Level-up luck

[Gandalf]

I've devised a system which gives you a score. If you have perfectly average luck you will end up with 0.

You take each level-up on your team and give it a number, sum up the numbers and there's your score. There has to be an assumption that you chose the "highest quality" choice available - there's no way to know, for example, that you chose a double skill after rolling a double 6, rather than a +ST.

+ST = 68.421 (1 in 36 chance)
+AG = 34.2105 (2 in 36 chance)
+MV/AV = 22.807 (3 in 36 chance - assumes you take MV/AV on a double 5)
double = 13.6842 (4 in 36 chance, ie double 1, double 2, double 3, double 4)
normal skill = -10

How did I get these numbers? Well I started off with giving normal rolls minus 10, then worked the other ones out to be proportional to each other and so that the sum from 36 average rolls would be 0.

This gives the following example scores (worked out manually so might be a bit wrong)
Everglade "what's a double" Wanderers = -181 (the luck they get with injuries is counterbalanced by awful level-ups)
Orkly Ducklingz = -23.5 (1 +ST, 1+AG, 1+MV, 3 doubles and 19 others)
It's A Sign! = -19.2 (becomes 24.9 when you realise that I actually turned down a +AG roll)
Cheesosophers = 14.21 (1 + ST, 2 +AG, 2 doubles ON THE SAME THRALL WHO GOT A +AG!!! rest normal skills)
Bad For Thy Elf = 38.42 (1 +ST, 3 normal skills)

You can suggest alternatives to this system but I may not incorporate them Hope to get this in over the weekend.

[Raveen]

Looks good to me. Thanks!

[lawastooshort]

That doesn't include dead players? - I had a thrall with a double who was retired. I also had a double on a vampire but why would you take it (you wouldn't). Did the +AG thrall really get 2 doubles? Perhaps I should play vampires for next season to catch up on what is going on with them.

The Undercheesers have had 2 doubles, 2 +STs and a +AG.

So where do you get the data from - continually evolving from the current team?

But anyway, I quite like that.

[notjarvis]

Dark Elves have only managed 3 or 4 skill ups so getting + str was pretty lucky, and has made a big difference to the stuff i can do with them (sadly my star ST4 blodging blitzer is injured this weekend, probably means I can afford Morg though...)

My Lizardman team did quite well having 3 +Mvs and quite a few doubles IIRC (The Krox had Block as did a few skinks having Block/wrestle), but no other stat increases.
My old Chaos team did generally far worse with their skill ups

[Gandalf]

That doesn't include dead players? - I had a thrall with a double who was retired. I also had a double on a vampire but why would you take it (you wouldn't). Did the +AG thrall really get 2 doubles? Perhaps I should play vampires for next season to catch up on what is going on with them.

It should include dead/retired players, so long as they are in BBDB. I think your AG player only got 1 double actually (pass), sorry.

So where do you get the data from - continually evolving from the current team?

Yeah... I have a set of procedures/tables that are "bits of data generated from other bits of data" and they get refreshed every time a match is added. There's not much there at the moment, but one thing that is there is star player's stat lines (ie, sum up all matches they have played). So I will write this code and add it in there.

Please bear in mind the example numbers I generated above were from a quick manual look at the BBDB website, not from the actual data so I may have mis-calculated.

I've revised the mapping numbers, it was coming up with unfair numbers in a few scenarios I tried (eg 1 double and 2 normal skills resulted in a negative score). It's not perfect but I think this is better.
+ST 50
+AG 30
+MV/AV 23.333
double 20
normal skill -10

Also, this was Raveen's idea so credit to him too. Felt it worthy of its own thread in case people go "how do you work out those numbers?" - I can point them here.

