Improving Strategy’s Trailing Stop Loss

• 02/15/2021 at 1:16 PM #161480
  deletedaccount051022

  Hi,

  I’m hoping for input from people to see if the trailing stop loss section of this strategy which I have been working on can be improved before publishing it.

  This strategy is taken from the book Beyond Technical Analysis by Tushar Chande, the strategy is called Bottom Fishing.  It was intended for use on a daily time-frame but I have adapted it to a 3 minute time-frame to generate more signals.  It has just a few Entry criteria of which I optimised a couple on 50k bars before then testing on 200k bars.  The Exit strategy is taken from the excellent Vectorial strategy, a very big thank you to the author and contributors of that.

  This strategy is good at taking small profits on a consistent basis, which accounts for 80% of the winning trades, and only a few losing trades overall.  hence I believe the trailing stop loss might be too tight.

  I would welcome any ideas or feedback regarding the trailing stop loss methodology to see if the performance can be improved without impacting the equity curve too greatly.  I have looked through the Snippet library and this is still work in progress, but hoping someone might have a better idea.

  Notes:

  Instrument       Dax

  Account             IG Spreadbetting, assumed spread 2.6

  Capital/Risk    10,000 and 10 per point

  Money Mgt      None utilised in this code.  I manage this manually, increasing risk per trade at month end based on original capital plus performance to date and keeping the 1,000 of capital to 1 per point ratio.

   

  Thank you very much,

  S

  1 user thanked author for this post.

  Midlanddave

  

  deletedaccount051022

  Participant

  Topics: 26

  Replies: 139

  Been thanked: 97 times
  02/15/2021 at 1:17 PM #161483
  deletedaccount051022

  With second image which didn’t load.

  

  deletedaccount051022

  Participant

  Topics: 26

  Replies: 139

  Been thanked: 97 times
  02/15/2021 at 1:55 PM #161493
  nonetheless

  What do you want the trail to achieve? to start when you’re $3500 in profit? That doesn’t seem to make sense, as you already have an ATR trail.

  I think what you want there is

  SET STOP %LOSS or SET STOP pLOSS

  1 2 3 4 5

  SET STOP %LOSS   or   SET STOP pLOSS

  with a more realistic stoploss of around 1.5% (for example)

  You do realize that the Breakeven is switched off ? did you mean to do that?

  Otherwise, tsminstop should be 5 for the DAX – this is set by IG (although they can also vary it at different times of the day).

  Also, for backtest you have to enter something for Spread, at least 2 if it’s running 24hrs

  

  nonetheless

  Participant

  Topics: 83

  Replies: 1237

  Been thanked: 978 times
  02/15/2021 at 2:26 PM #161496
  nonetheless

  This is how it looks on 1m bars

  Spread = 2

  SET STOP %LOSS 0.6

  Breakeven = 1

  positionsize = 1

  tsminstop = 5

  No other changes

  1 user thanked author for this post.

  

  nonetheless

  Participant

  Topics: 83

  Replies: 1237

  Been thanked: 978 times
  02/15/2021 at 2:33 PM #161498
  deletedaccount051022

  Hi nonetheless,

  Thank you for the flurry of ideas, feedback and observations.  That’s exactly what I was hoping for.  In reply to each of your suggestions/comments;

  1 – Trailing stop to start when you’re $3500 in profit?

  Probably too high based on average winning trade being so much lower.  But I could reconsider a lower profit target.

  2 – %LOSS at 1.5%

  I like this idea, it will create a wide enough stop to allow it to run.  However, as it trades on a 3 min time-frame, I might refine your idea further – thank you

  3 – Break even switched off, did you mean to do that?

  Yes, purely for the purposes of showing the raw performance.  Enabling the break even increases returns (see attached image), but then the majority of the winning trades are from break even profits which I feel partly clouds the analysis, but also suggests that the break even might be to tight.

  Does optimizing the Stop Loss also equate to curve fitting?

  4 – for backtest you have to enter something for Spread, at least 2 if it’s running 24hrs

  I have this set to 2.6, please see first post.  I based this on an article you had kindly posted on the MoD strategy, showing how you arrived at a suitable spread for back testing based on how the spread with IG changes throughout the different hours.

   

  Thanks once again, very helpful.

