I am beginning to gain an appreciation on how difficult and how much skill
is needed to come up with a strategy.
I just realized that all the parameters to be optimized are integers. That
means all the indicators and the formulas must keep everything scaled
appropriately for it to work. For example in the Sample code:
double force = balanceVelocity - scale * priceVelocity;
if (force >= entry && balanceVelocity > 0 && priceVelocity < 0) {
To have resolution you would like scale to be ~ 100 or higher, which means
that balanceVelocity intrinsically needs to be much bigger than
priceVelocity as to not get overwhelmed. And then entry also ought to be in
the 100's range etc. (my line of thinking to get 1% resolution in
optimization).
Well, I coded and implemented my unity gain scaling on priceVelocity,
created an artificial 'test case' of prices increasing at a rate of 0.1 per
second, and verified the velocity was calculated correctly for any filter
period. But it was all for naught, as now the range of priceVelocity is
about +-0.005. So it throws the scaling on everything out of kelter. I
include the code below, just if anyone is interested.
I'll work on it some more to see if there is any inherent benefit. I was
hoping that it would make optimizations independent of filter period.
OK here is a two parameter velocity unit gain indicator:
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
package com.jbooktrader.indicator.price;
import com.jbooktrader.platform.indicator.*;
/**
* Velocity of price
* One parameter, Unity gain version
* Modified: A. Vanags
*
* @author Eugene Kononov
*/
public class PriceVelocity_1 extends Indicator {
private final double alpha;
private double price_ema;
public PriceVelocity_1(int Period) {
super(Period);
alpha = 2.0 / (Period + 1.0); // tau/dt=1/alpha, dt=1
}
@Override
public void calculate() {
double price = marketBook.getSnapshot().getPrice();
price_ema += (price - price_ema) * alpha;
value = (price - price_ema) * alpha;
}
@Override
public void reset() {
price_ema = marketBook.getSnapshot().getPrice();
}
}
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
And here is the two parameter version:
package com.jbooktrader.indicator.price;
import com.jbooktrader.platform.indicator.*;
/**
* Velocity of price
* Two parameter, unity gain
* Modified: A. Vanags
*
* @author Eugene Kononov
*/
public class PriceVelocity_2 extends Indicator {
private final double fastMultiplier, slowMultiplier, Gain;
private double fast, slow;
public PriceVelocity_2(int fastPeriod, int slowPeriod) {
super(fastPeriod, slowPeriod);
fastMultiplier = 2.0 / (fastPeriod + 1.0);
slowMultiplier = 2.0 / (slowPeriod + 1.0);
Gain = 1.0 / (1.0/slowMultiplier - 1.0/fastMultiplier);
}
@Override
public void calculate() {
double price = marketBook.getSnapshot().getPrice();
fast += (price - fast) * fastMultiplier;
slow += (price - slow) * slowMultiplier;
value = (fast - slow) * Gain;
}
@Override
public void reset() {
fast = slow = marketBook.getSnapshot().getPrice();
}
}
--
You received this message because you are subscribed to the Google Groups
"JBookTrader" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/jbooktrader.
For more options, visit https://groups.google.com/groups/opt_out.
