Post by hodsopa on Feb 26, 2017 12:00:28 GMT
I thought it would be interesting to compare troop types not on the basis of how often that particular type of element kills and gets killed (which is what I tried first), but rather on how often they are part of a winning or losing army, and by what margin of victory. In this way the valiant enemy psiloi who (too often for comfort) charge into the bad going and immobilise half my mounted force, without ever actually fighting, can also get recognition.
I have collected data on 23 competitive battles and 68 solo games (is it embarrassing to play so many solo?). To describe the margin of victory, if the loser scored no kills, I valued the battle at 4 points, however many kills were scored by the winner. Otherwise, I reduced its value according to the ratio between the winner's and the loser's kills. (For example, a game in which the winner scored 5 kills and the loser scored 4 kills was valued at (4x1/5) = 0.8 points.)
I counted an extra kill for a general or the first double element, and 0 points for Hd etc.
(On reflection I am not convinced that this is the best way to describe the margin of victory; and maybe it might anyway be better to count all wins as +1 and all losses as -1, whatever the margin; comments welcome.)
I then awarded the value of each battle as a "plus" to each element on the winning side and a "minus" to each element on the losing side. So in a 4-point battle with 3 Cv on the winning side and no Cv on the losing side, the total score of the "Cv" troop type would increase by 12. In a 0.8 point battle with 2 Ps on the winning side and 3 Ps on the losing side, the total score of the "Ps" troop type would decrease by a net 0.8.
To compare the troop types, I took all those for which I had records for at least 30 battle participants. I divided the total score for each of these troop types by the total number of battle participants. I did the same for the subset of competitive battles. The results are shown in the table.
Comments:
1) Obviously this is a work in progress. The sample, though it feels large, is too small to give reliable results.
2) Still, it confirms my broad impression that 3Bd are the best troop type (I haven't used El much) and LH are woefully underpowered.
3) The good performance of Cm and 3Wb comes from the fun I have had, both solo and in competitive games, with the Tuareg, nomadic pre-Islamic arabs and Welsh. I wonder if the bad performance of the 4Bd comes from the successes of the Welsh against Viking-type armies (where I had a lot of luck in the last PAWS of the year).
4) The better performance of the Bw and Lb in competitive games than solo ones is because others use them better than me!
I will keep this up and would welcome comments.
The numbers can roughly be interpreted as meaning "If you had an army of 12 4Pk, you could expect to score 0.9 more kills, by the end of the game, than the average army". But obviously this ignores the need for balance in armies; and the way I calculated victory margins mean that this description is anyway only approximate.
Paul H
I have collected data on 23 competitive battles and 68 solo games (is it embarrassing to play so many solo?). To describe the margin of victory, if the loser scored no kills, I valued the battle at 4 points, however many kills were scored by the winner. Otherwise, I reduced its value according to the ratio between the winner's and the loser's kills. (For example, a game in which the winner scored 5 kills and the loser scored 4 kills was valued at (4x1/5) = 0.8 points.)
I counted an extra kill for a general or the first double element, and 0 points for Hd etc.
(On reflection I am not convinced that this is the best way to describe the margin of victory; and maybe it might anyway be better to count all wins as +1 and all losses as -1, whatever the margin; comments welcome.)
I then awarded the value of each battle as a "plus" to each element on the winning side and a "minus" to each element on the losing side. So in a 4-point battle with 3 Cv on the winning side and no Cv on the losing side, the total score of the "Cv" troop type would increase by 12. In a 0.8 point battle with 2 Ps on the winning side and 3 Ps on the losing side, the total score of the "Ps" troop type would decrease by a net 0.8.
To compare the troop types, I took all those for which I had records for at least 30 battle participants. I divided the total score for each of these troop types by the total number of battle participants. I did the same for the subset of competitive battles. The results are shown in the table.
| all games (91) | competitive games only (23) | |
| Cm | 1.0 | 0.6 |
| 4Pk | 0.9 | n.a. (not enough examples) |
| 3Bd | 0.6 | n.a. |
| Sp | 0.5 | -0.6 |
| 3Ax | 0.4 | n.a. |
| 3Pk | 0.2 | n.a. |
| 3Wb | 0.1 | 0.5 |
| 3Kn | 0.0 | 0.3 |
| Cv | 0.0 | 0.1 |
| Ps | 0.0 | -0.5 |
| LH | -0.3 | -0.7 |
| 4Lb | -0.3 | n.a. |
| 3Bw | -0.4 | 0.1 |
| 4Bd | -0.8 | -0.3 |
| LCm | -1.0 | n.a. |
Comments:
1) Obviously this is a work in progress. The sample, though it feels large, is too small to give reliable results.
2) Still, it confirms my broad impression that 3Bd are the best troop type (I haven't used El much) and LH are woefully underpowered.
3) The good performance of Cm and 3Wb comes from the fun I have had, both solo and in competitive games, with the Tuareg, nomadic pre-Islamic arabs and Welsh. I wonder if the bad performance of the 4Bd comes from the successes of the Welsh against Viking-type armies (where I had a lot of luck in the last PAWS of the year).
4) The better performance of the Bw and Lb in competitive games than solo ones is because others use them better than me!
I will keep this up and would welcome comments.
The numbers can roughly be interpreted as meaning "If you had an army of 12 4Pk, you could expect to score 0.9 more kills, by the end of the game, than the average army". But obviously this ignores the need for balance in armies; and the way I calculated victory margins mean that this description is anyway only approximate.
Paul H
