stevehopwoodforex.com
https://www.stevehopwoodforex.com/phpBB3/
Print view

Macd Stochastic Fractal MA5/25 Daily Marylin
https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=1448		 Page 64 of 141
Author:  auvergnat [ Fri Jun 28, 2013 5:39 am ]
Post subject:  Re: Macd Stochastic Fractal MA5/25 Daily Marylin
Thks dudest, I Need tout remove my dark glasses :D
Author:  Radar [ Fri Jun 28, 2013 2:51 pm ]
Post subject:  Re: Macd Stochastic Fractal MA5/25 Daily Marylin
Hey Phil,

Nice system you've got there! :) I started reading this thread on Wednesday night, and put the EA on a EU chart on Thursday morning, (I usually wait until the weekend before setting up EA's).

I have a few questions and general ideas for you about her...

1) Aside from the W1 RSI, all other indicators are getting their values from the D1 chart, yes? So it is possible to place this on a M15 chart, and change W1 RSI to something like H1 or H4 RSI? (This would only be to speed up development of filters to keep us out of the losers, without killing the winners).

3) How long is a signal valid for? A few days? Until W1 RSI signals an exit?

2) Have you considered creating "Shadow trades" (in the same way that Bunker's "Double Big Mac" creates shadow baskets, and doesn't send any orders until the "shadow basket" is in profit)?

If we were to use shadow trades, we then have to be able to determine when to bring them out of the shadows, so we would need to know...

a) What percentage of winners run into profit without ever going into DD?

b) What percentage of losers run into profit first, then turn around and hit the virtual SL? (And what is their maximum and average positive excursions?)

c) What percentage of losers run to the SL without ever seeing any profit?

d) What percentage of winners start out in DD before turning around to take profit? (And what is their maximum and average negative excursions?).

e) Are there consistant discernable differences between them?

With type a and b trades, we can bypass the shadowing system, and send the orders as the signals are triggered.

With type c trades, well, you've been asking this question from the start... "How to filter them without filtering the winners?"

Now, with type d trades, here's the fun bit with using shadow trades... You have your virtual stop at 300 pips from the original signal price, but you end up sending the real trade when it is much closer to that stop. For example, PA is at 1.7000 when you get a buy signal, so you create a shadow trade with VSL at 1.6700, the average negative excursion is 150 pips, so you send the real trade at 1.6850, leaving you with an SL of 150 pips, (if price runs in your favour before hitting average negative excursion, you can enter at or before 1.7000).

I'd like to be able to help with this analysis, but I don't have the hardware, the software, or the wetware :( There are, however, quite a few people on this site that have them all :)

Take care,

&

Have fun!

Radar =8^)
Author:  phil_trade [ Fri Jun 28, 2013 3:33 pm ]
Post subject:  Re: Macd Stochastic Fractal MA5/25 Daily Marylin
I have a few questions and general ideas for you about her... I'm happy it's only a few :):)

1) Aside from the W1 RSI, all other indicators are getting their values from the D1 chart, yes? So it is possible to place this on a M15 chart, and change W1 RSI to something like H1 or H4 RSI? (This would only be to speed up development of filters to keep us out of the losers, without killing the winners).? Could you clarify this ?

3) How long is a signal valid for? A few days? Until W1 RSI signals an exit? from a few days to a few week..depending the trend ! And of course, if you use W1 RSI as a exit filter... when param level is reached

2) Have you considered creating "Shadow trades" (in the same way that Bunker's "Double Big Mac" creates shadow baskets, and doesn't send any orders until the "shadow basket" is in profit)? No. This strategy try to catch very early the trend. So generate some bad trades, but good trades are very remunerative :) I'm afraid that shadow trade would avoid this early catch I want ? Remember €/$ 2009-2013 ratio bad/good trade is about 50%. but Good trade are running while bad trade cutting.

If we were to use shadow trades, we then have to be able to determine when to bring them out of the shadows, so we would need to know... I can send an XLS file with all FT trade from 209-2013 if someone could do this sort of analyse

a) What percentage of winners run into profit without ever going into DD?

b) What percentage of losers run into profit first, then turn around and hit the virtual SL? (And what is their maximum and average positive excursions?)

c) What percentage of losers run to the SL without ever seeing any profit?

d) What percentage of winners start out in DD before turning around to take profit? (And what is their maximum and average negative excursions?).

e) Are there consistant discernable differences between them? I never saw one ! but still searching. Just now with CSS

With type a and b trades, we can bypass the shadowing system, and send the orders as the signals are triggered.

