The idea ROI is not convincing. If i understand it correctly,
ROI = I/P where I is the investment and P is the pay back per turn
LZ prefers strategy 1 over 2 if ROI_1 < ROI_2
However, actually what we need to compare is the GLOBAL profit R:
R = P*N-I = P*(N-ROI) where N is approximately the total number of turns.
ROI~10 << N~300, therefore P might be the dominant factor for the global return.
IMHO, the integration of f(t) does not make sense either since
1) f(t) is essentially discrete (N/ROI ~ 30 is sufficiently large???)
2) Even if f(t) can be viewed as a continuous function, ROI(t) does not only affect df(t)/dt, in fact df(t)/dt should be a function of ROI(tau) where tau ranges from 0 to t
[ 本帖最后由 binary 于 2010-2-15 08:50 编辑 ] |