Signal clarity

[FibbGann]

Did we drop the RSI and put the MACD in its place? Unless I am reading this wrong. the coding is refering that the Stochastics were dropped. If not, I apologize. I think I am spiraling into an abyss here.





[PHP]//long cci
if (cci1 < cciL && cci0 > cciL ) a1= +1;
//short cci
if (cci1 > cciS && cci0 < cciS ) a1= -1;
//long rsi
if (rsi1 < rsiL && rsi0 > rsiL ) a2= +1;
//short rsi
if (rsi1 > rsiS && rsi0 < rsiS ) a2= -1;
//long stoch
if (sto1 < stochL && sto0 > stochL ) a3= +1;
//short stoch
if (sto1 > stochS && sto0 < stochS ) a3= -1;
//long macd
if (macd1 < macdL && macd0 > macdL ) a4=+1;
//short macd
if (macd1 > macdS && macd0 < macdS ) a4=-1;

//BackTest code by Alexis
if(Strategy.isLong()==false&&a1+a2+a3==+3 && dir == -1)
Strategy.doLong("Long",Strategy.CLOSE, Strategy.THISBAR);
if(Strategy.isShort()==false&&a1+a2+a3==-3 && dir == +1)
Strategy.doShort("Short",Strategy.CLOSE, Strategy.THISBAR);

//long signal = +3?
if(a1+a2+a4==+3 && dir == -1) onAction11();

//short signal = -3?
if(a1+a2+a4==-3 && dir == +1) onAction14();

[ACM]

One suggestion I would submit to this group is that before we add stops for profit (or loss) we try to have a system that performs reasonably well in its raw mode (ie as a SAR).
As is the MACD does not appear to have enhanced FibbGann's original idea, on the contrary. Maybe we should go back to that one and review what settings of RSI may make a difference and backtest several configurations.
Just an opinion
Alex

[Doji3333]

Profit Targets..

David,

The only suggestion I'll make is the following..

If you are trying to pass a "Strategy" function with a specific price, make sure that price is actually available on the bar you are trying to execute the order on. If not, then the order will not be executed.

This is why I suggested using my code example to verify the HIGH and LOW are within the range - or above the range that you are trying to execute at.

I'm speaking from personal experience with this. I strongly suggest you adopt my style (to prevent any hours of hunting for potential bugs).

Brad

[ACM]

FibbGann
That seems to be an older version. In this message I had posted an update to FGall.efs that uses the MACD in place of RSI.
Alex

[FibbGann]

I agree with Alex. I have been tweaking the RSI levels for better signals instead of the macd.


Fibbgann

[dloomis]

Re: Reply to post 'Signal clarity'

Doji3333

I see, thx.

Somehow I thought a MARKET order would get filled right away.

>

[dloomis]

Re: Reply to post 'Signal clarity'

Alex -

I agree we should get the rest of the system running first. I was just
looking into the stop idea for future reference in this project. Let's
leave it as a commented section for now.

>

[dloomis]

Re: Reply to post 'Signal clarity'

opps FibbGann

You must have missed that point - I thought it was your idea - to use the
macd instead of the rsi.

>

[FibbGann]

Some was else had suggested we move to the macd from the Rsi and I said yes. But, you're being dragged through the mud on this. and the macd is just not giving good signals. (RSI was much better. I am going to go back to the orignal efs and play with the settings on that one.If I can figure out which one that was) If we find it does'nt work we can try another study. I just don't see any others that have a good fit at this time. I just think were floating farther away from our orignal objective.


Fibbgann

[dloomis]

Re: Reply to post 'Signal clarity'

Just posted a rsi FGall - macd is on idle.

>

[FibbGann]

Here is probably the simplest way to do this. ( I should have thought of this before I brought it to the table) Lets take the RSI, CCI,STOCHASTICS. (If any others think they have a good indicator fit, they can do this too) And backtest them one by one for their best signals. Then come back and impliment them as one. Do the testing on same symbol,interval, time frame.


What do you think?

Fibbgann

[dloomis]

Re: Reply to post 'Signal clarity'

Why would the optimized parameters you develop for each indicator be the
correct parameters when you add them together?

>

[FibbGann]

Maybe some others could weigh in on this and give us their opinions. Alexis, Garth, Brad any others?


Fibbgann

[ACM]

FibbGann
I would tend to agree with David that testing/optimizing the individual components does not necessarily provide the best results when these are used as a combination.
Alex

[Gavishti]

I appologize for bringing up MACD. You seemed frustrated with RSI and I was just offering an alternative. From what you have been saying I don't think we are using MACD the same way. I'm getting better signals than from RSI.

Regardless, sticking to the original plan is probably your best way to go. You can always introduce other indicators later.

I wouldn't spend a whole lot of time trying to optimize. TradeStation has a nice optimizing feature. What I found is it is difficult to find the perfect balance. Conditions change. What was best a year ago is not likely to be right today. It is an easy trap to over-optimize for a particular set of data.

If you are really going to backtest you need to find all three market conditions (up, down, flat) plus the transitions between them or your results will be skewed. Results should be tracked by market condition as well as short and long. Some indicators do better in a particular market. I think that is what the conversation with gspiker was all about.

Good luck.