With type c trades, well, you've been asking this question from the start... "How to filter them without filtering the winners?"

Now, with type d trades, here's the fun bit with using shadow trades... You have your virtual stop at 300 pips from the original signal price, but you end up sending the real trade when it is much closer to that stop. For example, PA is at 1.7000 when you get a buy signal, so you create a shadow trade with VSL at 1.6700, the average negative excursion is 150 pips, so you send the real trade at 1.6850, leaving you with an SL of 150 pips, (if price runs in your favour before hitting average negative excursion, you can enter at or before 1.7000).

I'd like to be able to help with this analysis, but I don't have the hardware, the software, or the wetware :( There are, however, quite a few people on this site that have them all :) I'm always waiting for someone to help...that's the reason why I publish this code !

Thks for your analyse and time ! Hope a team will be formed to propel Marilyn in the firmament :)
Author:  Radar [ Fri Jun 28, 2013 4:49 pm ]
Post subject:  Re: Macd Stochastic Fractal MA5/25 Daily Marylin
Hey Phil,
phil_trade wrote:I have a few questions and general ideas for you about her... I'm happy it's only a few :):)

1) Aside from the W1 RSI, all other indicators are getting their values from the D1 chart, yes? So it is possible to place this on a M15 chart, and change W1 RSI to something like H1 or H4 RSI? (This would only be to speed up development of filters to keep us out of the losers, without killing the winners).? Could you clarify this ?
I mean, aside from the RSI exit being hard-coded to use the W1 timeframe, do the rest of the indicators automatically use whatever timeframe the EA is on? Backtesting is fast, but forward testing takes many months, so if we had a version that ran on M15, we can test and refine different methods to improve the win/loss ratio in a much shorter time, then transfer the best ones up to the D1 version.
3) How long is a signal valid for? A few days? Until W1 RSI signals an exit? from a few days to a few week..depending the trend ! And of course, if you use W1 RSI as a exit filter... when param level is reached
So, if I put the EA on a new chart, and see that there was an entry signalled a couple of days ago, I can just wait for a retracement, and then enter on the return to trend?
I can send an XLS file with all FT trade from 209-2013 if someone could do this sort of analyse
Does it show hi-low of each trade between open and close? Is it possible to generate that info with your backtesting? If so, could you do a backtest without any scale-in trades? That would answer questions a - d, and give us some basic analysis to start with. Once the brains trust here gets ahold of things, we can then look at scale-ins.
e) Are there consistant discernable differences between them? I never saw one ! but still searching. Just now with CSS
I'll ask Niepce if it's non-trivial to convert the EA to lower timeframes, to see what we can test

Thanks for the time, and the system :)

Have fun!

Radar =8^)
Author:  phil_trade [ Fri Jun 28, 2013 5:38 pm ]
Post subject:  Re: Macd Stochastic Fractal MA5/25 Daily Marylin
So, if I put the EA on a new chart, and see that there was an entry signalled a couple of days ago, I can just wait for a retracement, and then enter on the return to trend?
Yes, if MACD/Stoch/MA5-25 still align !
I mean, aside from the RSI exit being hard-coded to use the W1 timeframe, do the rest of the indicators automatically use whatever timeframe the EA is on? Backtesting is fast, but forward testing takes many months, so if we had a version that ran on M15, we can test and refine different methods to improve the win/loss ratio in a much shorter time, then transfer the best ones up to the D1 version.
I think it's possible if you change the parameters. Please confirm with Niepce
Does it show hi-low of each trade between open and close? Is it possible to generate that info with your backtesting? If so, could you do a backtest without any scale-in trades? That would answer questions a - d, and give us some basic analysis to start with. Once the brains trust here gets ahold of things, we can then look at scale-ins.
I think I can write code for this. "without any scale-in trades" -> Could you explain more ?
Author:  Radar [ Sat Jun 29, 2013 8:41 am ]
Post subject:  Re: Macd Stochastic Fractal MA5/25 Daily Marylin
Hey Phil,
phil_trade wrote:
Does it show hi-low of each trade between open and close? Is it possible to generate that info with your backtesting? If so, could you do a backtest without any scale-in trades? That would answer questions a - d, and give us some basic analysis to start with. Once the brains trust here gets ahold of things, we can then look at scale-ins.
I think I can write code for this. "without any scale-in trades" -> Could you explain more ?
I was thinking of initially looking at only the first entry of each set of 3, but now remember that I can hide/reveal scale-in trades in excel, (On Friday nights/Saturday mornings I tend to lose my thinking-cap).

