Phoenix optimization - page 9
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Please do JB_007, I look forward to the results. I simply have too much going on :/ If you use autohotkey it is important that
#1 you use a screen resolution other people can use (1024x768 is what I use.. ugh)
#2 the applications are full screen.
You can switch applications by name so sending ALT-TAB isn't needed. Are you aiming for 2 rounds or 3? Can the date ranges be specified in Excel or Macro? (I'd recommend Excel, copy paste).
PContour's Excel to .set file writing should probably be useful for loading up results and importing them if you want to do a 3+ stage optimization. I documented a suggested 6 weeks optimize, 2 week test, 4 week test out of chronological order:
4 weeks (test3 of best results)
6 weeks (optimized, test1)
2 weeks (test2 of best results)
Longer time frames yield better results, but take longer. A better answer might be test1: 4 months test2: 1 month test3: 1-3 months
After testing AutoHotKey for few hours, I believe it is the best tool for the job.
I liked the fact that rather than just recording mouse actions and keyboards, it generates the script (macro) which can then be customized by hand.
I use 1024x768 screen resolution so that shouldn't be a problem.
I haven't been following the optimization thread so I do not fully understand the optimization strategy you described - I think I understand what you are talking about. This is a similar to strategy used for testing Neural Networks - 2/3 of data is used for training the network and the remaining 1/3 data is used for testing.
I can assure you that I will exhaustively read the optimization thread before continuing.
re using Excel and Metatrader together - that was just an idea I came up with to fully automate the optimization process. I will have to investigate it further.
Please do JB_007, I look forward to the results. I simply have too much going on :/ If you use autohotkey it is important that
#1 you use a screen resolution other people can use (1024x768 is what I use.. ugh)
#2 the applications are full screen.
You can switch applications by name so sending ALT-TAB isn't needed. Are you aiming for 2 rounds or 3? Can the date ranges be specified in Excel or Macro? (I'd recommend Excel, copy paste).
PContour's Excel to .set file writing should probably be useful for loading up results and importing them if you want to do a 3+ stage optimization. I documented a suggested 6 weeks optimize, 2 week test, 4 week test out of chronological order:
4 weeks (test3 of best results)
6 weeks (optimized, test1)
2 weeks (test2 of best results)
Longer time frames yield better results, but take longer. A better answer might be test1: 4 months test2: 1 month test3: 1-3 monthsYou are right about the NN reference, that is where I got the strategy from. The main reason to the 3rd round of testing is to more clearly delineate long term settings from short term settings.
Trend or not
Here is a conundrum for you. As I understand it, Phoenix is not trend aware, which I interpret to mean that it works the same way whether the market is trending or not.
Now a trend is usually a fairly large scale (time and amplitude) phenomenon, which we might term a metawave, whereas Phoenix presumably makes its living off wavelets and even ripples (i.e., two or three orders below a metawave in amplitude and time).
So it makes no sense to me that Phoenix should post systematically different results in a trending or non trending market. Instead, if Phoenix has good weeks and bad weeks, could this not simply be a statistical fluctuation? If so, then there must be considerable room for improvement in its signal interpretation, because an efficient EA should not be subject to large statistical fluctuations. In other words, it should be able to consistently separate bad trading situations from good trading situations, and enter and leave the market accordingly.
Comment please.
As a trend reversal system, Phoenix should be aware of all things related to trend reversals. First and foremost is indications of a strong trend but Phoenix does not currently do this. Secondly, Phoenix needs to be aware of support and resist lines (fibonacci, gann, murrey) and trade accordingly.
Nearing S&R isn't a requirement to trade, but it would be considered a confirming signal and more lots might be placed, or a confirming signal would overcome a weak signal exhibited by other indicators.
If a strong trend is detected, Phoenix shouldn't try to fight it (trade filter).
All of these are partially implemented in my Phoenix 6 code. I'll resume work on it this coming week. My first goal for Phoenix was to make it stable and make similar trades under similar conditions. Using the contributed signal count code along with the bug fixes makes Phoenix 5.7.0 perform in a way that I would trust money on. Phoenix 6 is designed to be more adaptable to different market types and not loose as much (or make money even) when a different market type is detected (strong trend most important).
Here is a conundrum for you. As I understand it, Phoenix is not trend aware, which I interpret to mean that it works the same way whether the market is trending or not.
Now a trend is usually a fairly large scale (time and amplitude) phenomenon, which we might term a metawave, whereas Phoenix presumably makes its living off wavelets and even ripples (i.e., two or three orders below a metawave in amplitude and time).
