Friday 13 December 2013

Forge Father Rules Continued now with added STATS

The results are in! Well some of them are. Take a look at this lovely collection of graphs and numbers.
Disclaimer first. This is all purely numbers it cannot replace actual play test but does offer a base to work from.

Following our recent work on fixing Forge Fathers, our Chris went and got his geek on with spreadsheets and did some theoretical modelling...

Firstly lets discuss the model used.
This test has used a random number generator to model 1000 results of a Human Corp Guard Slamming a Forge Father Striker. In the test the guard gained the + 1 dice for moving into a slam.

We have modelled 5 different suggested fixes and applied them to 5 Forge Father Strikers. Each one uses the same set of dice rolls so we have a picture of what would of happened to each in each individual case.

The dice rolls are all created by using a random number between 1-6 generated by excel (=randbetween(1,6) for those that want to know.) we have modelled exploding sixes to a degree of 40 possible 6's chained together (Which is a 1/6^40 = 7.481E-32). We generate separate sets of dice rolls for "The Slam", "The Dodge/Brace" and "The Armour Saves".

The five suggested fixes that we have modelled are:

FF Normal = Current Forge Father striker (Control group)
FF Grizzled = Forge Father Striker with Grizzled
FF can Brace = Forge Father striker who can use the "Brace" slam reaction as mentioned in our previous post.
FF CFaT = Forge Father with "Cant Feel a Thing"
FF Speed 4+ = Forge Father with speed 4+


My Thoughts
Initially Speed 4+ is too good. The striker ends up winning when being slammed 28% of the time this just doesn't seem right we were trying to make them survivable not Judwan.

CFaT while it looks good stats wise, they still lose quite a bit but don't die or get injured. This needs further testing as it then brings in the problem that FF cant get up very well, and of course if they fall over they lose the ball.

Brace seems to come out as predicted it makes them less likely to get doubled 32% so hold onto the ball for longer something needs to be looked at as they are also winning 28% of the time like speed 4+

Grizzled seems to be the best fit at the moment reducing the amount of times they get taken out for 3 or more. Worryingly they end up doubling the guard the most at 18% but they have the lowest just winning 6% this is also reversed they have a low chance to only get beaten but quite a high chance of being doubled without injury.

Next
We are going to try and model the effects of giving these rules to Forge Father guards and Jacks to see what happens when they slam themselves. Particularly interested in Grizzled and CFaT (mini buzzcuts).

Then we can take the results to the tables and see how it changes actual play.

If anyone would like to generate their own results and compare them, or check that we haven't done anything wrong below is a link to a blank .xlsm doc use the buttons to generate your own results.

  
Please discuss.

8 comments:

  1. The brief test I conducted on weds before a playtest (basically explaining the new idea) we did 3 each slams using normal, grizzled and brace and found that grizzled seemed more 50/50. Obviously your test was more conclusive. Numbers cant lie.
    We're due to playtest sunday so i'll give an update then ;)

    ReplyDelete
    Replies
    1. Playtest didnt happen due to Hard Grounding at the Outpost.
      Here are my results from Blackburn's test. Slightly different, especially after only 250slams, but corroborates his results in the end:
      https://www.dropbox.com/s/m0sr5vv9kltudxh/FF%20Testing%20250.xlsm

      Delete
  2. How did u get 50/50 on the test if u only did 3?

    ReplyDelete
  3. I think you need to run the numbers with a 3+ strength as well. ALOT of teams have 3+ strength now. 6% may day against 4+ but I loose strikers constantly against 3+ strength.

    In addition it's worth noting that these numbers don't take into account - dice that will almost always be given to ensure death.

    Great work so far. Lots more variables to considder.

    ReplyDelete
    Replies
    1. 3+ is on the list of things to do. But it will only show a more more successful set of results for the guards.

      "In addition it's worth noting that these numbers don't take into account - dice that will almost always be given to ensure death."

      I dont know what you mean? If you mean coaching dice then no it wont include them as that is a pretty unique case.

      Delete
    2. He means consider the chance of less than 3 dice rolled by the Striker.

      Delete
    3. This is what actually playtesting is for it is very difficult to model every situation. Think this as a test case. No models were used just excel and maths.

      Delete
  4. With Grizzled we got FF win, Human win, Draw.
    With Brace we got Draw, Human win, Human double.
    But as i said, Mr Blackburn's test is more conclusive...

    ReplyDelete