Trading Silver with PSAR SUPERTREND EA plus ZigandZag & PSAR Oscillator Confirming - page 5

To add comments, please log in or register
JJJohnson
154
JJJohnson  

Tzuman:

Just today I read your email asking for the URL to the brokers (end of page 3).

You may have them already, but here they are -- just in case:

Online forex trading with leading online forex broker Tadawul FX

Forex | Forex Broker | Trade Forex | Trading Point

Sorry about the delay!!

JJ

Bob Fisher
316
Bob Fisher  

Hi Joe

Your results are in line with those of yesterday, really good. Were they with Signal Reversal? I'll put the SL TP into the system defaults. I think we should look at using a multiple of the ATR or Standard Deviation so that when we switch Time Frames or Currencies, it should make less of an impact.

Regarding optimization, I believe in it and think that the near term history is the best to use, like 4/1 or 5/1 through 5/30. Also think it can be overdone if you keep adding and optimizing more and more parameters, then you go past optimization and into curve fitting for a specific time frame. I suspect that weekly optimization is best, perhaps on a 4-8 week rolling interval.

Today was a lost day; I had an ophthalmologist exam and the dilation drops prevented me from using the computer. Tomorrow I plan to try to get Zazzy and 202 tested to the point that they produce identical trades. After that, I'll implement ATR SLs and TPs, also want to incorporate ZAZ correctly. I'll use the time period from 1/1 through 5/31. Do we agree that 203 is not worth investing much more time into it? I'll also release an update tomorrow as well.

Bob

JJJohnson
154
JJJohnson  

Well Bob, other than the dilation, I hope your eye exam went well!

I've been busy this afternoon, but am going to have to stop for today, as I have a meeting this evening.

In response to your questions:

The EURO v202 test in post 38 IS reversed; the EURO test below is NOT. I don't see a pattern emerging yet to explain the when and why. The SILVER tests are NOT reversed.

Testing v202 this afternoon it appears that the EA is NOT reading the market direction. I'll try to find and attach that test as well.

I like the idea of stabilizing the indicators for when we change TF or currencies, etc, but I'm unsure as to how best to do that. Without hard facts I will defer to your judgment as to what to try. Then we can test it and get the hard facts needed.

If we get the ZAZ and PSAR-O signaling properly, I've always thought using the longer TF is helpful, even for those prefering scalping, etc, in short timeframes.

Finally, I ran some tests using the EURO 4H and it performed poorly. That had been my experience with SILVER in the past; the 1H seems to work best with this EA. We might find some way to make it work, in time.

Tomorrow I'll send you reports on how SILVER did with the TP and TS added.

Meanwhile, have a good night!!

Joe

Bob Fisher
316
Bob Fisher  

Version 202a & 1.04

Hello JJ,

I have been busy and made very good progress today. First, a general confirmation on your astute observations in previous posts. The PSAR 202 was indeed only taking buy orders. The PSAR for a sell always occurs 1 bar before the SuperTrend sell is set. Consequently sell orders are never processed; this has been fixed. The original 202 version did not use the ZAZ and the Leeb 203 had a serious logic error using it. The ZigZag line, indicator 0, only has values for each inflection point, otherwise its zero. I added code to properly compensate for this and now the order criteria only place orders in the direction of the slope of the ZigZag line. A problem and its a big one is that if a reversal goes far enough to create an inflection in the ZigZag line, the selection logic will issue an order. Subsequently if the reversal reverses and takes out the minor reversal, we will have whiplash. Maybe we need trend verification from ZAZ from a higher time frame. Sometime I am going to look into the "jittering" arrows that move from bar to bar and back again on each tick; I suspect it is a program bug caused by improper initialization.

I updated PSAR 202 to create 202a by removing the martingale and correcting a lot of minor issues. To my knowledge, both Zazzy v1.04 and PSAR v202a are using the same logic for identifying orders. Zazzy has a lot of bugs in the handling of SLs and TPs. I ran several tests with Zazzy and got excellent probabilities of success, over 60% for sales and 70% for buys. Unfortunately, the expected gains for each were very low and the expected losses were much greater. I did my testing from 1/3 to today and never got better than a 0.5% gain, pretty dismal. To eliminate the SL processing errors, I switched to PSAR and the runs generally netted out to around zero. I am running an optimization now, never ran one before, and it is taking an extremely long time. Any tips, techniques, etc you can give me on using the Strategy tester will be appreciated.

I found a lot of SuperTrend variants last night and I'm going to see if they are useful. I am also going to try and fix the TP SL problems in Zazzy tomorrow.

We do not need the SAR Oscillator anymore. Determining when the PSAR goes from below to above or vice versa is the same as the Oscillator going through zero. I also enhanced the SAR indicator to add an option to display either dots or a line. This version uses two up and down buffers, one set identifies high quality values and the other set marginal ones, generally at the start of the trend. I verified that the initial marginal values are identical to the built in iSAR in Metatrader.