  

  deletedaccount051022

  Participant

  Topics: 26

  Replies: 139

  Been thanked: 97 times
  02/15/2021 at 2:45 PM #161500
  deletedaccount051022

  Thank you for running this over 1m bars.  Please could you provide a screenshot of the equity curve above the DAX market price chart.  As this is a long only I would imagine it might require a simple filter to avoid the drawdown period.

  Thank you,

  S

  

  deletedaccount051022

  Participant

  Topics: 26

  Replies: 139

  Been thanked: 97 times
  02/15/2021 at 2:58 PM #161501
  nonetheless

  The equity curve will be v similar to the gross performance graph, as shown.

  

  nonetheless

  Participant

  Topics: 83

  Replies: 1237

  Been thanked: 978 times
  02/15/2021 at 3:18 PM #161506
  deletedaccount051022

  Quick question – you have backtest on 1m bars testing the IG German CFD contract.  Are you a client of ProRealTime, and through them trade with IG, or a client of IG directly?  Thanks

  

  deletedaccount051022

  Participant

  Topics: 26

  Replies: 139

  Been thanked: 97 times
  02/15/2021 at 3:34 PM #161507
  nonetheless

  Both. I opened a second IG account through PRT in order to get Premium.

  

  nonetheless

  Participant

  Topics: 83

  Replies: 1237

  Been thanked: 978 times
  02/15/2021 at 8:43 PM #161535
  Mattias

  nonetheless wrote:

  Both. I opened a second IG account through PRT in order to get Premium.

  Are you allowed to have two accounts? When I talked to PRT the other day they said I had to cancel my main IG account first in order to open a Premium account through PRT.

  

  Mattias

  Participant

  Topics: 9

  Replies: 76

  Been thanked: 19 times
  02/15/2021 at 8:51 PM #161538
  nonetheless

  This was about a year ago, maybe they’ve changed their policy. I actually have 4 IG accounts in 3 different countries (don’t ask).

  1 user thanked author for this post.

  Mattias

  

  nonetheless

  Participant

  Topics: 83

  Replies: 1237

  Been thanked: 978 times
  02/15/2021 at 9:10 PM #161543
  Paul

  no trailingstop & no breakeven, but a smaller profittarget compared to the stoploss 0.6%, re-optimise on 200k, showed here results on 1M.

  // strategy //bullish = close>lowest[15](low)[5] bearish = lowest[15](low) //ORIG:20 OPT:12 on 5ok bars PullBack = (low[5]=bearish) X = high-low Y = close-open // Conditions to enter long positions indicator1 = Average[13](close) indicator2 = Average[6](close) c1 = (indicator1 CROSSES UNDER indicator2) condbuy = c1 AND Pullback AND X > 1 AND Y > 2 condsell= 0 // ctime = time >= 090500 and time < 213000

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

  // strategy //bullish = close>lowest[15](low)[5] bearish = lowest[15](low)  //ORIG:20   OPT:12  on 5ok bars PullBack = (low[5]=bearish) X = high-low Y = close-open   // Conditions to enter long positions indicator1 = Average[13](close) indicator2 = Average[6](close) c1 = (indicator1 CROSSES UNDER indicator2)   condbuy = c1 AND Pullback AND X > 1 AND Y > 2   condsell= 0   // ctime = time >= 090500 and time < 213000

   

   

   

  1 user thanked author for this post.

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  02/15/2021 at 9:38 PM #161549
  Paul

  @samsampop  Performance looks good in your code. But this part

  SET STOP $TRAILING StopLoss

  1	SET STOP $TRAILING StopLoss

  I believe it can’t be used reliable because it doesn’t read tick by ticks.

  Perhaps it’s better to use a stoploss like nonetheless suggested.

  To me ts is secondary. It’s interesting to test this code in ORT layout. Thanks for posting.

   

   

   

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  02/16/2021 at 11:09 AM #161618
  nonetheless

  Here’s another possible treatment, I added a 6min Stoch/RSI and %trail.

  This is optimized on 1m bars at 70/30, preferring Lowest Max DD. Interesting that I got very similar values to Paul with his ORT approach.

  Still patchy in places, but getting there.

  2 users thanked author for this post.

  Midlanddave, Paul

  

  nonetheless

  Participant

  Topics: 83

  Replies: 1237

  Been thanked: 978 times
  02/16/2021 at 11:37 AM #161622
  Paul

  nice results nonetheless. Using mtf  makes it slow but worth it!

   

  1 user thanked author for this post.

  nonetheless

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
• Author
  Posts