| stevehopwoodforex.com https://www.stevehopwoodforex.com/phpBB3/ Print view |
|
| Stretchy Slope. Blame Bunker for this one. https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=716 |
Page 11 of 14 |
| Author: | feucht [ Mon Oct 08, 2012 2:28 pm ] |
| Post subject: | Re: Stretchy Slope. Blame Bunker for this one. |
Hi effluvium, I just leave the EA running all the time on a machine at home on about 20 charts, just really to collect the tick data. There are a few short periods where I have lost ticks due to unexpected restarts so the dataset isn't absolutely perfect. With all the talk about repainting etc. with TMA Slope, I wanted to make sure that I was testing based on what the EA had at that tick in realtime. From there I have loaded all the tick level data into a database, and written routines (stored procedures for the db techies) that simulate the ticks coming into the EA and trades being opened and closed, based on the parameters I send into the routine. I can set up loops so overnight I can test say TP between 30 and 180 in 10 pip steps versus SL at various levels, and then I store what the result of each trade was, when it started, slope values at the time, max drawdown during the trade etc. I load this table of trades into Excel and use PowerPivot to create the table I post on here. Database programming and business intelligence is my day job, and what I feel most comfortable working with, other people might think this is a mad way to do it, and I can respect that! The EA I am running at any one time defaults to whatever I think has the best chance of success at the time I publish it, but that is just out of interest really and to make sure it works as planned, all on demo, one or 2 weeks' performance isn't really going to tell you much either way. So no real set files as such. Hope this helps! |
|
| Author: | feucht [ Mon Oct 08, 2012 3:37 pm ] |
| Post subject: | Re: Stretchy Slope. Blame Bunker for this one. |
OK I have been looking at HamFX's idea for closing trades based on the stretch value, and I have run through a number of scenarios. In the first image I am looking at 3 different ways of controlling trade exit. These are all based on defaults from the current version of the EA as posted last night, with different SL and TP where relevant. The yellow figures show the traditional closure method at either SL or TP. The green figures show a run with closure based on the first to hit out of SL, TP or a contraction of the stretch to 0.1. Overall pip haul is down, but the second image shows the same results but for losing trades only, and this method does seem quite good at trimming losses. The purple figures show closure based solely the contraction of the stretch. I was getting better results the closer the target stretch neared 0, but the stretch is usually an absolute value (i.e. is always positive), so I started looking at the stretch in non-absolute terms (so HTF-LTF for buys, LTF-HTF for sells), and targeting negative stretches to see what would happen, so here you can see stretches that have contracted to -0.5 through to +0.15. At a contraction to -0.4 we see better pips with less losers (than the default yellow method at 90/130). Note we only have one line of results here, as the TP and SL are irrelevant for this method. It's interesting to note that these contractions are approaching the sort of levels where we would be considering a trade in the other direction, so that's quite neat in itself, and this would be easier to manage than individual TPs and SLs per pair based on something like ATR (?) |
|
| Author: | Pipley [ Tue Oct 09, 2012 1:21 am ] |
| Post subject: | Re: Stretchy Slope. Blame Bunker for this one. |
Wow! This just keeps getting better and better! (And I keep feeling more and more stupid since I still have no clue about how you are able to do this!....even after I read your explanation three times!) I am so glad that you are able to do this and that you, Steve, and Bunker have brought this all together! I truly feel that you are on to something real big here! Thank you all! |
|
| Author: | Pipley [ Wed Oct 10, 2012 11:21 am ] |
| Post subject: | Re: Stretchy Slope. Blame Bunker for this one. |
I luckily turned off NB a while ago and have only been running Stretchy. Results are not so good, but the losses seem to be slowing down and I have a few starting to turn around. Anyone else seeing the same? |
|
