Post
by Maddog » Wed Sep 05, 2012 2:34 pm
Nune, I'll give an explanation below, but why make this so difficult.
It does not need to be. What number, or group of numbers do you toss the most? Answer that and then build betting that covers those numbers with either a flat bet, a regression, and/or a progression as your bankroll and temperament dictate. It really is as simple as that.
Developing the skill to repeat your results, now that is the difficult part!
----------------------
The DiceTool screen shows results based on the FF Model. You say you understand that in the post above. Your tracked dice toss results are associated with one of the 5 FF buckets (PFH, SP, DP, etc). So, yes, they are tracked results, but only in the sense that they are tracked into the appropriate FF bucket.
The lower right corner of the screen, labeled as "Place Bets as Tracked", does not use the FF model mappings, but instead uses the direct probabilities for each number as calculated from your sample/tracked rolls.
So for example, in your test rolls, lets assume you tracked 100 Primary Face Hits (PFH) using the basic permutation of the 3V set (3/2-3/6).
As described before the FF model distributes those results as:
~ 25 of the 3-3’s (sixes)
~ 25 of the 2-6’s (eights)
~ 25 of the 4-4’s (eights)
~ 25 of the 5-1’s (sixes)
But when you look at the actual tracked results, lets imagine that what you actually tossed was something like:
~ 15 of the 3-3’s (sixes)
~ 30 of the 2-6’s (eights)
~ 35 of the 4-4’s (eights)
~ 20 of the 5-1’s (sixes)
So in the actual results you have 65 eights, and 35 sixes... whereas the FF Model allocates 50 eights and 50 sixes. Obviously your probability of tossing eights is actually higher then tossing sixes, so when calculated out you would see a greater edge over the eight then the six, where as the FF Model will say they are about the same. That is the difference, and is why there are different results between the "FF" section and the "as Tracked" section, and why the tracked section treats each number individually vs the "or" description on the FF side.
I disagree that Foundation Frequency Metric shows what you will do in "Future" sessions, any more then what your actual tracked results will show regarding what you can do in future sessions. What you do in future sessions is completely, 100% dependent on your ability to replicate your tossing ability from one session to the next.
However the FF model, IMO, is a more "forgiving" model in terms of smoothing out the variance you would see when tracking specific number distribution/probability from one practice session to the next.
Probably the simplest way to put this is like this: If you DO have a measure of PITCH control, or dice correlation, then DiceTool will mask it in the distribution modeling. The "As Tracked" section gives you the comparison.