[Gandalf]

score team id team name skill count
46.6666 2 Team Shrek 16
20.0000 14 Bad for Thy Elf 4
19.9999 18 Saur'd Of Justice 20
16.6666 4 It's A Sign! 12
13.3333 11 Hambridge Archers 2
10.0000 13 A Fistful of Boomstick 3
-20.0000 1 The Mighty UnderCheesers 24
-26.6667 15 Orkly Ducklingz 25
-26.6667 8 Azgorh Rock-Knockers 12
-30.0000 7 The Cheeseosophers 20
-40.0000 10 Evil Lacerating Orcestra 7
-40.0000 6 Club Twenty Seven 16
-56.6667 16 Isa Black Stars 15
-56.6667 17 Fables! 19
-60.0000 5 The 13th Doomwheel 19
-60.0001 3 The Brave Harts 22
-76.6667 20 Cangurus 21
-90.0000 9 Gojira's Army 9
-116.6667 19 Pinewood Carrions 21
-176.6667 12 Everglade Wanderers 35
*******************************
observations....
team shrek are a bit inflated as I fire any ogre whose first 2 level ups are normal rolls, and those ogres aren't in BBDB. Though I fully admit ag4 ogre is hilariously unlikely.
not suprised to see saur'd up there.
Vindication for Rav & myself, who complain most about level ups, for we are at the bottom
overall scores are a bit negative? Can this be explained only by people not choosing the "better" skill, like la's vampire? Maybe apply some scaling upwards based on # of skills to account for this?

[lawastooshort]

Those jammy jammy halflings.

[Raveen]

Yep, fully half of my level ups with that team are +MV.

Interestingly the Rock Knockers would be positive if I had taken a second double on one of the bulls.

[lawastooshort]

Not choosing the "better" skill doesn't happen that often, really. Vampires are a fairly special case. Perhaps AV isn't taken often (I can't think of anything I'd take it on) but then you can choose MA.

I'd like to see - in all of the luck calculations - a statistical significance rating of some kind. How bad does your luck have to appear before you can actually call it bad justifiably? We're not really talking about large numbers of dice being rolled.

[lawastooshort]

Oh yes and as Rav hints and I had considered, there's quite a few players you'd probably decline a second double on.

[Gandalf]

Not choosing the "better" skill doesn't happen that often, really. Vampires are a fairly special case. Perhaps AV isn't taken often (I can't think of anything I'd take it on) but then you can choose MA.

I'd like to see - in all of the luck calculations - a statistical significance rating of some kind. How bad does your luck have to appear before you can actually call it bad justifiably? We're not really talking about large numbers of dice being rolled.

Well in this one, it's a sum of all your dice roll scores, so it has built-in statistical significance. Hence for example Saur'd of Justice are ahead of Hambridge Archers as they have had good luck over a larger number of rolls. Albeit it's not the most scientific of methodology... it seems broadly correct to me - IMO it correctly shows the Wanderers as having, by a long way, the worst luck level-ups, show me any other team who comes close... it was even worse, after 25 or so level-ups I think I had 1 double 1 +MV and 23 normals.

I am going to use proper statistical equations on some of the other dice rolls - it's easier for in-game rolls as there are clear success/failure criteria, there are very few rolls that aren't one or the other (a pass being accurate pass/inaccurate pass/fumble being the only one I can think of). The difficulty with applying proper statistics to level-ups is that there isn't such a thing as a "successful roll".

More examples of not using "the best roll" - I gave a line elf a normal skill rather than MV/AV. The team is fast enough as it is and the AV won't help much. I turned down +AG on a rotter as I already have one +AG and I was worried about TV. I *suspect* most veteran teams will have turned down "the best roll" at least once... so I will apply some sort of scaling factor. Maybe only for teams who have at least 10 skill-ups.

[lawastooshort]

Oh, yes, this is probably one of the least appropriate things to ask for a statistical significance measurement on, but it's been generally bugging me for a while. That said, for the most part I have stopped looking at the luck ratings in any case.

[DaigaroOgami]

Vindication for myself, who complain most about level ups, for we are at the bottom

Dude... Can't you interpret your own data?

you coach two of the teams in the top 4!

I feel vindicated sunshine, you and Nuffle's favourite son are the jammy gits!

[lawastooshort]

Well, 28 level ups in the two top teams vs 35 in the bottom one team, probably pretty even. Whereas I have 44 level ups in 2 teams both below 0 - I am pretty much the unluckiest coach in the league

[DaigaroOgami]

The Ducklingz and the Black Stars are you below you squire, leave the self-pity to those who need it... add this to 'who's had the worst injuries to the key players on their teams' and we'll see who's been having the bad luck eh??