Player and Skill Costs: least-squares analysis

Got a great idea and/or proposal for BloodBowl?

Moderator: TFF Mods

Post Reply
fidius
Experienced
Experienced
Posts: 121
Joined: Fri Feb 18, 2011 11:03 pm

Player and Skill Costs: least-squares analysis

Post by fidius »

I recently ran a mostly unrestricted least-squares analysis of all players from the official races against their rookie skill values (just because I was curious if it could be done and if it would have any intuitive appeal), and came up with a very interesting set of results for *implied* skill, stat, access, and misc costs (and therefore player costs). There are limitations simply because in many cases the number of data points is low (eg. Strong Arm appears only on the Treeman so tends to act like a plug figure for std error; there are many such examples). However the nice thing is that, despite its imperfections, it still yielded the following players as "most mispriced", which seems more or less appropriate:

1. Dwarf Runner (overcosted by 9)
2. Blitz-Ra (over by 9)
3. Marauder (under by 7, even with M access valued at 0)
4. Orc Thrower (over by 5)
5. Human & Skaven Throwers (under by 5)
7. Wood Elf Thrower (over by 5)
8. Norse Lineman (under by 5)
9. Orc Lineman (over by 5)
10. Underworld Skaven Thrower (under by 4)

This isn't meant to be critical of Galak's sytem -- obviously there are reasons why a MA6 player should be more valuable on a slow team than on a fast one, for example.

Note that the model also generated odd results, such as this: characteristics (positive or negative) worth 0 value included Passing Access, Mutation Access, Mutation on Doubles, any ST greater than 5 (ie ST6 and ST7 are worth the same as ST5), General on Doubles (neg), and No Access to Apo (neg). Also Leap was valued at 0 -- explained by the fact that Blodge Premium (12) and Very Long Legs (13) tended to distort Leap's value as when it appears it appears only with those skills. I'll have to manually tweak the restrictions to correct for this and see what happens.

Also, interestingly, positive skills yielded an average value of 11, suggesting perhaps skills are valued too highly across the board. How would it change the game, I wonder, if all regular skills cost 10k instead of 20k?

Reason: ''
adhansa
Experienced
Experienced
Posts: 151
Joined: Tue Jan 04, 2011 11:44 am

Re: Player and Skill Costs: least-squares analysis

Post by adhansa »

fidius wrote: Also, interestingly, positive skills yielded an average value of 11, suggesting perhaps skills are valued too highly across the board. How would it change the game, I wonder, if all regular skills cost 10k instead of 20k?
My guess is that skilled-up teams would have a substansial advantage agaist low-leveled ones. When you buy players you get the skills you get, but when you can choose, thats definatly worth something.

Reason: ''
Wulfyn
Emerging Star
Emerging Star
Posts: 323
Joined: Mon May 19, 2014 9:33 pm

Re: Player and Skill Costs: least-squares analysis

Post by Wulfyn »

This is a really interesting analysis - nice to see some mathematical rigor applied. There's definitely like you say some anomolies that can never be hoped to be covered in a simple pricing structure, so it shouldn't be taken as a criticism of either Galak's system or the CRP, and to be fair Galak even covers the need to adjust some prices to achieve balance which I imagine is for exactly that reason.


I wouldn't be surprised to see that the average skill came out well below 20k, because I imagine that the average skill selected is some way sub-par to what you'd pick yourself. I mean Safe Throw on a High Elf Thrower is nice and all, but if they cost 70k instead of 90k and came without that skill I doubt you'd see many people pick it; certainly not in the first 3 choices.

Did you do any analysis on skill combinations? Maybe consider block and dodge as a single blodge skill to see how it compares? Also was there anything regarding the total number of skills a player had? I imagine that any combination starts to have synergy at some point.

(the questions are not to pick on your analysis, because I think it is great, it is more to selfishly try and steer you into doing some more!)

Reason: ''
Post Reply