So it makes no sense to me that Phoenix should post systematically different results in a trending or non trending market. Instead, if Phoenix has good weeks and bad weeks, could this not simply be a statistical fluctuation? If so, then there must be considerable room for improvement in its signal interpretation, because an efficient EA should not be subject to large statistical fluctuations. In other words, it should be able to consistently separate bad trading situations from good trading situations, and enter and leave the market accordingly.
Comment please.I've been thinking about this issue for some time, particularly, since Hendrik's contest version of Phoenix, ran into a series of losses in the closing stages of the contest. One way to make it through a bad stretch is, to reduce the amount you trade when the percentage of trades you are losing on, rises. Daraknor has some great adaptive ideas for changing settings. My update for the MM section in much more primitive all it does is cuts back on the lot size based on the last 3 trades.
I have posted the code https://c.forex-tsd.com/smiles/nerd.png an update to Phoenix_5_7_2a and Phoenix_5_7_2_W. Please click the link for details.
Hi everyone
Daraknor, I'm glad, you are feeling better!
... because an efficient EA should not be subject to large statistical fluctuations. In other words, it should be able to consistently separate bad trading situations from good trading situations, and enter and leave the market accordingly.
About separating bad trading from good trading conditions, and because once Daraknor asked for some (indicators/help) with determining the trend, I thought to attach this interesting indicator. It may help, it may not...
Author is Luis Damiani, he was second in the championship and his EA was long time with Phoenix exchanging first and second place. If I heard correct, this indicator is part of his 2nd place winning EA Dumasa.
I got indicator here:
http://www.damianifx.com.br/indicators.php
Author's note about this indicator: "Used to filter whipsawed markets out. When noise line (grey) is above signal line (green) even trading ranges strategies will have poor performance - by Luis Guilherme Damiani"
Bertbin and Azbo, great spreadsheet, thank you! Btw, you are doing good work with your's testings!
I can't forward testing latelly, because my laptop is sometimes with me in the car so I don't have it on the internet 24/7.
I was trying to do some optimization few times, but looks like, I must of have some problems with my historic charts. No matter how I was 'optimizing' or even if I'm using current data, results are bad. I was trying to reload alpari data, had to manually delete old data, something went wrong there, I'm sure, shouldn't be so complicated... so I gave up and I'm now waiting on settings from someone else (LOL) or on new version Phoenix 6.
I'm also trying to learn some manual trading strategies (you know, just in case)
but due of lack of free time, I'm still at the very first step.
Anyway, just though to step in and say hi to everyone.
Have a nice day everyone. (and you Bertbin, be carefull, that you don't get some sun burns. (here we go again, I'm jealous again LOL)
Mario
P.s. Daraknor, do you plan, to put on Phoenix fund new version when will be ready or you will start with 5.7.x? Just curious...
Hi Mario
Good to see you back here
I went on the Damiani Web site as you suggested in your long post.
On the graphs that Damiani shows, there is another indicator named 1Damitrix or 1Damitrix adaptsys that seems great for the trend (and reverse/bounce moments)
But unfortunatly I didn't find it in the avalaible indicators...
Any idea(s) ???
See you soon I hope...
(Send you some warm degres from here at 12h30 GMT -07h30 in the morning here - air : 26 Celsius water : 23 Celsius and poor wind :2/5 knots coming from east)
I plan to trade PhoenixFund with 5.7.2a. I will probably set up 5.7.2W temporarily for USDJPY, but only USDJPY until a demo validates some other settings in forward testing.
Thank you for the indicator! Do you know the redistribution policy? It is currently configured as an indicator for manual trading. It is original enough I would want to maintain copyright even if I rewrote the code. I will probably email Mr. Damiani regarding the code if it works well for integration.
My current workload is based around unpacking, so I can find things I'm looking for in under 2 hours. Once most of the unpacking is complete, I'll resume work on Phoenix 6. My major task this week is adding new hosting customers.
Happy Day: PhoenixFund was approved as an FXDD IB. See PhoenixFund thread for details
I plan to trade PhoenixFund with 5.7.2a. I will probably set up 5.7.2W temporarily for USDJPY, but only USDJPY until a demo validates some other settings in forward testing.
Thank you for the indicator! Do you know the redistribution policy?
Tnx for the answer and sorry, don't know the policy
My current workload is based around unpacking, so I can find things I'm looking for in under 2 hours.
Lol, you are too funny! I'm glad you are feeling better
Mario