Need help backtesting 200k

• 09/27/2020 at 1:30 PM #145532
  Monochrome

  Hi i tried to code a strategy i found on the internet. Not a very good code but it makes a nice equity curve. Probably because ustech100 has been on a long uptrend.

  If someone can help test with 200k units that’ll be super helpful.

  UStech 100 on 2 min tf

  attached the itf

   

  

  Monochrome

  Participant

  Topics: 17

  Replies: 116

  Been thanked: 56 times
  09/27/2020 at 2:14 PM #145547
  Mauro T. “Algorithm System”

  Done …. I hope it will help you.

  1 user thanked author for this post.

  Monochrome

  

  Mauro T. “Algorithm System”

  Participant

  Topics: 18

  Replies: 336

  Been thanked: 88 times
  09/27/2020 at 6:06 PM #145585
  VinzentVega

  I changed the tradetiemes from 2.30 p.m to 19 p.m. (german time) and the results seems to be more comfortable. But i can´t backtestet it with 200k.

  

  VinzentVega

  Participant

  Topics: 7

  Replies: 145

  Been thanked: 29 times
  09/27/2020 at 6:25 PM #145590
  VinzentVega

  Another point is, the trailing stop ist not profitable. Take a look at MFE, some trades are loosing 90% profit.

  1 user thanked author for this post.

  Monochrome

  

  VinzentVega

  Participant

  Topics: 7

  Replies: 145

  Been thanked: 29 times
  09/27/2020 at 8:19 PM #145599
  Monochrome

  Thats a good idea actually thanks. saves on overnight costs too. Ill look into MFE.

  

  Monochrome

  Participant

  Topics: 17

  Replies: 116

  Been thanked: 56 times
  09/28/2020 at 8:34 AM #145614
  Nicolas

  Long only strategy in a straight upward market since May. Ok 😉

  

  Nicolas

  Keymaster

  Topics: 83

  Replies: 20898

  Been thanked: 4866 times
  09/28/2020 at 9:28 AM #145629
  VinzentVega

  Nicolas wrote:

  Long only strategy in a straight upward market since May. Ok 😉

  Right, that was my first thought. Nobody knew when the trend will change. It´s better to build a strategy which works on boths sides.

  

  VinzentVega

  Participant

  Topics: 7

  Replies: 145

  Been thanked: 29 times
  09/28/2020 at 10:08 AM #145639
  Vonasi

  VinzentVega wrote:

  It´s better to build a strategy which works on boths sides.

  Not on a strategy on an equity index in my humble opinion.

  Trying to short those is a tough thing to do and this game is hard enough already without trying to fight the markets natural tendency. Better to be long only but ensure that your strategy is capable of being out of the market when those annoying recessions and major events cause a temporary correction…. and ensure that there are some of these in your back test. It is also essential to robustness test to ensure that you didn’t just get lucky and get out before a big drop just because of when your strategy was started. Starting the strategy a day later or even the day before a big drop might have mean’t that you bought in just before that large drop but your back test just got lucky because it started trading on the right day/time.

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1812 times
  09/28/2020 at 10:28 AM #145644
  Mauro T. “Algorithm System”

  The problem with these strategies with low time frames is to have little data available …. indeed I would say very little.
  Then you should always use the Walk Forward test as often suggested by Nicolas … use it as you see fit, but using it is important …. for example, I use it this way.
  In this example I have 30 test starts that are taken at random and in a time frame of 7 months it always remains positive … to me this gives me an extra certainty of not having a trading system with a lucky start as he told you above Vonasi.
  Hello

  

  Mauro T. “Algorithm System”

  Participant

  Topics: 18

  Replies: 336

  Been thanked: 88 times
  09/28/2020 at 4:36 PM #145725
  deletedaccount051022

  Hi <span class=”bbp-author-name”>monochrome,</span>

  Thank you very much for sharing,  it’s been an interesting strategy to experiment with.  I made the a series of minor changes, tested in a simple Walk-Forward and am just analyzing the results which I thought you’d be interested in seeing.

  I have added exit methodology based on market structure (thank you <span class=”bbp-author-name”>nonetheless)</span>, optimized the Ichimoku settings slightly and added a Max Position function as changing CumulateOrders=False resulted in failiure early on.  Will keep testing and keep you posted.

  Best,

  S

  1 user thanked author for this post.

  Monochrome

  

  deletedaccount051022

  Participant

  Topics: 26

  Replies: 138

  Been thanked: 99 times
  09/28/2020 at 5:33 PM #145736
  deletedaccount051022

  Meant to add, Sharpe ratio of 2.7 which is great, but Max DD still needs improving.  Thanks once again for sharing.

  

  deletedaccount051022

  Participant

  Topics: 26

  Replies: 138

  Been thanked: 99 times
  09/28/2020 at 5:35 PM #145738
  Paul

  interesting, but a 5% stoploss & pt is pretty big for 2m timeframe.

  cumulate orders is true, but if using tradeprice(1) it is only based on the last trade. What happens if your long multiple positions and also to the first position? Does the breakeven work as intended for all?

  one other point. Breakeven starts when having 60 points. If pointstokeep is set bigger then zero, you have a nice winning percentage, I like it too, but it’s cheating the statistics. If you set that at 0, it wouldn’t finish wf.

   

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  09/28/2020 at 8:05 PM #145765
  GraHal

  Mauro wrote:

  it always remains positive

  Did you optimise the strategy over the full period of the 30 WF Tests before you did the WF?

  

  GraHal

  Participant

  Topics: 222

  Replies: 9290

  Been thanked: 2460 times
  09/28/2020 at 8:06 PM #145766
  deletedaccount051022

  I agree, 5% is too wide.  I have tested it with 2.5% and the results are still very respectable.  Your point regarding the multiple orders is something I had overlooked, well spotted.

  Regarding the setting the Points To Keep to 0, the results aren’t too far off.  This issue which VinzentVega pointed out, is that the trailing stop is still not profitable. Looking at the MFE, some trades are loosing 90% profit.  Trying to see how best to resolve that one.

  

  deletedaccount051022

  Participant

  Topics: 26

  Replies: 138

  Been thanked: 99 times
  09/28/2020 at 8:36 PM #145775
  Paul

  alright, looking forward to idea’s.

  it’s interesting look at signals regardless of a position and if these are well placed.

  IF trigger= 1 and close>spanb and c1 and c2 and cc and timeEnterBefore AND timeEnterAfter AND not daysForbiddenEntry and countofposition < 10 THEN BUY 1 CONTRACTS AT MARKET SET TARGET %PROFIT 5 SET STOP %loss 5 trigger = 0 test=1 else test=0 endif graph test

  1 2 3 4 5 6 7 8 9 10 11

  IF  trigger= 1 and close>spanb and c1 and c2 and cc and timeEnterBefore AND timeEnterAfter AND not daysForbiddenEntry and countofposition < 10 THEN BUY 1 CONTRACTS AT MARKET SET TARGET %PROFIT 5 SET STOP %loss 5 trigger = 0 test=1 else test=0 endif   graph test

   

   

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
• Author
  Posts