|
Post by arnopov on Feb 5, 2020 15:08:47 GMT
Careful here, the Bd can still kill the Kn the old-fashioned way, by doubling them (2/36 at base 3v3). So that gives 15/36 kill for the kn, and 8/36 kill for the Bd. For a quick and dirty "first order" estimation (very loose use of the term here, don't go all Cauchy on me on that), that leads to Kn win 65%, Bd win 35%. Contrast this with the (perceived at this stage) outcome of Bd vs Bw (I expect Bd to win very bigly, 95%??), AND Kn vs Bw (I expect Bw to win big as well -not as much as BdvBw, but more than KnvBd- 80%??), and you can see that in our game of probabilitic rock/paper/scissor (Kn/Bw/Bd), there is a shift to stronger scissors. And without even taking terrain into account, which favours 3Bd and 3Bw even more.
I'll try to provide sounder numbers later, but it's already pretty clear cut to me (and others in the tourney scene)
|
|
|
Post by macbeth on Feb 6, 2020 2:09:58 GMT
Can’t fault the maths there Macbeth. In a Kn vs Bd fight it is indeed 42% vs 16% in GG with no overlaps. I know you previously posted the details of how you calculated your handicap system several months ago but I’m useless at being able to find it - even though I remember reposting the link on another thread. Could you repost it. In particular does it take account of how troops move? I ask this because one of the key advantages of 3Bd is their 3BW move and ability to trot over RG with impunity while the Kn are forced to skirt around. Hi Paddy,
the basic details about how I calculated my rating scores is given in the Magister Militum Per Capitoline Territorialis thread - the one below this one at the time of writing.
I did take into account the movement rates - The sum of Good Going and Bad Going movement rates was added to the base score - so a Kn element would have +4 added but a Fast Element would have +6 added
Put additional bonuses in for multiple moves, not recoiling, having to pay more than 1 PIP, and shooting range as well.
Cheers
|
|
|
Post by macbeth on Feb 6, 2020 2:27:29 GMT
Careful here, the Bd can still kill the Kn the old-fashioned way, by doubling them (2/36 at base 3v3). So that gives 15/36 kill for the kn, and 8/36 kill for the Bd. For a quick and dirty "first order" estimation (very loose use of the term here, don't go all Cauchy on me on that), that leads to Kn win 65%, Bd win 35%. Contrast this with the (perceived at this stage) outcome of Bd vs Bw (I expect Bd to win very bigly, 95%??), AND Kn vs Bw (I expect Bw to win big as well -not as much as BdvBw, but more than KnvBd- 80%??), and you can see that in our game of probabilitic rock/paper/scissor (Kn/Bw/Bd), there is a shift to stronger scissors. And without even taking terrain into account, which favours 3Bd and 3Bw even more. I'll try to provide sounder numbers later, but it's already pretty clear cut to me (and others in the tourney scene) My apologies for leaving off the extra 5.5% chance for the Bd to kill by doubling. My bad - but it still close to a 2:1 ratio of kills in favour of the Kn
Worth noting that Kn kill Bd - 15/36 Bd kill Kn - 8/36 Bd recoil Kn 13/36 - and after this happens the odds shift in a line so that the overlapped Kn has only 10/36 of a kill with the Bd getting up to 11/36 in return.
And you are right Arnopov - as a member of Oz's tourney scene I am fully aware of the power of fast Bd (after all I took Tamils and Siamese to Cancon) BUT
I believe Kn have the upper hand, and if we nobble the Bd without doing something to the Kn then we further cement their dominance. At which point we need to fix each and every element
Someone on this board once said that DBA is like knitting - Very Hard to knit the Jersey but easy to pull at a couple of loose threads and unravel the whole thing.
My element rating system has some way to minimising the effect of the super troops. We have a couple of much sought after prizes at my tournaments awarded to the highest "Adjusted" score where each round's score is adjusted by the ratio of Opponent's Rating/Player's rating. So when My Tamils - rated approx. 500 took on the Sumerian Great Revolt approx. rating 350 my score was adjusted down by a factor of 0.70 whilst my opponent was adjusted up by a factor of 1.43. Quite embarrassing when he beat me.
The Magister Militum Prize (or "Maurice") is handed out at the three competitions that I run (Collision Course/Wintercon/Landwaster). At Cancon we hand out the "Grand Maurice" which looks at a scaled average of the best three results across the 5 ACT tournaments (my three plus the two days of Cancon).
A number of the better players have chosen armies that might have a lower rating.
There is more detail about my rating system and the Magister Militum prize in the Magister Militum Per Capitoline Territorialis thread on this board (the one below this one at the time of writing).
