Trading Silver with PSAR SUPERTREND EA plus ZigandZag & PSAR Oscillator Confirming - page 4

To add comments, please log in or register
Bob Fisher
316
Bob Fisher  

Hello Again,

Hope you check in. I have modified Zazzy to add a new input OrderStrategy

which will allow us to switch between the old v202 order selection (value = 1) and the newer one (value = 2). These are the only two values, 0 is not used.

The old selection criteria was

Buy Signal

If the close of the bar is greater than the SuperTrend Buy

and the close of the bar is greater than the PSAR

and the close of the preceding bar is less than the preceding PSAR and

and the PSAR Oscillator is greater than the preceding PSAR Oscillator and ZAZ is greater than the preceding ZAZ

and then there is a test to determine if the current SuperTrend Buy is greater than a million, set it to the Supertrend sell value and if the Supertrend sell value is greater than a million, then set it to the buy value. Also do for the Preceding Supertrend Buy and then compare these two as a final criteria. I think someone was looking for an inflection point where the slope changes from up to down or vice versa.

The sell signal reversed the logic from greater to less than.

Notice there is no check for the PSAR oscillator crossing the zero line, just that the current value is greater than the previous, either negative or positive.

I haven't tested these changes but I'm quite sure they will work.

I am going to download Tadawulfx.com's demo system and use the same test parameters that you are using

I'll check back before signing off

Tzuman

Files:
Bob Fisher
316
Bob Fisher  

Hello JJ,

I loaded TDFX last night and tried to use the strategy tester and found there is no data which I guess is due to never using it before. Switched to FXCM and ran several tests with Strategy Option =1 and the results were atrocious. Switched to 2, the old way, and got some interesting results. All the buys appear to be stopped out for small losses or gains. The sells show extremely large losses and profits, one was for $660 loss and they run for days. When I look at the chart, I cannot see why a sells would be placed there, ZAZ trending up and transaction occurs in the middle of the ZAZ trend but PSAR oscillator is dropping. I think the low profit of the buys may be due to the stop loss being too tight. Right now I have no suspicion why the sells are being placed and why they remain open for so long. I'l work on this Monday and provide an update Monday night.

Bob Fisher
316
Bob Fisher  

Hello JJ,

I am stumped. I have been testing Zazzy in the FXCM strategy tester to analyze why the sell occurred that created a big drawdown. The strategy tester is either buggy or it does not operate as I expect. At 04:00 I have highlighted an output that shows the Order strategy is 2 The trade signal is 0 the ZAZ signal is -1, a sell and the ZAZ shows indicator 6 with a value of 1.3015 indicating a sell but the PSAR oscillator is above zero but descending. The next line shows the main code got the no trade signal=0 and then notice the next line shows a time of 12:40 or 6 hours 40 minutes later with an order placed.

What we should have seen were the entries between the two times with my diagnostics still printing, e.g. Signal & TradeSignal= on an hourly basis until a diagnostic indicating a trade was taken and then tick by tick entries until the trade closed.

2011.06.19 14:22:03 2011.01.12 12:40 ZAZ PSAR V1.02 EURUSD,H1: OrderModify error 130

2011.06.19 14:22:03 2011.01.12 12:40 ZAZ PSAR V1.02 EURUSD,H1: OrderModify error 130

2011.06.19 14:22:01 2011.01.12 04:00 ZAZ PSAR V1.02 EURUSD,H1: TradeSignal=0

2011.06.19 14:22:01 2011.01.12 04:00 ZAZ PSAR V1.02 EURUSD,H1: Signal 2 0 -1 0 ZAZ 1 0 1.3015 Occ 0.0078 0.0085

2011.06.19 14:22:01 2011.01.12 03:00 ZAZ PSAR V1.02 EURUSD,H1: TradeSignal=0

2011.06.19 14:22:01 2011.01.12 03:00 ZAZ PSAR V1.02 EURUSD,H1: Signal 2 0 0 0 ZAZ 5 0 0 Occ 0.0085 0.0098

2011.06.19 14:22:01 2011.01.12 02:00 ZAZ PSAR V1.02 EURUSD,H1: TradeSignal=0

2011.06.19 14:22:01 2011.01.12 02:00 ZAZ PSAR V1.02 EURUSD,H1: Signal 2 0 0 0 ZAZ 4 0 0 Occ 0.0098 0.007

My conclusion is that the output from the Strategy tester is extremely unreliable and cannot be used as the basis to form any conclusions about the profitability of any EA. Tomorrow, when the markets are open, I will rerun the same test on both FXCM and TDFX to determine if the problem is related to weekend processing.

JJJohnson
154
JJJohnson  

Got the modified ZAZZY EA, Tzuman. Thanks!!

Will test it both ways tonight and Monday and let you know if I have any better results.

You may well be right that the data is suspect, as Strategy Tester isn't pulling real time data.

You likely already know this, but in case not, check and see if your Reports are showing a Modeling Quality of 90% or better.

If not, go back to the history center and re-load the data a couple of times. What I do is load 1M, 5M, 15M, etc, through all of the TFs, then re-load the 1M TF again. Maybe overkill, but it gives me the 90% quality; anything less has data errors that render the results unreliable.

