Great Leaders from Great Battles?

[T-Slyce]

It makes inherent sense that Great Leaders should appear after great battles where one side is greatly outnumbered. As a matter of fact, I lightly assumed this was incorporated into the game, but never bothered to test it. With the advent of patch 1.21, I tried out a game with the random seed generator turned off, and found some rather interesting results:

I took an elite Immortal (yep, Persians) and attacked a regular French warrior jovially sucking his thumb on a plains tile. When the battle was completed, I reloaded and attacked again, continuing until a leader poped up. It required 22 reloads before I obtained my first GL (and my elite Immortal never lost once to the warrior!)

I then used the GL to rush a wonder and took an elite warrior over to the French side. I found a city on a hill guarded by a fortified regular spearman and attacked. It took 6 tries before I even won the battle, but the first time I won, a GL appeared!

To be quite honest, I haven't done any further experiments to see if this is just a coincidence or genuine - the game got a little carried away (after all, this "experiment" dictated I have GLs before 1000 BC, hehehe...) and I figured that this was likely discussed earlier by someone, somewhere out there...

SO, any conjectures? Oh, and another factor: The second GL appeared as I was in my Golden Age (just won with the Immortal, obviously). Does this have any effect?

 

[punkbass2000]

Not to be the Deity of Statistics and Empirical Reasoning (which is all I feel like I'm doing in the Strategy Forum these days), but I would presume it was just coincedence. Some have theorized that it depends on if the unit takes no damage, wins with one hp, etc. None have been verifies, IIRC.

 

[Daaraa]

I read on some thread that the quality of the unit has something to do with it as well.
This seems to hold true (no testing) in my games. It is harder for me (Militaristic civ with Heroic Epic) to get a GL using Cavalry against Spearmen. However, if I use cavalry against riflemen I seem to get them more often. My wife who is quite observant noticed this and told me about it. I tried it and I have been in a war for over 100 turns and I have gotten one GL from Cavalry vs pikemen and 5 GLs from Cavalry vs Riflemen! Throughout the whole war I have had the Heroic Epic.
Again all this is without testing.

pb2000: why not put that line in your sig with a lol smily

 

[punkbass2000]

The problem with some of these improbable notions is that they turn out to be true! I never would have guessed you couldn't get more than one leader from the same unit, and would have argued against it as much as any of these topics until it was proven beyond a shadow of doubt (or, in that case, confirmed by Firaxis).

 

[T-Slyce]

Hmmm...

It would appear that there are really a LOT of variables to check on then. I can, however, personally negate the theory about elite units requiring full health to generate a GL. My warrior had a mere one health unit remaining when the slaughter was through :soldier:

Quite honestly, I agree with you pb2000, that it's likely just coincidence. But it just makes so much sense for a GL to appear after a battle where the unit is hopelessly outnumbered, that I would have thought it imperative to weave it into the game somehow. After all, what makes a great leader IS a great battle (or a series of them), is it not?

Maybe they did something simple. For example, if the odds are 1 to 4 against, chance of GL appearing +15%. 1 to 4 in favor, chance of GL -15%

Unfortunately I have no idea how to test this quantatatively. It would be nice to know when playing a GOTM type game, so you can choose the best attack sequence to max the odds of obtaining a GL. Maybe a Firaxian will pick this up???

 

[Daaraa]

If the elite unit is defending there is a 50% less chance that a GL will appear. That is a statistic that I found interesting. So to increase the chances, attack, attack, attack.

 

[Smoking mirror]

I have been experimenting and I think that there is a small chance of a GL being created if an Elite unit wins a combat. I think that chance is tested for each combat exchange (each HP lost on either side) so if an elite cavalry beats a conscript warrior 5-Hp-2 without losing any HPs that is 2 rounds of combat, twice the very small chance of GL creation. If a cavalry attacks elite riflemen and loses all but one hp before finaly killing the rifleman that is 9 rounds of combat; 9 times the very small chance of GL creation.

This would make some sense; most of my GLs have been created in attacks against other equal units with high combat experience. But then some times its not; yesterday I got my first GL with my elite russian cavalry attacking a regular pikeman out side london, and my second one five turns later attacking regular pikemen in Ireland. I must admit however that I had nearly 30-40 cossak units in my war aginst the english persians and germans, the chance of a GL apearing some where was going to be quite high, with two or three elite attacks every turn, one of them was bound to produce a GL (or two).

 

[eg577]

Im not a statistician, but typicaly when performing random measurements (such as attacking 22 times with the elite immortal) you can expect that what the "actual probabilty" predicts will be somewhere within plus or minus the square root of n of your result, with n being the number of times you performed the test. (If anyone is really good at this stuff please post it here)

One leader out of 22 (victorious) battles predicts about four and a half percent of elite to leader promotions, but how do we measure the accuracy of this precentage? Well, we have n=22, square root of n=4.690 so lets that round to 5. Using this approximation, the actual probabilty of promotion would likely predict somewhere between 1 leader plus or minus 5, which turns out to be -4 and 6 leaders. The reason we got a negative number is because the square root of n, 5, is much greater than the number of the measured occurances, in other words it is a sign telling us that we need to perform the test more than 22 times to get any accuracy. In fact, you can clearly see that 6 out of 22 is a LOT more frequent that 1 out of 22 (a difference in 23%!).

With the warrior attacking the spearmen, there was only one victorious battle, the fact that it was the first battle or that it lost six before it doesn't help us.