Don't take it as offensive, to predict highs and bottoms there could be 'some' need like adaptive technology, not only on fixed past movements, but also what could(would) else be as change (but of course not to go far in extreme as should be that indicator), what can worst happen (or -What's the Worst That Could Happen-)
It is just what is happening (almost always in real) without adaptive things on changes. (just my thoughts, no-one to take it as offensive).
I think that, a good system shouldn´t have adaptative parameters.
1) The best idea that I met during my Forex life is the KG model, which find the Fibbo lines at the important TF,
2) Comput the bollinger bonds as if you are at TF=1 hour chart then the Time you look back is corresponded to
H4 (time back = 4), H8 (time back = 8 - which is the third part of the day), D1 (time back = 24) etc.
Find the KG model at the forex forums (the Indonazian one)
Please check me.
You can find good geometric entry points without volatility parameters with regression channels and Gann / Fibo lines...