Forex Factory (https://www.forexfactory.com/)
-   Trading Systems (https://www.forexfactory.com/forum/71-trading-systems)
-   -   MGH System (Martingale Grid & Hedging) (https://www.forexfactory.com/thread/497448-mgh-system-martingale-grid-hedging)

Paracelsus Aug 13, 2014 6:27am | Post# 441

Welcome Back, Kfx!

As you have indicated and as I mentioned in my "game theory approach", we should concentrate on the "worst case scenario" with any setup, i.e. the chance of being forced to go through one of the many "Death Trail" price patterns.

Also excellent observation by you that the grid size should be dynamically adjusted to the given pair's volatility.

Kind regards,

Paracelsus
(alias FxMasterGuru)

spyderman Aug 13, 2014 7:25am | Post# 442

1 Attachment(s)
Hey Kfx...Welcome back.

Here's a shot of the GbpJpy I'm trading on a 50 pip grid. I missed some trade last night with the big move on the Gbp. I had been thinking the grid might be small for the GJ and now this morning you guys are talking about adjusting the grid to the ATR. I think that's a good idea.

Click to Enlarge

Name: gbpjpy!h1.png
Size: 59 KB

spyderman Aug 13, 2014 7:31am | Post# 443

1 Attachment(s)
Hey all.

In line with what Hanover was talking about, I wanted to do a longer term backtest on the system. I went back and started with 2001 and got some interesting results.

Just the 2nd series in, I hit some major DD. 11,800 pips worth. Now in 2001 the ATR on EurUsd was much larger, so again, IMO a 50 pip grid would not have been big enough. Before I go any further, I'm going to figure out the historic ATR (or at least an approximation) and modify the grid as I move through the history and see how that may change things.

Anyways...food for thought.

Click to Enlarge

Name: EURUSD!H1.png
Size: 103 KB

Kfx Aug 13, 2014 7:52am | Post# 444

Hey all. In line with what Hanover was talking about, I wanted to do a longer term backtest on the system. I went back and started with 2001 and got some interesting results. Just the 2nd series in, I hit some major DD. 11,800 pips worth. Now in 2001 the ATR on EurUsd was much larger, so again, IMO a 50 pip grid would not have been big enough. Before I go any further, I'm going to figure out the historic ATR (or at least an approximation) and modify the grid as I move through the history and see how that may change things. Anyways...food for thought....

Great spyderman.

It will be hard work to make this long backtest manually.

Paracelsus Aug 13, 2014 7:55am | Post# 445

Hey all. In line with what Hanover was talking about, I wanted to do a longer term backtest on the system. I went back and started with 2001 and got some interesting results. Just the 2nd series in, I hit some major DD. 11,800 pips worth. Now in 2001 the ATR on EurUsd was much larger, so again, IMO a 50 pip grid would not have been big enough. Before I go any further, I'm going to figure out the historic ATR (or at least an approximation) and modify the grid as I move through the history and see how that may change things. Anyways...food for thought....
I am also eager to test my "modified approach" on more distant historical charts to see the "worst case scenarios" in more volatile times.

Could you please advise Spyderman where do you get your charts? Thanks!

Kfx Aug 13, 2014 7:58am | Post# 446

Just some words on general Trading, about the expiry date of the systems.

One of the keys that have helped me in the way of Forex has been diversification.

I am sure that a system will not win forever, so we always must have several systems and diversify our capital. Retire those that not work and put more money in those that having more success.

And above all, think this is always an ongoing research and development of new systems and improving what we already have.

spyderman Aug 13, 2014 8:00am | Post# 447

{quote} I am also eager to test my "modified approach" on more distant historical charts to see the "worst case scenarios" in more volatile times. Could you please advise Spyderman where do you get your charts? Thanks!
I got the data from fxhistoricaldata.com and then imported it into MT4. They only have H1 or daily though.

Does anyone know of an easy way to show ATR for a given time period...say year to year.

Paracelsus Aug 13, 2014 8:01am | Post# 448

Hey all. In line with what Hanover was talking about, I wanted to do a longer term backtest on the system. I went back and started with 2001 and got some interesting results. Just the 2nd series in, I hit some major DD. 11,800 pips worth. Now in 2001 the ATR on EurUsd was much larger, so again, IMO a 50 pip grid would not have been big enough. Before I go any further, I'm going to figure out the historic ATR (or at least an approximation) and modify the grid as I move through the history and see how that may change things. Anyways...food for thought....
It seems that if you had applied Kfx's original method (i.e. moving TP to Level1 after Level2 entry) your results would have been significantly different and better.