Cheers
|
|
|
Post by paddy649 on Feb 6, 2020 7:37:42 GMT
the basic details about how I calculated my rating scores is given in the Magister Militum Per Capitoline Territorialis thread - the one below this one at the time of writing. Thanks Macbeth - I found the thread - but did you intend to send a link to the updated version. You have obviously put a lot of work, thought and effort into this. It looks great and is certainly best of breed at the moment. What I am really interested in is getting a better understanding of the algorithm used to get to the values and not the values them selves. It is by understanding that algorithm that we can see what respective values it places on different troop abilities. For example you include movement rates but also the fact that fast are recoiled by solid - is it this that places 4Bd on 43 and 3Bd on 40? My current view of the U.K. competition scene is that this may overstate the combat results and underestimate the movement flexibility. I would have expected the rankings to be reversed. On the face of it what I think you have produced here is a weighted ability system for troop types. So if troop type A scores 40 and troop type B scores 20 then on average across all DBA players in an open tournament then Troop type A is twice as effective as troop type B. Does this system imply that troop type A to appear twice as often in army lists selected by competitors as troop type B? If not should it? Surely the ultimate discriminator here is troop selection frequency. For example under the ability system Bw come in at 23/24 points while Ax score an impressive 31/34 points. Yet just looking at the army lists and photos from the Welsh open 3Bw appear with a far higher frequency than 4Bw and Ax are almost non existent. Similarly LH at 28 points appear far less frequently than the ubiquitous 3Bd (possibly a 5:1 or 6:1) despite the points system implying an anticipated 2:1 ratio. ......or am I over-thinking this?
|
|
|
Post by Baldie on Feb 6, 2020 18:03:54 GMT
the basic details about how I calculated my rating scores is given in the Magister Militum Per Capitoline Territorialis thread - the one below this one at the time of writing. Thanks Macbeth - I found the thread - but did you intend to send a link to the updated version. You have obviously put a lot of work, thought and effort into this. It looks great and is certainly best of breed at the moment. What I am really interested in is getting a better understanding of the algorithm used to get to the values and not the values them selves. It is by understanding that algorithm that we can see what respective values it places on different troop abilities. For example you include movement rates but also the fact that fast are recoiled by solid - is it this that places 4Bd on 43 and 3Bd on 40? My current view of the U.K. competition scene is that this may overstate the combat results and underestimate the movement flexibility. I would have expected the rankings to be reversed. On the face of it what I think you have produced here is a weighted ability system for troop types. So if troop type A scores 40 and troop type B scores 20 then on average across all DBA players in an open tournament then Troop type A is twice as effective as troop type B. Does this system imply that troop type A to appear twice as often in army lists selected by competitors as troop type B? If not should it? Surely the ultimate discriminator here is troop selection frequency. For example under the ability system Bw come in at 23/24 points while Ax score an impressive 31/34 points. Yet just looking at the army lists and photos from the Welsh open 3Bw appear with a far higher frequency than 4Bw and Ax are almost non existent. Similarly LH at 28 points appear far less frequently than the ubiquitous 3Bd (possibly a 5:1 or 6:1) despite the points system implying an anticipated 2:1 ratio. ......or am I over-thinking this? Yes
|
|
|
Post by macbeth on Feb 6, 2020 22:42:53 GMT
Hi Folks,
unlike the simple algorithm for my original rating system that was published in the old Fanaticus resources page (and in fact had its own long thread on the original Fanaticus BB), the core of this one was a very complex and large spreadsheet that I cannot collapse onto a page.
The basis was to work out the likely chance in 36 that each element could destroy each other element and then the chance in 36 that each element would be destroyed by each other element. This chance was scaled by the maximum frequency of each element occurring (so being able to kill SCh (there can be only one) is not as big a deal as being able to kill 3Bd (where there can be 12). The next step was to convert this to a percentage by dividing the sum of these chances by the total combination (can't access my personal stuff here so can't give the actual number).
Then I make minor tweaks based on movement, shooting, pip cost, rear/side support, loss effect and recoil effect. Recoil effect is included because if the chance of being recoiled increases, the chance of being destroyed (because you are flanked) goes up. Not recoiling means you don't die from being flanked.
It is a measure of how effective each element is against all other element types, but it does assume an even distribution and does not take into account things like * The environment - in a themed event where certain troop types are banned, elements that kill them are over-valued * Player Choice - certain troop types are liked better by players or are perceived to be better (Fast Bd are talked up atm but some of the up talking is (the qk of Kn on a tie) is over-hyping). So there are more of these on the table than others this biases the scoring. * Interactions - Solid Ax are pound for pound better than Solid Bw, but because the Bw do better against Kn, they are more likely to be chosen for a tournament army.
