I'm not sure if the error is on the "Overrun %" prediction or in the actual combat resolution mechanics, but after noticing how underwhelming my "> 80%" Overrun results were, I kept track of outcomes through an entire German Campaign in Black Turn. When my "Overrun %" prediction chances were either 80% or 90%, here's the results I got:
[*] Infrantry attacks: 1 Overrun out of 6 attacks (if my high school stats math is still holding up, the odds of this happening are < 0.2%)
[*]Armored attacks: 5 Overruns out of 17 attacks
Am I missing something about the way these predictions are calculated? I'm happy to provide save files, etc.
Wonky Math on Overrun
Re: Wonky Math on Overrun
This is very strange, as far as I can understand the code, the overrun results aren't randomized (attacks are!), but I suggest to run another campaign, and compare this. Maybe you were very unlucky

 Cadet
 Posts: 3
 Joined: Sun Feb 28, 2016 4:30 pm
Re: Wonky Math on Overrun
That's what I thought before I started doing my tick sheet, which was about 3 scenarios in. Anybody who bothers to run the probability math here will know a guy doesn't get that UNLUCKY!
I strongly suspect there's different logic being used in the "predict your outcome" screen than in the "resolve your battle" calculations. All it takes is omitting something like weather or terrain shifts in the Overrun combat odds on the Prediction screen to make this happen.
I'll test it again in another campaign one of these days (I'm almost done with Taiphun!), but this doesn't feel like the kind of bug that's unique to my own install. Anybody else have similar disappointing results?
I strongly suspect there's different logic being used in the "predict your outcome" screen than in the "resolve your battle" calculations. All it takes is omitting something like weather or terrain shifts in the Overrun combat odds on the Prediction screen to make this happen.
I'll test it again in another campaign one of these days (I'm almost done with Taiphun!), but this doesn't feel like the kind of bug that's unique to my own install. Anybody else have similar disappointing results?
 Tomislav Uzelac
 2x2 Games
 Posts: 1716
 Joined: Mon Apr 04, 2011 11:24 pm
 Location: Zagreb, Croatia
Re: Wonky Math on Overrun
Hi guys!
The way it works is, you first roll for retreat. You only roll for overrun if the defender retreats.*
So if there's 50% chance of retreat and 80% of overrun, that should work out to an overrun actually happening with about 40% probability.
* Actually, to be precise: you roll for overrun if retreat roll was succesful. Defender may be cornered or destroyed.
The way it works is, you first roll for retreat. You only roll for overrun if the defender retreats.*
So if there's 50% chance of retreat and 80% of overrun, that should work out to an overrun actually happening with about 40% probability.
* Actually, to be precise: you roll for overrun if retreat roll was succesful. Defender may be cornered or destroyed.

 Cadet
 Posts: 3
 Joined: Sun Feb 28, 2016 4:30 pm
Re: Wonky Math on Overrun
That math is helpful, thanks! In every case below where I was looking at Overrun %s 80%, my Retreat % was either 90% or 100%.
So, while there's a big difference in the way the Overrun % plays out in less probable overrun scenarios (and I thank you greatly for that tidbit!), it doesn't change the math much for the kind of outliers I'm looking at. Unless my probability math is screwed up (I don't think it is), the odds of only one overrun in 6 attacks is still < 0.3%.
I'd chock it up to bad math and move on, if the results weren't largely comparable for the larger sample set of 17 attacks with armored results (the probability of 5 Overruns or less looks to be < 0.4%).
I'll start scrutinizing the Prediction %s and see if I can replicate those manually based on the Retreat and Overrun % values given in the manual. Thanks for the advice!
So, while there's a big difference in the way the Overrun % plays out in less probable overrun scenarios (and I thank you greatly for that tidbit!), it doesn't change the math much for the kind of outliers I'm looking at. Unless my probability math is screwed up (I don't think it is), the odds of only one overrun in 6 attacks is still < 0.3%.
I'd chock it up to bad math and move on, if the results weren't largely comparable for the larger sample set of 17 attacks with armored results (the probability of 5 Overruns or less looks to be < 0.4%).
I'll start scrutinizing the Prediction %s and see if I can replicate those manually based on the Retreat and Overrun % values given in the manual. Thanks for the advice!