PDA

View Full Version : Battle - archery tests (Bow Ashigaru vs unit stances)



Tamur
04-21-2011, 21:36
therother was kind enough to meet me on the multiplayer field of battle and let me shoot his poor Yari Ashigaru to pieces.

One thing we can conclusively state from this test is that 3000 corpses do not pile up over time.

After that morbid start, we can only go uphill! In a metaphorical sense, not in a totalwar sense because that would give us combat penalties. OK, research, right!

Hypothesis:

We had three test hypotheses:

1. Units in spearwall formation take less damage from arrows (shot frontally)
2. Units in loose formation take less damage from arrows (again shot frontally)
3. Samurai units of the same type as Ashigaru will take less damage from arrows

Method:

therother brought 10 units of Yari Ashigaru and 3 units of Yari Samurai to the battle. I brought 15 units of Bow Ashigaru.

It is hard to find a level spot on the two maps we tried, but we did manage to find a fairly flat spot covering about 100 yards. therother marched units to the same spot each time, and I replaced archers in the same spot, as they ran out of ammunition.

therother placed his General so that the unit being shot at was within the General's influence ring. This was not because he was nice -- it was so we would have more data to work with.

Finally, we ran between three and five test "mini-battles" per unit configuration, so the figures below reflect standardised averages over those tests.

Results:

Our results showed a fair bit of randomness, but some simple conclusions emerge. The following graph shows kills per volley.

http://cicero.modwest.com/images/tw/archer_results_02.png

Discussion:

Our hypothesis about the Spearwall stance was incorrect. They suffered the same loss rate as the non-stanced Ashigaru. The one difference was that Spearwall formation led to the unit routing an average of two rounds later than non-stanced Ashigaru.

Our hypothesis about units in Loose formation seems correct. They suffered an overall lower kill rate than either normal or spearwall formation Ashigaru. They also routed at a later time (2 rounds longer than Spearwall, 4 longer than non-stanced).

Finally, our hypothesis about Yari Samurai proved correct. They suffered the lowest casualties of all unit configurations tested. Not only that, but the standard deviation on kills was significantly lower than any other unit.

Conclusions:

1. If the enemy have archers, and you will be holding a position with standing troops, put them in loose formation. While this is one of the basic premises of command in TW, it bears repeating. This may be especially useful at bonus buildings on multiplayer maps.

2. Spearwall formation leads to higher morale and longer time on the battlefield, even when facing arrows. If you have peasant units, it may be worth it to put them in this formation at all times when not on the move.

3. The higher cost of Samurai units seems to be very well worth it. In some of our tests, the Samurai suffered half the kill rate of the Ashigaru.

Further Research:

1. The longer staying power of Spearwall formation needs to be studied in more detail. It could be that we simply had lower kill rates those last few rounds, and therefore the troops didn't have the "fast kills going on" morale penalty.

2. The stable kill rates with Yari Samurai deserves more attention. Does this hold up in an infantry vs infantry battle?


OK, that's enough text for one day!

gauch0
04-22-2011, 02:54
I would like to know if archery units in loose formation have a lower kill rate than archery units in standard formation. (Does spreading out the archers make them less deadly?)

therother
04-22-2011, 12:20
Nice write up, thanks Tamur!

We did do a small number of tests on number of archer ranks. Not enough draw to draw any firm conclusion, but number of ranks didn't seem to have a major impact. We'll add archers in loose formation to the list of things to test.

chrisj
04-25-2011, 15:54
I would like to know if archery units in loose formation have a lower kill rate than archery units in standard formation. (Does spreading out the archers make them less deadly?)

This is what is really interesting.

Tamur
04-25-2011, 16:38
Based on data we got from another round of testing this past weekend, it looks like archer formation (both number of ranks and loose/normal stance) makes no difference in kill rates.

Data:

We tested three different conditions, averaging multiple runs with each configuration. Each archer unit fired at a single unit of Yari Ashigaru, 5 ranks deep, in normal formation.

Condition 1: Archers in 2 ranks, normal stance
Condition 2: Archers in 7 ranks, normal stance
Condition 3: Archers in 3 ranks, loose stance

I love graphs, so here's another one showing results:

http://cicero.modwest.com/images/tw/archer_results_03.png

(X axis is volleys, Y axis is kills, so any one point represents the number of men killed in a single volley)

As you can see, there is a slight variation in the kill rates but not much --- certainly not enough to say there's a significant difference in kill rates with any of these configurations.

7-rank formation seemed to have a slight edge, or at least seemed more consistent.

However, that could be down to experimental anomaly rather than real finding. Not enough data to say for sure. Someone else doing this test (requires multiplayer) would help verify the result.

gauch0
04-25-2011, 19:59
Wow, this is great! It looks putting your archers in loose formation during sieges (on attack, and on defense when there isn't room at the walls) is a no-brainer.

frogbeastegg
04-26-2011, 20:16
Interesting stuff, well done gentlemen. :bow:

Are you intending to test the effect of formation depth on the spearmen? If the missile engine works as it did in the first two TW games then a deeper formation should mean more casualties due to the higher chance of a missed arrow hitting one of the men behind the target. Recent TW games seem to have the arrows behave more as the result of a calculation, leading to a kind of homing effect, so misses may not be as free to accidentally hit.

Tamur
04-27-2011, 06:02
That would be a logical next step.

therother led me astray yesterday by starting in on a study of elevation effects with archers... which turned into an ugly cav/sword melee when our control group results went all wonky. We'll finish that up hopefully soon, and move into target rank depth.

Rothe
04-27-2011, 06:55
Could you also test what is the "splash" effect of the arrows. Lets say you order fire on one unit, and then you move another unit to the same space (interlaced formations) - would both take identical casualties from subsequent volleys, or does high accuracy really mean more hits on the actual target unit?

neoiq5719
04-28-2011, 22:05
Let´s see cos to me that´s a wierd representation and if u like graphs then u should label them properly. Could u label the x and y axis so we know what is what? 5,10,20 etc what is that anyway? secondly what does the x axis represent? thirdly why there are lines going up? if we are testing the strenght of a formation the best the graph could do is level off but not go up or maybe I´m mistaken cos i dont know what the axis mean?

Tamur
04-29-2011, 02:59
neoiq5719, sorry for not labeling the axes -- I guess I'd been looking at them so long that they just made sense to me. I'll get that done with some more data additions tomorrow.

For now though, I can at least tell you that the Y axis is kills, and the X axis is volleys. So if you look at the lower graph's yellow line, the 3-rank loose formation yielded on average a bit under 16 kills for the first volley, a bit over 17 kills for the second volley, and so on.

neoiq5719
04-29-2011, 06:47
Then all u have to do next is to break down the x axis into the number of volleys which according to the graph are 8 right? But still I would represent on the y axis not the number of kills but the percentage of kills and do the same experiment with other unit sizes to see if they take the same kind of punishment. And also I still dont get how the graph goes up if like u said the y axis is the number of kills, as u lose men it should go down right? cos if u represent the men killed then it should go up all the time and if u represent men remaining then it cant go up.

Ituralde
05-02-2011, 16:21
It shows the men killed per volley. Sometimes more were killed in one volley sometimes less. At least that's what I understood. Nice testing it's good to have 'common sense' backed up by data!

Tamur
05-02-2011, 17:34
Ituralde is right, it is men killed (Y axis) per volley (X axis), rather than total men killed over time. I'll make sure to be much clearer in the future -- my apologies for the confusion.