Odds prediction is seriously screwed in late years

[Plaid]

Well, here goes picture.
Need little comments, but its just the brightest example.



(USA tank effectiveness in 71, you can check with right click on picture -> open picture)
But generally in late years (past ~1942) odds become wilds and have little in common with reality.
Almost any attack against high effectiveness entrenched unit gives you odds like 2:8 1:10 etc in defender favour, while in fact result is usually much more acceptable (like 2:4 3:3 etc). In the given example (0:14 in USA favour) actual result was 3:2 in german favour.

Is it only me, or something is wrong here?

[gchristie]

I see this frequently on the Eastern Front and I have to admit the odds presented psyches me out and makes me much more cautious.

[ncali]

It's because it's not really a combat result calculator. It doesn't run the combat to see what would happen - it compares the adjusted strengths of the units rather than estimating losses by going through the combat sequence. The sequence for land combat - suppression, defending unit fires, attacking unit fires is critical. This is why I think the calculator tends to be more accurate as giving you an idea of losses in air to air combat, which is simultaneous (so the adjusted AA strengths of the units is all that really matters).

[Cybvep]

I noticed that, too. Very annoying.

[Peter Stauffenberg]

We havent changed the odds calculation code in CeaW except actually taking into account combat sequence. There are some strange issues with the vanilla game odds calcuation. E. g. I used to see odds like 50:1 when late game armor attacked enemy armor. That's impossible when the max damage you can have is 14. Usually the result you got was 10:1 or something.

There is nothing in the odds calcuation code that indicates why this would be happening, but I believe it's happening when either the attacker or the defender gets a very high antitank value. That means the game would see the enemy survivability as lower than it actually is. E. g. if the antitank is as high as the enemy survivability then the effective survivability would become 0 and all attacks would score a hit. But somewhere in the code I think there a limit to the real survivability so it can't drop below 2.

[ncali]

For some future version of GS, it might be nice to turn the odds calculator into a true combat results calculator. In other words, it would run a certain number of sample combats and then average the results in terms of the expected losses for each side.

[Plaid]

Would be great to see it fixed.
Looks like what was called "extensive roll randomness" when combat result is very different from the odds is actually some flaw in the very mechanics how odds are calculated.

With the current mechanics I use to load savegame and go attacking with different unit types vs different unit types, then close a game. I repeat it several times before any major operation in later years, to get some reliable data about how relatively strong units are. It seems slightly unfair maybe, but I deffinetely cant plan anything with this 2:20 odds showed.

[Diplomaticus]

I don't worry about odds. I just keep quitting and reloading the game until I get good *results.* If you run the same battles 20-30 times, sooner or later you get really awesome luck! And the best thing about it is that your opponent thinks, "Wow, this guy's like the best general ever!"

[Kragdob]

I don't worry about odds. I just keep quitting and reloading the game until I get good *results.* If you run the same battles 20-30 times, sooner or later you get really awesome luck! And the best thing about it is that your opponent thinks, "Wow, this guy's like the best general ever!"

I like your tactics! I use it from now on

[gchristie]

I don't worry about odds. I just keep quitting and reloading the game until I get good *results.* If you run the same battles 20-30 times, sooner or later you get really awesome luck! And the best thing about it is that your opponent thinks, "Wow, this guy's like the best general ever!"

-_-

Speaking of future fixes for this game...

[BuddyGrant]

I don't worry about odds. I just keep quitting and reloading the game until I get good *results.* If you run the same battles 20-30 times, sooner or later you get really awesome luck! And the best thing about it is that your opponent thinks, "Wow, this guy's like the best general ever!"

Haha . Yeah, it would be great if there were some game mechanism to stop saved game reloads in human versus human play. Even a counter to indicate how often your opponent loaded the save turn. Something like that would be possible based on the load game action combined with the time stamp of the saved game in question. The only workaround to that style of saved game security would be to run the game on multiple PC's, and even that would be a big limiting factor to turn reloads.

[rkr1958]

I don't worry about odds. I just keep quitting and reloading the game until I get good *results.* If you run the same battles 20-30 times, sooner or later you get really awesome luck! And the best thing about it is that your opponent thinks, "Wow, this guy's like the best general ever!"

Haha . Yeah, it would be great if there were some game mechanism to stop saved game reloads in human versus human play. Even a counter to indicate how often your opponent loaded the save turn. Something like that would be possible based on the load game action combined with the time stamp of the saved game in question. The only workaround to that style of saved game security would be to run the game on multiple PC's, and even that would be a big limiting factor to turn reloads.

The only way I see that you could have a timestamp or counter is if you had to load your turn directly from a server (e.g., Battlefield Academy, Panzer Corps). With game turns that are emailed all one has to do is save the attached game turn from the email again back into your game directory.

Well, a thought just did occur to me. One thing that might could be done for PBEM games is that an email would be sent out to your opponent when you loaded a game turn. This feature would have to be robust enough to allow the game turn to be opened only after the email was successfully sent out. This would mean that each PBEM game would have to store each player's email address. If we did this right, then when you finished your turn then GS would automatically email your turn out.

[richardsd]

I don't worry about odds. I just keep quitting and reloading the game until I get good *results.* If you run the same battles 20-30 times, sooner or later you get really awesome luck! And the best thing about it is that your opponent thinks, "Wow, this guy's like the best general ever!"

Haha . Yeah, it would be great if there were some game mechanism to stop saved game reloads in human versus human play. Even a counter to indicate how often your opponent loaded the save turn. Something like that would be possible based on the load game action combined with the time stamp of the saved game in question. The only workaround to that style of saved game security would be to run the game on multiple PC's, and even that would be a big limiting factor to turn reloads.

whilst this could potentially be an issue, is it really? I don't know how many times I have been frustrated by the 1 remaining step but do people really reload for better combat results?

and how do you then cope with the constant danger of MSoft tecnology and children!

I think you have to have a level of trust, as for example I only use one password and have given that to Staffenburg for debug purposes and its the password I use when I play him (maybe thats why he is better than me , no he just is better than me!)