JJ

Bob Fisher
316
Bob Fisher  

Zazzy v1.03

Hello JJ,

To bring you up to date. Yesterday before I dropped testing efforts, I tried to use the strategy tester in the visual mode and it worked! It is also so slow that it is marginally useful only when you know the exact date to use. There were no gaps and I could see what was happening. I ran with Strategy Option 2 and determined that the large losses were actually placed correctly but the SL was not being set, consequently the orders stayed open until a reversal order was placed. What is happening on all time frames is that a slight downturn will flex the ZAZ line and create an order arrow. If the trend then reverses, when the ZAZ flex is eliminated, the order arrow disappears. Strategy 2 was reacting properly in placing the orders. Since the EA knows which bar contains the signal, I can have it confirm that the signal is still present before placing an order and also have it close the order if the signal disappears. What I also found was that the calculation of the Stop Loss was not being calculated and the order stop loss never set. I corrected this problem today.

I have created the first full release, Zazzy v1.03 for you. The Zip file contains all the sub-directories and files you need to fully set up the system, e.g. Experts,Templates, Profiles, Indicators, Includes, & Libraries. If you use Vista or Windows 7, you can double click on the zip file and then copy the directory directly onto your Metatrader. For example copy, ctl c, Experts and paste, ctl v, it onto the main Metatrader directory and the OS will unzip and copy Experts, its sub directories and all the files you need. Do the same for Profiles and Templates and you will (should be) all set. I have used !$_ as a prefix for all files to keep them together at the beginning of a file list. There is a Template that you can use to set up a screen and a Profile that will set up everything. I have tried to make sure that all the Indicators and the EA are internally consistent with regards to default settings so for example the SAR & SAR Oscillator & EA use the same step and maximum default values and colors. I'm sure I missed a few but it's a start. In addition, I included AllAverages which is a Swiss army knife type of moving average calculator. There are over 20 separate MA types as well as some band calculations. I prefer the Hull Moving Average, period=16 or 25 as a good indicator. I also included the ADX as this may be useful for identifying non trending periods and also acting as a confirmation for orders.

Cheers & Happy Testing, I will be doing the same.

Files:
june19.gif 26 kb
zazzy_v1.03.zip 135 kb
JJJohnson
154
JJJohnson  

Hi Tzuman!

Why do I get the impression you are having fun???!!! :-)

I'm looking forward to playing with the new EA!!

As you were doing that, I was taking the approach of possibly reverse engineering the performance of the v202. You might still find the results interesting, so I will attach them here.

First thing you might notice is how very different the settings are for the EURO and SILVER. In fact, to get profitable results on the EURO I had to set 'reverse system' in the EA inputs to TRUE.

Did it for only one short period in January, but figured that gave us a starting point.

I'll put v1.03 through it's paces tonight!!

Joe

By the way: Did you fix v202 so that it reads the ZAZ trend reversals? It sure acts like it.

If you did, is there any reason we couldn't label it v202_Tzuman?? Like to give credit where credit is due.

JJ

Bob Fisher
316
Bob Fisher  

Hello JJ,

I made a post last night around 9:30 that seems to be missing. I'm very enthused about your finding with the reverse trade on 202. The 202 criteria are very simple, closes less than Supertrend and PSAR. No ZAZ and no PSAR Oscillator.

Made a version of Zazzy that incorporated 202 criteria and reverse trade. Ran it last night but it did not trade, neither did 202

More later this evening

Cheers

JJJohnson
154
JJJohnson  

To Tzuman:

"So many toys, so little time!!"

As you can imagine, I haven't been idle either. Attached are the 1 month tests of the ZAZZY EA, both strategy 1 & 2. I used the ATR as stoploss and then ran the same period using the Turtle stop, both strategies. ATR beat Turtle every time.

One problem I'm having testing ZAZZY is that it maxes out my CPU capacity. It's got 1.24 GB RAM running at 2.8 GHz, which was near the top when I bought it, but probably trailing the pack today.

Anyway, the PC labors and starts running slow and hot. I'd like to keep testing the various permutations of ZAZZY; is there any way you could strip it down to the essentials for trading?

Also, I'm wondering if you were successful in attaching the trend change indicators to any of the v202 versions. You might have said, but its not clear to me which one. We can name it after you! :-)

More attachments are on their way of the results testing v202.

Back soon!!

Joe

JJJohnson
154
JJJohnson  

Well, as I was testing ZAZZY, my mercenary instincts overwhelmed me and I set up the v202 on SILVER.

As promised, Tzuman, here are the results attached. Pretty mercenary!

These results were obtained after re-optimizing the inputs. Surprisingly, a $35.00 Take Profit and a $75.00 Trailing Stop proved most profitable.

Also, when I tried starting the test on July 1, 2010 the results were negative. Didn't blow the whole account, but still bad.

Later I re-optimized for April 20 to May 20; got good results (over 50% return in one month) then tested it from May 20 to June 20 with equally pleasing returns.

My conclusion is that since markets are constantly changing (which is why most EA don't work long-term) we will need to optimize regularly. Don't know yet the optimal interval, but I think it prudent to watch the EA like a hawk and re-opt any time it starts making poor trades. Some might just want to run the optimizations every week or month.

What are your thoughts on optimization?

Anyway, I'll 'kick the tires' on the latest versions and let you know how it turns out.

Make it a GREAT day!!

Joe

To add comments, please log in or register