Signal clarity

[ACM]

I uploaded a backtest report to the FileShare group.
The strategy is based on the Close of the bar that generates the signal and tests over 120 days of ES #F,5 (9:30-16:15).
Alex

[dloomis]

Can you add the code to the FGall.efs and repost?

77 points since July 1? Did I read that right?

[ACM]

David
Attached is the efs with the strategy. Aside from the values (the strategy uses Close instead of Open) the trades seem to correlate to the signals on the chart.
Alex

[FibbGann]

I have already found one flaw that will improve the statistics of this efs nicely. And will address this tommorow. A hint is totally exclude RSI out of the equation.

Fibbgann

[FibbGann]

Have I struck a nerve here for the longs? Or am I beating a dead horse? This is just an extreme rough draft of what I was heading towards.


Fibbgannn

[dloomis]

Re: Reply to post 'Signal clarity'

... an extreme rough draft of what I was heading towards.


Great! Let's go for it.


>

[FibbGann]

Take a look at the RSI factor in this strategy. Because of the recent side ways trading range we are in, The RSI is barely ever providing a good signal. Any suggestions of a possible replacement? Or maybe just have to take it out.

Fibbgann

[gspiker]

Fibbgann,

You really need to understand how an indicator (and system) behaves in the long term (which is why we need years of data to do proper backtesting). What if in the 20% of the time the market is trending your RSI (with the parameters you use) your RSI is making 3x what the other indicators do.

I'm not saying not to take it out...all I'm saying is that by using the word "recent" you have raised a common mistake people make with systems. They tune for the last few trades and not for how the market behaves over time. By definition such systems will tend to fail over time.


Garth

[Gavishti]

It is widely acknowledged that indicators have their favorite markets. No system will "always" work. Go with the one that works now. When it stops working switch to a different indicator. Anyway isn't the point of using multiple indicators to have a system that works most of the time?

MACD crossing the signal line works pretty well most of the time. I use 12,26,9,HLC/3. It is a slightly different type of indicator than CCI, RSI, and Stoch. They are more oscillators and are sort of different versions of the same thing. MACD is still an oscillator but closer to a moving average. This might help with different market conditions.

In general I have found that HLC/3 generates more reliable signals than the alternatives. With regards to the Stoch a length of 13 will generate slightly stronger signals. Might help with entry and exit if you want to try it.

Also, I have a Vol*Price Oscillator that was tested on 72 years of daily data for the $DJ with remarkable results. 1,495,436.39% better than buy and hold according to Robert Colby. Personally, I think it needs work. It can whipsaw a lot in low volume. Perhaps a percentage change or voume filter would take care of the noise. One of the things on my projects list. You are welcome to a copy of it.

Another indicator that I use a lot for intraday is the $TRIN. The direction is usually where the market is headed. I reverse it for ease of reading (1 - $TRIN). The only issue is you must stick to a time template corresponding to the NYSE floor.

Hope you guys have switched to entering and exiting on close. It is pretty tough to execute on open after the close. Looks good in backtesting, but hard to get a fill in the real world.

When you get further along you may want to add a stop feature to take care of the outliers. It will help cut the losses and improve the overall system. If you are going to intraday trade with this system, an exit on close of the day will also help the statistics.

Thanks for all of your hard work. Hope my two cents is of benefit.

[FibbGann]

I completely agree... It was mentioned to test this on a 5 min chart. What I had seen was that the RSI becomes less effective in shorter time frames. Thats all I was implying.

Fibbgann

[FibbGann]

The reason I dicided to leave the MACD out is because I always seen it as a lager compared to the others. And yes the trin works nice I just use a 10 or 20 MA on a five min chart depending on market conditions. It works very well. I am also waitng for esignal to impliment a put/call ratio.

Fibbgann

[dloomis]

Re: Reply to post 'Signal clarity'

http://share.esignal.com/download.js...le=FGall15.png

has some revisions to the GUI for the 3signal system we are working on.

Comments?

>

[Gavishti]

I use 5 min bars for setups and 1 min bars for executions. From what I can see the MACD gives better signals than the RSI. The RSI is the one that looks slow to me.

[gspiker]

Go with the one that works now. When it stops working switch to a different indicator.

For most people, this rule will make sure they are almost never going to make a good trade. Some people seem to have an almost intuitive feel for when the market is changing characteristics, but they are by far the minority. For the rest of us it takes many wrong trades in a row - and by the time you are aware of the market change and switch indicators, it is time for the market to change character again.

The approach they works for more people is to pick a system that uses a few indicators that work well togther in most markets...some of them will work better in trending and other better in trading. Together they might work less well in any one market type than any one of them would...but togther they work well enough to make money in most market types.

This takes the subjective nature...which is what kills most trades...out of the equation.


Garth

[gspiker]

David,

The new GUI looks cleaner - great job! What might be interesting is to add a button(s) that would allow the indivdual indicator signals to be displayed as well...

Garth