[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: RJP < .a real deal ? i think not !



PureBytes Links

Trading Reference Links

cow pie in the sky thingy
 
first signal trigger:
momentum > momentum[displ] and 
momentum < buy zone (30) and
smoothed(stochastic) > midpoint (50), buy
 
momentum < momentum[displ] and
momentum > sell zone (70) and
smoothed(stochastic) < midpoint (50), sell
 
second signal trigger:
momentum > buy zone (30) and
smoothed(stochastic) > midpoint (50), buy
 
momentum < sell zone (70) and
smoothed(stochastic) < midpoint (50), sell
 
greatly oversimplified system logic, but you get the idea.  custom
smoothed stochastic lengths for slow and fast d are generally 3 or 4
times larger than the momentum length.  parameters shown are only for
illustrative purposes. 
 
system continues generating signals in one direction as long as the
trend filter remains in that direction and the other conditions are
true. 
 
while the displaced momentum length is fixed, the momentum length may
be optimized.  but that's what gets you into big trouble on the ts
backtest.  at first blush, it's looks great, catching highs and lows,
and wonderful stats.  but after a little real time forward testing,
the thing falls apart rapidly.  and you end up on the "must constantly
reoptimize the thing" treadmill.  trust me, it ain't worth the hassle.
 
TJ
 
 ---cvinton  wrote:
Hmm... What exactly is a momentum crossover system? If it's analogous
to moving average crossovers, then the buy signal would be when, for
example, the 5-day momentum crosses over the 20-day momentum. Is that
the idea, or am out in left field here?
 
Hmm... Suggestions on how to use stochastics to keep you in the trend?
Perhaps use a larger time-frame and then make sure the value is within a
certain range?