates of casualties all the way around from a low strength unit surrounded and absorbing multiple attacks from quality units.

[willgamer]

ORIGINAL: DigIn

There is no true randomness in a computer generated die roll. The seed determines the results and the seed is determined by.....usually, the clock, sometimes a number picked from a table. This means that once the seed is determined the entire sequence is no longer random and OFTEN resultes in long strings of numbers within a narrow range. "Good" or "Bad" depending on whose side you are playing.

I would call it bad regardless. [:-]

If the random method does gen long, narrowly bounded sequences, I hope the devs will take notice and corrective action. [&o]

ORIGINAL: DigIn
Unless someone wants to load large numbers of random number tables we either cap the results or live with it.

DigIn, another orginal PG Forum guy.

Or use a method, or combination of methods, that yields a better algorithm. [:)]

If Dragoon's observation is confirmed, I believe the random generator needs fixing.