Stretchy Slope. Blame Bunker for this one.

[feucht]

Anyone still looking at/interested in this one? I have been logging the slope values tick by tick across about 20 pairs, so I can run some scenarios and try and work out the best parameter values to use. My first run through was flawed but gave me some ideas for improving performance, and after leaving it running for a couple of weeks while on hols, am seeing some good results (basically one big loss, then 30 small winners). Running with SLPips at 105, StretchTarget 1.05, HTF buy and sell levels at +-0.45. BE 25/10 JS 5. Average winners nets less than 25 pips, so may look at moving to a TP of 25, but will investigate further.

Currently rerunning the analysis with fixed code, and a month's worth of data for most pairs (>1m ticks!), will report back if anyone is still following this!

[Bunker]

Anyone still looking at/interested in this one? I have been logging the slope values tick by tick across about 20 pairs, so I can run some scenarios and try and work out the best parameter values to use. My first run through was flawed but gave me some ideas for improving performance, and after leaving it running for a couple of weeks while on hols, am seeing some good results (basically one big loss, then 30 small winners). Running with SLPips at 105, StretchTarget 1.05, HTF buy and sell levels at +-0.45. BE 25/10 JS 5. Average winners nets less than 25 pips, so may look at moving to a TP of 25, but will investigate further.

Currently rerunning the analysis with fixed code, and a month's worth of data for most pairs (>1m ticks!), will report back if anyone is still following this!

Hi feucht,

I'm still testing various configurations with this one. When the market is trending, this seems to be a winner!

Here is my test from this week. So far so good. I am using 1.2 StretchDistance.

Cheers!

Bunker

[feucht]

Glad to hear you are still keeping the faith, I like the logic behind this idea. Today not so great though, we all knew it would be wild today, should have turned off, though from a logging and analysis point of view it will make an interesting case study. Am still crunching figures for the last month, but would like to look at what bad trades look like, and whether they could be detected before a big SL is hit by their stretch profile. Hopefully will have something to post tomorrow or over the weekend...

[Bunker]

feucht,

I’m glad to see your dedication to this EA. In my opinion, this system is one of the best way to catch retracement during strong trends. My results use to be the same as yours. A LOT of BE and one (or 2) big lost. We just have to work on MM. Thanks for your help!

Cheers !

Bunker

[feucht]

Hi,

Attached are the results of my analysis looking back on all the data I have collected, some pairs go back further than other, but there is about 3 weeks at least on all of them, 5 or 6 on some. I have included the action for Thursday and Friday, which was pretty dire for me on the aforementioned settings. For each pair I have run 715 combinations of parameters

Each pic shows results for all the pairs, the rows are the StretchTarget 0.8-1.8, the columns are the SLPips 30 to 150. If you look at the BSLevel slicer in the top right you can see the HtfBuyLevel (and HtfSellLevel as its negative), I have included a view with all values of this, so you can see what influence the StretchTarget and SLPips have on their own. In every case I am using a 25/10 JS5 breakeven strategy.

I have colour coded the pips profit in each case to make it a bit easier to read. In general, lower StretchTargets and higher SLPips seem to be more profitable. The BSLevel seems to be better lower, as you will get more trades. Strangely the average pips profit per trade is good at 0.4 and 0.8 and poor in between, so I would prefer 0.4 as you will get more of them!

Another really interesting thing IMHO is that the big StretchTargets are unsuccessful in almost every case. I imagined that when the stretch was that big, a ping back would be more likely, but you just wouldn't get many trades. You don't get many trades, but when you do, they tend not to work. This makes me wonder if you could monitor trades that you have taken at a lower StretchTarget to see if they get to an extended stretch, and close them then before they hit SL to save pips. As though once the rubber band gets stretched too far it breaks rather than pinging back.

I will look next at capturing more details of individual trades in my analysis and seeing how/if we could make this work. I also need to look at my BE strategy, as my average win seems to be less than 25 pips, so a 25 pip TP could work better, or something else.

If anyone is wondering, I made the attached with PowerPivot, which is an add-in (free) for Excel 2010 and is baked in to the preview of Excel 2013. If anyone would like the original file to play about with let me know and I will share it somewhere.

[feucht]

Following some offline discussions with pipley I attach my slight reworking of this EA. I have added some wrinkles with parameters to control them:

LogToCsv will make the EA create a CSV file in your /experts/files folder named for the pair plus the contents of the parameter CsvSuffix. For every tick you get a line in the file with the time and the HTF and LTF values for the slope. I have been collecting this data and using it as the basis of some scenario analysis.

What this suggests to me is that high stretch values don't seem to be very successful. Even if you take a trade at a lower stretch, when it closes, if the stretch is above your target the EA will trade again. I have added a MaxStretch parameter that lets you create a range with StretchDistance where it will take trades. I use 0.9 to 1.2.

I also thought looking at my analysis that trades that go off into big stretches don't come back any time soon, and usually end up hitting the SL. I wondered if it might help to close trades in drawdown that have reached a high degree of stretch. The param StopAtStretch controls when this happens (defaults to 1.5).

I'm also looking at the best Breakeven settings, though the figures I have so far look a bit unrealistic, suggesting a Breakeven target of 95, which isn't reached on average until a trade has been running for 72 hours. This inevitably means positions held over the weekend, which no-one is keen on, so I will look at this again.

All the warnings from the original EA from Steve count double for mine. I program for a living, but this is my first go with Empty4, and I don't know all the pitfalls yet. It seems to work as expected from my experience, but handle with care!

[Bunker]

Thanks Feucht

I will certainly give it a whirl

I am still amazed with the ratio win/lose (or should I say BE/lose) using the ootb setting though. For 1 lose, I get 10 BE. But the results are still red pips. What I was planning to do is as simple as increasing the BE (probably too simple) . I’ll put it on my test plan as Plan B. Right now, SSF it is ! Thanks for that EA. BTW, brilliant your MaxStretch parameter.

Cheers!

Bunker

[feucht]

Thanks Bunker, I will keep on crunching the figures and seeing how best we can make this sing. Not a great day today, though the unfavourable conditions are better for trying to find ways to improve this!

[Pipley]

Thanks Feucht! I am really looking forward to learning about this! It sure makes a lot of sense to me!

I'm starting pretty much from scratch, but I hope that I will learn quickly and can, eventually, contribute to the thread!

[Bunker]

Thanks Feucht! I am really looking forward to learning about this! It sure makes a lot of sense to me!

I'm starting pretty much from scratch, but I hope that I will learn quickly and can, eventually, contribute to the thread!

Welcome aboard Pipley !