Dax long term adaptable strategy

• 04/28/2017 at 1:16 PM #33863
  Francesco78

  Dear all

  I have tested a slight variation on the them of my previously described strategy on short TF.

  This time I took a very long timeseries, used Daily timeframe and modified the exit strategy in term of number of bars, all is optimized with Reiner’s seasonal parameters

  Although the return of 180%  with a drawdown of  ~20k on a such a long timeseries is not great, I though it was worth posting because of the ability of this strategy to survive all the 1998/2001/2008 shocks and because it’s relative smoothness.

  Any idea to reduce the drawdown even more would be greatly appreciated.

  Best Regards

  Francesco

  1 user thanked author for this post.

  Nicolas

  

  Francesco78

  Participant

  Topics: 43

  Replies: 242

  Been thanked: 77 times
  04/28/2017 at 1:26 PM #33869
  Maz

  If the system flattens off in 2014 = it doesn’t work any more! Re-optimize for the last 2 years and see what happens. Keep in mind the spread in 1999 wasn’t 1 point 🙂

   

  1 user thanked author for this post.

  Francesco78

  

  Maz

  Participant

  Topics: 14

  Replies: 121

  Been thanked: 138 times
  04/28/2017 at 1:40 PM #33876
  Francesco78

  Thank you Maz! will do, I have a lot of concern on overoptimization on this code, thats why I didnt want to limit on the recent past.

  True, I didnt think about the spread tightening.

  Have a great day

  

  Francesco78

  Participant

  Topics: 43

  Replies: 242

  Been thanked: 77 times
  04/30/2017 at 11:54 AM #34020
  Francesco78

  Optimized for the last 5 years,shorter than that I think is not significative given the nature of the strategy as the number of trader becomes too small.

  Regards

  Francesco

  

  Francesco78

  Participant

  Topics: 43

  Replies: 242

  Been thanked: 77 times
  05/10/2017 at 5:12 PM #35101
  Francesco78

  Hello everyone,

  I did a little bit more work on the strategy and the results now are really good in my opinion

  Please let me know your thoughts.

  Regards

  Francesco

  

  Francesco78

  Participant

  Topics: 43

  Replies: 242

  Been thanked: 77 times
  05/10/2017 at 5:13 PM #35104
  Francesco78

  backtest details

  

  Francesco78

  Participant

  Topics: 43

  Replies: 242

  Been thanked: 77 times
  05/11/2017 at 10:19 AM #35178
  victormork

  Hi! Nice work Francesco. I haven’t looked to much in detail at your system but I was wondering about if you turn off the seasonal multiplier the equity curve gets very unstable, what are your thoughts about that?

  1 user thanked author for this post.

  Francesco78

  

  victormork

  Participant

  Topics: 19

  Replies: 165

  Been thanked: 57 times
  05/15/2017 at 2:09 PM #35591
  Francesco78

  Hi VictorMork, my apologies for such adelay in answering, the point of seasonal multiplier is exactly to make the equity curve more stable, so it is not surprising the fact that removing them worsen the performance, it actually a proof that they do what they should do.

  The concern is instead to cosider if the multiplier provide an overfitting or not. I still dont have a clear idea about it.

  Regards

  Francesco

  

  Francesco78

  Participant

  Topics: 43

  Replies: 242

  Been thanked: 77 times
  05/15/2017 at 3:27 PM #35597
  Nicolas

  To get rid of overfitting, a strategy should first be tested with a fixed lot size. If the strategy is so-so or already good, then the seasonal multiplier should be added, why not .. 🙂 It should not be considered as a way to make a strategy better, because it only increase the risk exposure of your account (I did not test your strategy Francesco, it is just a general idea).

  

  Nicolas

  Keymaster

  Topics: 81

  Replies: 20828

  Been thanked: 4738 times
  05/15/2017 at 4:16 PM #35611
  Francesco78

  Thank you Nicolas for your coments, I totally agree with what you wrote. Only thing Id like to point out is that in this specific case I renormized the positions size to take into account the risk magnifying effect of the seaso als parameters. Specifically i divided the position size by the average if the factors (2.666in this case).

  

  Francesco78

  Participant

  Topics: 43

  Replies: 242

  Been thanked: 77 times
• Author
  Posts