I'll check in from time to time but I don't expect too much progress in the next few days

Cheers and good testing,

Bob

Bob Fisher
316
Bob Fisher  

oops

Here is the system

Files:
zazzy_v1.04.zip 152 kb
Trade4gain
6
Trade4gain  

Hello Tzuman, JJ and @all,

i`m a new german member to this forum and hope there will be no communicationsproblems with my english (american) Languageknowledge. I found this thread because i`m searching for a psar-EA, so i found you.

First i want to gratulate to this fine work. So i be mightily impressed.

I followed the thread since opend but i was much progressed in my job. Now i got a little more time and want to assistance my help for backtesting etc. In programming i`m just a student with low level to metatrader. Usually i use ExpertAdvisorBuilderforMT or MolanisSB.

For only Backtesting i use two Accounts (Alpari and AT) with modellingquality from 90% and monthly history update.

For livetrading i use Hotforex and Gomarkets (ECN).

O.k. I just want to introduce short myself and want to ask how i can assitance you.

By

Bob Fisher
316
Bob Fisher  

Hello Trade4Gain,

Welcome aboard. Its good to get another contributor,. First let me emphasize that the EAs on this thread are not ready for live trading. They are concepts that should be used on demo accounts only. Any testing you can do and any conclusions you can reach based on your testing will be greatly appreciated.

Right now the EA is a combination of the PSAR, ZigAnZag and SuperTrend. We are trying to identify the best combination of the three for entry and also to look at various strategies for Stop Losses and Take Profits. Both Zazzy and PSAR SuperTrend should open the same orders. However, how they handle the SLs and TPs is entirely different. Of the two, Zazzy is more complex with more options, however there are a lot of software issues to identify and rectify.

Cheers,

Tzuman

Trade4gain
6
Trade4gain  

@tzuman,

o.k. I use the weekend for learning the functions of both ea`s and make some different trades in the strategytester to see the principal working of the both ea`s. I put the ea`s separatly in two mt-accounts for testing. If i have generaly question i will post it here.

So i think that i can start next Monday with Forexopening to assist you.

But first i must learn the generaly functionality of this forum to send attachments, links,etc.

By

T4G

JJJohnson
154
JJJohnson  

Trade 4 Gain!

Allow me to add my 'Welcome' to that of Tzuman!! Its always good to get the observations and contributions of others. We look forward to collaborating with you on this project!!

JJ Johnson

JJJohnson
154
JJJohnson  

Tzuman!

Found out why the Jan to June SILVER test wouldn't upload: It ran over a million bits or bytes or whatever the max allowable size measurement is. I'll try to run a shorter one so we can look at it trade by trade.

Meanwhile, here are the settings:

Period 1 Hour (H1) 2011.01.03 00:00 - 2011.06.17 22:00 (2011.01.01 - 2011.06.20)

Model Every tick (the most precise method based on all available least timeframes)

Parameters S1="---------------- Entry Settings"; Step=0.06; Maximum=0.6; TrendCCIPeriod=45; S2="---------------- Money Management"; Lots=0.1; RiskMM=false; RiskPercent=1; Martingale=false; Multiplier=2; MinLots=0.01; MaxLots=100; S3="---------------- Order Management"; StopLoss=0; TakeProfit=0; HideSL=false; HideTP=false; TrailingProfit=35; TrailingStop=75; TrailingStep=3; BreakEven=0; MaxOrders=100; Slippage=3; Magic=2009; S6="---------------- Extras"; Hedge=false; HedgeSL=0; HedgeTP=0; ReverseSystem=false;

Bars in test 3762 Ticks modelled 2862513 Modelling quality 90.00%

I agree, we can set the v203 and the v202leebmod EAs aside. Over a six month test I couldn't get either of them to take a trade.

Last time I tried optimizing all of the settings together it would have taken nearly two hours with my CPU running at max.

Now I generally run one input at a time. Also, if I've had good results with certain settings, I will use that as a starting point and test to either side of that value. Hope that helps!!

Did I understand that you will have v202a for us to test this weekend?

I look forward to working with it.

Taking the Martingale out makes complete sense. I played with it and even with profitable settings, the Martingale strategy resulted in losses.

I agree that using higher TFs should help stabilize the ZAZ signals. Using the buffers to avoid whipsaws also sounds like the right idea. Right now I'm running a visual mode test (Feb 1-28, 2011) on SILVER that will take about twelve hours to complete. When the market is ranging the EA takes losses.

One question we need to ask ourselves: If we had an EA that NEVER traded in ranging markets and thus took NO trades for weeks at a time, would we (could we) be happy with it?

Seems we could be if the EA could follow several currencies simultaneously, thus always having something to trade. Alternately, if an investor had a separate account for each currency, most times at least one EA would be trading.

That requires self-knowledge; are we REALLY in it to make money (its a business to us) or are we in it for the action? HMMM!!

I pose this question for everyone following our thread.

Make it a great day, Bob!!

Joe

To add comments, please log in or register