| Author: | HamFX [ Thu Oct 11, 2012 4:38 pm ] |
| Post subject: | Re: Stretchy Slope. Blame Bunker for this one. |
Great feedback, glad to see it might have some has legs  I set it running on EU and got a winning trade, a nice start. V interesting to see your idea/result of it "pinging" past zero. Of course this happens with an elastic band too if you pull it hard enough. I'm still wondering if all the parameters (stretch, close etc ) optimise with different values on different pairs. |
|
| Author: | ianp [ Sun Oct 14, 2012 8:43 am ] |
| Post subject: | Re: Stretchy Slope. Blame Bunker for this one. |
After 2 good weeks we have had a big fall, equity drop from $10711 to $10104............ |
|
| Author: | feucht [ Mon Oct 15, 2012 8:32 pm ] |
| Post subject: | Re: Stretchy Slope. Blame Bunker for this one. |
OK, a bit late but here are the updated results for up to Friday. Hardly any change this week on the yellow (SL and TP based) figures. This is with the MW filter in my coding (monthly pivot above weekly for a buy). The purple figures are for trade closure with a contraction in the stretch. These figures aren't really comparable to last weeks, as I found a problem with my calculations for sell trades, I had the "real" stretch for them (not the normal stretch which is always positive) as LTF-HTF when it should be (of course  ) -(HTF-LTF). Again, closing at around -0.3 or -0.4 seems optimal.I found the bug at open last night when I started up my new version of the EA that incorporates trade closure on stretch contraction, and it promptly closed all my sell trades, partly because of my bad algebra, partly because I defaulted the contraction target to 0.35 instead of -0.35  One way to get a nice clean slate to test the new EA on!The other figures in the top line refer to a slight variation on MW called MDW where the daily pivot has to be above the weekly as well the monthly above the weekly for a buy. Results not as good but worth monitoring. Ian I don't know which version of SS you are using, but this seems to be the pattern, lots of wins in a row, then a glut of losers that wipes most of the gains away. I am hoping to stop SSF from trading in these conditions, and if it does still trade, I hope that the contraction of the stretch method will get us out for smaller losses. What this does mean is periods of no trades at all for days (and weeks sometimes), but when the stars align my simulations show some really stellar gains. I will post the new EA once I have done some tidying up. |
|
| Author: | feucht [ Mon Oct 15, 2012 9:04 pm ] |
| Post subject: | Re: Stretchy Slope. Blame Bunker for this one. |
Here is version 5. There are some big changes to the defaults on this, so be careful. I have added params called CloseOnContraction (defaults to true) and CloseOnContractionTarget (def -0.35). No TP or SL is specified, so CloseOnContraction is taking on all trade closure duties, so nothing on the server, trades won't close unless your terminal is connected. You can of course set "emergency" SLs, or mix and match on a whatever hits first basis of SL, TP and CloseOnContraction. I have also opened up the allowable HTF slope level to 1.0, though the pivot filter tends to stop trades that are above 0.8 in my testing. Of course, all other disclaimers apply as usual. Have fun. |
|
| Author: | feucht [ Tue Oct 16, 2012 6:36 pm ] |
| Post subject: | Re: Stretchy Slope. Blame Bunker for this one. |
Spent last night revisiting some earlier versions of SSF to see how they fared recently. With the standard settings and a range of TPs and SLs, but no filter on pivots, this is what the trades that would have been opened and closed in the last 4 weeks looked like (first picture). Compare this with the second image with the weekly pivot below the monthly for a buy (MW) for the same period (no trades opened in week 38). Much healthier using TP and SL, and only trivial losses closing with the contraction of the stretch. |
|
| Author: | Pipley [ Thu Oct 18, 2012 11:43 pm ] |
| Post subject: | Re: Stretchy Slope. Blame Bunker for this one. |
I have only had one trade (USDCAD) since I started over with Feucht5. Does that sound about right to others, or am I forgetting something? I am all for staying out of bad trades and am quite happy to wait for the good ones, but I just wanted to make sure. |
|
| All times are UTC | Page 11 of 14 |
|
Powered by phpBB® Forum Software © phpBB Limited |
|