And of course there are the Doug Melvilles, and David Kujits of the world who can demolish most other players with whatever army they choose and my effectiveness system does not take the calibre of the general commanding into account.
However it does provide a basis for adjusting scores, and has worked well here in Oz. It mitigates the use of Dismounting troops where I rate that element pairing as the sum of both elements.
I think it also works here because our scoring system differs from elsewhere
8 points for a win - flat rate whether you spent the whole game wiping each other out or an El(gen) fell to Art fire in bound 1 and then two elements of stranded Ps were ridded down.
For the other side 1 point for each element killed (Max 3) - here an element is an element, a 6Kn(gen) or a 7Hd both contribute 1 point to the score 1 point for killing a general 1 point for taking a camp
and 1 point for finishing the game.
So it is possible to score 6 points for a loss - but I have yet to see the game where a player killed 3 elements including the general AND took the camp BUT still lost.
A draw will range from 0 to 3 in general with the possibility of a 4, or 5
This flattens out the top half of the table, whilst putting plenty of variance in the bottom half, and it encourages games that end with casualties.
I remember some time ago there was a player who liked to lay down a terrain fortress then prevaricate through the round, until close to the end of time when he would spring out and try to kill 1 or 2 exposed elements. I am told he said that the scoring system was "biased against his style of play"
(I am tempted to say "Yes! Yes it is!"
Cheeers
|
|
|
Post by arnopov on Feb 7, 2020 14:38:16 GMT
It's all a bit of fun, especially fort those with more geeky inclinations. But if I was a reviewer for "Annals of the Royal Tetrapilectomy Society A: Wargames Analysis and MathHammer", I would point out that the Macbeth's model has a gross built-in positive bias towards elements that do well against foot in GGo (23 foot vs 7 Mtd skews the averaging),an unrealistic emphasis towards CC in GGo, and a very unbalanced weighting for shooting. It wouldn't be too hard to address these.
In fact, I think that at the minute, Simon's model (sum of combat factors -El 9 ... Ps 4-) is less misleading, without a veneer of stat.
But it doesn't matter, organisers do what they want, their absolute prerogative.
And it rises some interesting questions, about hard stats, but also perception. What's the proportion of element lost due to shooting? Lost in CC in DGo, etc... Good stuff!
|
|
|
Post by Baldie on Feb 7, 2020 18:46:53 GMT
Does anyone try fighting a game as a defender with one force then swapping to be the attacker with your opponents force or fighting two equally matched pairs or the same figs against each other. I am pretty certain that unless I got some spectacular roles there are still masses of player including many on this illustrious forum that would still duff me up by simply out playing me.
|
|
|
Post by colinthehittite on Feb 8, 2020 8:51:52 GMT
Tetrapilectomy
Well, thank you for teaching me a new word, Arnaud… and such a fun one! It might be useful one DBAday.
Colin
|
|
|
Post by diades on Feb 9, 2020 10:17:21 GMT
Tetrapilectomy Well, thank you for teaching me a new word, Arnaud… and such a fun one! It might be useful one DBAday. Colin Heartily seconded! I also run a points system for seeding the armies at the Tarrington Tourney. It takes account of the specific elements present on the day. It makes assumptions about the frequency of support and distance combat and terrain that are undoubtedly imperfect.
|
|
|
Post by scottrussell on Feb 9, 2020 19:09:55 GMT
Tetrapilectomy Well, thank you for teaching me a new word, Arnaud… and such a fun one! It might be useful one DBAday. Colin In what context is it being used here? In medicine/surgery it would mean the removal (-ectomy) of four (tetra) hairs (pilo). Scott
|
|
|
Post by menacussecundus on Feb 9, 2020 19:12:29 GMT
Tetrapilectomy Well, thank you for teaching me a new word, Arnaud… and such a fun one! It might be useful one DBAday. Colin In what context is it being used here? In medicine/surgery it would mean the removal (-ectomy) of four (tetra) hairs (pilo). Scott Not sure I can take that much of a risk, Scott.
|
|
|
Post by martin on Feb 9, 2020 19:14:37 GMT
Tetrapilectomy Well, thank you for teaching me a new word, Arnaud… and such a fun one! It might be useful one DBAday. Colin In what context is it being used here? In medicine/surgery it would mean the removal (-ectomy) of four (tetra) hairs (pilo). Scott Scott, would it mean ‘splitting hairs’ (??)...never seen the word before, meself.
|
|
|
Post by colinthehittite on Feb 9, 2020 20:24:08 GMT
The art of splitting hairs four ways - I assumed!
Colin
|
|
|
Post by Baldie on Feb 9, 2020 20:56:17 GMT
Keeping out of this
|
|