Firstly thanks very much for your very thought provoking article.
I’d always thought I might be placing my stops too close but how close is too close? How far is too far? I think your article gives a possible answer. I like to understand the maths behind the reasoning before putting it into practice hence the questions.
Anyway here’s what I think is happening, let me know if I’m off track, which will probably be soon…
1) Compute the volatility, sigma (pips). This is the std dev of the difference between high and low for each bar (eg. M5), for a time period eg. 24Hrs. The period length is somewhat arbitrary…
2) If we assume no drift for the distribution of the random walk, mu (mean) = 0
3) Std dev, of the random walk at time t (num bars), = sigma * sqrt(t)
3) At time t, convert price to a Z score, where n=num pips gained, z=n/(sigma * sqrt(t)).
eg. Probability of price>+40pips, at time t, P[Z>+40pips] = 40 / (sigma * sqrt(t))
While P[Z>0] = 0.5 and P[Z>(some negative value)] > 0.5, I guess my question is how P[Z>(some positive value)] could be 82%, eg. P(TP)=82%, or is that not what that actually signifies/implies?
I realise that the above does not take into account the possibility of the num of pips gained/lost prior to time t, which could hit TP or SL, but simply P[Z>n], only at time t. But I thought that you dealt with that later ( P(SL only hit), P(TP only hit), etc.) and not at this point. If so, or not so, how do you derive P(TP) for your maximal curves?
Posted by byo2000