| |
BGonline.org Forums
weakness of volatility measurement as a doubling measure
Posted By: Bob Koca In Response To: Late bear off races - amazing discovery (Maik Stiebler)
Date: Friday, 16 June 2017, at 11:48 a.m.
Volatility is measured as variability in equity over the next sequence and is not the best way to measure if a double should happen. This has some weaknesses when being used as a measure of whether one should double.
Suppose that a position leads to a lot of positions in the next sequence where you then double and get a take anyways. Volatility in that set should be ignored since it does not influence the choice to double immmediately. What should be looked at instead is the regret not doubling sequences (market losers) and the regret doubling sequences.
The sum of those regrets is then the key thing. Increased variability is then not important.
For example suppose sequences A and B have the properties that i) 60% of the sequences lead to double/take next turns. Of those 60% for B they are all solid double/takes and for A there are some marginal doubles along with some close takes
ii) 20% lead to no double positions with an average doubling regret of .20. For A there is more spread here with some having huge regret and some being small regret with doubling then just barely being incorrect. For B's position they all exactly 20% regret (e.g Double equity is then .3 and ND equity is then .5).
iii) The other 20% are market losers for both. In A half of that 20% goes to DT equity of 1.05 and the other half goes to 1.65 and for B they all go to 1.35.
Volatility as measured by XG will be much higher for A. I think though that the correct doubling decision must be the same for the 2 positions.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.