By using larger grids, price has much smaller statistical chance to move an additional 50 pips to TP (after Level2/3 entries), than to travel an additional 10 pips, as I proposed in the modified version.

Paracelsus Aug 13, 2014 8:03am | Post# 449

{quote} I got the data from fxhistoricaldata.com and then imported it into MT4. They only have H1 or daily though. Does anyone know of an easy way to show ATR for a given time period...say year to year.
Thanks! I would need at least 15M to test my theory...

spyderman Aug 13, 2014 8:06am | Post# 450

{quote} It seems that if you had applied Kfx's original method (i.e. moving TP to Level1 after Level2 entry) your results would have been significantly different and better. By using larger grids, price has much smaller statistical chance to move an additional 50 pips to TP (after Level2/3 entries), than to travel an additional 10 pips, as I proposed in the modified version.

I noticed the same thing on the 2001 data. However on the current year, when I backtested, it seemed that your TP mod worked much better. It me out of a series much faster and without extending the grids to the higher multipliers.

moncsicsi78 Aug 13, 2014 10:41am | Post# 451

{quote} Great spyderman. It will be hard work to make this long backtest manually.
I've already written that we should have an EA to test this strategy. It would much more faster and easier than to testing manually.
Would anyone, who can programm, be so kind as to make an expert advisor for it?

xlcharly Aug 13, 2014 12:31pm | Post# 452

1 Attachment(s)
{quote} I got the data from fxhistoricaldata.com and then imported it into MT4. They only have H1 or daily though. Does anyone know of an easy way to show ATR for a given time period...say year to year.
something like that ? (daily base)

Click to Enlarge

Name: ScreenHunter_82 Aug. 13 18.28.jpg
Size: 170 KB

spyderman Aug 13, 2014 12:52pm | Post# 453

{quote} something like that ? (daily base) {image}
Yikes! That might be hard to read.

I actually downloaded the daily data into an excel file and ran some ATR formulas off them, so I can use that. Thanks

xlcharly Aug 13, 2014 1:06pm | Post# 454

{quote} Yikes! That might be hard to read. I actually downloaded the daily data into an excel file and ran some ATR formulas off them, so I can use that. Thanks
right....you can aggregate data for range and studying yearly frequency for each one.

spyderman Aug 14, 2014 10:03pm | Post# 455

2 Attachment(s)
This is like watching paint dry...

Click to Enlarge

Name: EURUSD!H1.png
Size: 82 KB
Click to Enlarge

Name: gbpjpy!h1.png
Size: 61 KB

GoldenEA Aug 14, 2014 10:13pm | Post# 456

This is like watching paint dry... {image} {image}
Yeah, 50 pips grid. But stress free though. How many pips you made so far?

BTW, Month August is range and boring market.

spyderman Aug 15, 2014 6:52am | Post# 457

{quote} Yeah, 50 pips grid. But stress free though. How many pips you made so far? BTW, Month August is range and boring market.

I'm tracking all the closed orders on screen as well, so it's 250 across both pairs.

spyderman Aug 15, 2014 4:03pm | Post# 458

Okay boys and girls. Been working a little bit on the 13 year EurUsd backtest this afternoon.

So first stage is 1/1/2001 - 5/20/2002. The reason I went into 2002 was that I had a long series in DD that it took until then to close.
Sooooo.... you may remember earlier that I tried this on a 50 pip grid and it blew up in 2001 right out of the gates.
This time fared better, but not great.

I took the current daily ATR at the beginning of each series and used that for the grid size for that series. Seemed to make a lot more sense adjusting the grid to the volatility of the pair.

So for stage 1 I went through 16 cycles (the last of which was 9 months long...
Total pips for the stage was 5,130
Curiously, there were only 2 of the series even went past the x1 multiplier.
The first had a DD of 900 pips and went to x2.
The second (already mentioned) had a DD of 5,060 pips and reached the x16 stage.

Draw what conclusion you will. I'll continue to work on the backtest as I'm able.

spyderman Aug 15, 2014 7:58pm | Post# 459

Stage 2
5/20/2002 - 5/19/2003
11 series...again the last one was 10 months this time...weird.
Anyways only 1 series (the last of course) went past x1. It only went to x4 but just wouldn't close due to a continued uptrend which didn't allow the TP's on the x4 to overcome the DD from the 3 open x2 trades. Total DD on that series was 4,880 pips with the yearly take at 1,915 pips

spyderman Aug 15, 2014 9:15pm | Post# 460

Stage 3
5/19/2003 - 12/31/2003
Nice clean 12 series. Got to x2 a couple of times.
Worst drawdown was 660 pips.
2,645 pips total.


© Forex Factory