I should be able to tag different entries for closer inspection. From there, we'll need to look at the data and charts on the approach to each tagged trade, to see what we can see.

edit... I just realised we'll need dates/times for open, high, low, and close of each trade, thanks.

Have fun!

Radar =8^)
Author:  phil_trade [ Sat Jun 29, 2013 10:48 am ]
Post subject:  Re: Macd Stochastic Fractal MA5/25 Daily Marylin
Radar wrote:Hey Phil,
phil_trade wrote:
Does it show hi-low of each trade between open and close? Is it possible to generate that info with your backtesting? If so, could you do a backtest without any scale-in trades? That would answer questions a - d, and give us some basic analysis to start with. Once the brains trust here gets ahold of things, we can then look at scale-ins.
I think I can write code for this. "without any scale-in trades" -> Could you explain more ?
I was thinking of initially looking at only the first entry of each set of 3, but now remember that I can hide/reveal scale-in trades in excel, (On Friday nights/Saturday mornings I tend to lose my thinking-cap).

I should be able to tag different entries for closer inspection. From there, we'll need to look at the data and charts on the approach to each tagged trade, to see what we can see.

edit... I just realised we'll need dates/times for open, high, low, and close of each trade, thanks.

Have fun!

Radar =8^)
Hi Radar

please find the standard XLS output that come with my FT test from 2009-2013. Let me know if I have to add something else. I will do if I could.

cheers

Philippe
Author:  dudest [ Sat Jun 29, 2013 11:13 am ]
Post subject:  Re: Macd Stochastic Fractal MA5/25 Daily Marylin
phil_trade wrote: ... This strategy try to catch very early the trend. So generate some bad trades, but good trades are very remunerative :) I'm afraid that shadow trade would avoid this early catch I want ? Remember €/$ 2009-2013 ratio bad/good trade is about 50%. but Good trade are running while bad trade cutting...
LOL! This reminds me of Ed Seykota's Whipsaw Song! --> "♫one good trend pays for them all!♫"

https://www.youtube.com/watch?v=LiE1VgWdcQM

:)
Author:  phil_trade [ Sat Jun 29, 2013 12:01 pm ]
Post subject:  Re: Macd Stochastic Fractal MA5/25 Daily Marylin
dudest wrote:
phil_trade wrote: ... This strategy try to catch very early the trend. So generate some bad trades, but good trades are very remunerative :) I'm afraid that shadow trade would avoid this early catch I want ? Remember €/$ 2009-2013 ratio bad/good trade is about 50%. but Good trade are running while bad trade cutting...
LOL! This reminds me of Ed Seykota's Whipsaw Song! --> "♫one good trend pays for them all!♫"

https://www.youtube.com/watch?v=LiE1VgWdcQM

:)

Yiiiihaaaaa :)

I adopt it as the official hymn of Marilyn !

Tks Dudest !
Author:  Radar [ Sat Jun 29, 2013 12:50 pm ]
Post subject:  Re: Macd Stochastic Fractal MA5/25 Daily Marylin
Hey Phil,
phil_trade wrote:

Hi Radar

please find the standard XLS output that come with my FT test from 2009-2013. Let me know if I have to add something else. I will do if I could.

cheers

Philippe
Thanks for that:) I'll need the following columns...

Ticket, Symbol, Trade-Type, Open Time, Open Price, 1st Extreme Time, 1st Extreme Price, 1st Extreme Pips, 2nd Extreme Time, 2nd Extreme Price, 2nd Extreme Pips, Close Time, Close Price, and finally, Close Pips. The 1st and 2nd Extremes are the highest and lowest points of each trade, in chronological order.

Eventually, I'll need to backtest on the same data, but having the info in the XLS will show me what trades to concentrate on when searching for any possible filters.

Have fun!

Radar =8^)
All times are UTC Page 64 of 141

Powered by phpBB® Forum Software © phpBB Limited