An over-performing strategy?

• 12/26/2020 at 6:56 PM #155262
  Mattias

  Here’s my story:

  18th August: Became a member of this very nice forum! A couple of weeks before I downloaded PRT and started to learn about algo-trading.

  26th August: I took my first algo live in demo-version. Continued to develop strategies and didn’t take too long before having around 15-20 algos running (all on demo). Some were taken from the forum, but most of them I developed myself.

  11th November: I launched two of my algos live! A big day for me! Both of the algos looked good in my initial backtest, and continued like that in demo.

  14/15th December: 2 more algos live.

  ---

  The rest of my algos that is still running on demo but not live is either in loss, or haven’t taken enough trades to be taken live yet. I have the maximum number of demo-trades now (it’s a pity you only can have 25 running at the same time), so every time I add a new strategy I have to remove one. 🙁

  My two algos from 11th November is performing well, one is doing extremely good (for my knowledge). But the thing is, this strategy is maybe one of those I didn’t think would perform so good. It’s a strategy that is not directly based on any TA that I learned, and I honestly don’t understand why it’s working. But the backtest looked good, and therefore I launched it in demo and then live.

  My question now is, since this is the case, should I consider this a lucky shot and perhaps stop it? Or wait for a draw-down and eventually it takes off again, or maybe it’ll go back to BE and then I stop it? I attach some screenshots so you perhaps can help me interpret the stats.

  Thanks!

   

   

   

  

  Mattias

  Participant

  Topics: 9

  Replies: 76

  Been thanked: 19 times
  12/26/2020 at 7:31 PM #155266
  GraHal

  Looks like you may have got lucky as you have 4 trades that combined show a profit more than your Total Profit?

  If we could see the price curve under the equity curve then we could give better advice.

  If you post the code on here then we may be able to give even better advice

  1 user thanked author for this post.

  Mattias

  

  GraHal

  Participant

  Topics: 211

  Replies: 8768

  Been thanked: 2179 times
  12/26/2020 at 8:03 PM #155268
  Vonasi

  Stop it when it is not on the market and add code that quits the strategy if the loss equals the profit that it has already made and then set it running live again and your worries are over. Worst case you lose nothing and best case it carries on and keeps making profit.

  1 user thanked author for this post.

  Mattias

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1811 times
  12/26/2020 at 8:50 PM #155270
  Mattias

  GraHal wrote:

  Looks like you may have got lucky as you have 4 trades that combined show a profit more than your Total Profit?

  If we could see the price curve under the equity curve then we could give better advice.

  If you post the code on here then we may be able to give even better advice

  Thanks for reply! Yes, that’s true. Is that bad? I attach a pic with the price curve as well. The code is so bad written I am too embarrassed to post it, sorry! But you’re probably right, it’s most likely luck.

   

  Vonasi wrote:

  Stop it when it is not on the market and add code that quits the strategy if the loss equals the profit that it has already made and then set it running live again and your worries are over. Worst case you lose nothing and best case it carries on and keeps making profit.

  Thanks for the tips! Yes, that’s a good idea I think. Too bad you can’t do it without having to reset the stats of the strategy though. 🙁 But you can’t have everything!

  Thanks again.

  

  Mattias

  Participant

  Topics: 9

  Replies: 76

  Been thanked: 19 times
  12/26/2020 at 9:23 PM #155273
  GraHal

  Be good to see the trades / positions also under the equity curve.

  I am trying to see if the strategy has a bias towards uptrends or downtrends?

  

  GraHal

  Participant

  Topics: 211

  Replies: 8768

  Been thanked: 2179 times
  12/26/2020 at 10:09 PM #155278
  Vonasi

  GraHal has a point. There are a lot of people on these forums who have spent a lot of time staring at charts and equity curves and position sizing that could at a glance make some suggestions as to why they think your strategy looks great over a small bit of data and perhaps why it has over performed in your expectations over an even smaller bit of data.

  Information is king…. back test screen shots with everything included, which market and time frame you run it on, what sort of robustness testing you did – and most important of all the actual code so that others can analyse it and test it and robustness test it. Without full information no one can tell you if you just got lucky or might have something that might actually work.

  1 user thanked author for this post.

  GraHal

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1811 times
  12/26/2020 at 11:12 PM #155282
  Eric

  First half of December nearly flat and half of total profit were made last week

  one short and one long did the most of it

  so i would say some luck but did catch the shorts and thats good

  1 user thanked author for this post.

  Mattias

  

  Eric

  Participant

  Topics: 18

  Replies: 441

  Been thanked: 84 times
  12/26/2020 at 11:51 PM #155285
  Mattias

  Eric wrote:

  First half of December nearly flat and half of total profit were made last week

  one short and one long did the most of it

  so i would say some luck but did catch the shorts and thats good

  Yes, good analysis, thanks!

  Eric, GraHal and Vonasi, I got the help I needed from all three of you, thanks a lot! I won’t bother you anymore with this now. 🙂

  To sum it up, and what I suspected (and perhaps was afraid of), it’s luck that made my strategy successful. Some lucky trades made most of the profit, and without them it’s most likely a quite poor algo.

  But what the heck, I have only done this for less than 4 months, so some beginners luck maybe is in place!? The other strategies I have taken live is also in profit (actually luck as well, but they are more traditional in it’s TA), except one that hasn’t taken any trades yet. The profit is not as big as the the one I posted though.

  Thanks again, I will continue to practice!

   

   

  

  Mattias

  Participant

  Topics: 9

  Replies: 76

  Been thanked: 19 times
  12/26/2020 at 11:59 PM #155288
  Mattias

  Vonasi, yes that’s perhaps a bit sad! The difference in the amount of pips just buying and holding that you have in your picture probably also describes how poor my strategy has performed, and I who thought it was over-performing, it’s the other way around!

  Thanks a lot though, good to have in mind.

   

  Edit: I see now when I have posted that your post has disappeared. What happened to it?

  

  Mattias

  Participant

  Topics: 9

  Replies: 76

  Been thanked: 19 times
  12/27/2020 at 9:11 AM #155299
  Vonasi

  Ooops – sorry but I realised just seconds after posting it that my analysis of the image was incorrect so I deleted my post but you were too quick and had already read it!

  Your strategy in fact has out performed buy and hold for that period of data.

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1811 times
  12/27/2020 at 12:06 PM #155315
  Mattias

  Vonasi wrote:

  Ooops – sorry but I realised just seconds after posting it that my analysis of the image was incorrect so I deleted my post but you were too quick and had already read it!

  Your strategy in fact has out performed buy and hold for that period of data.

  Ok, I checked and that’s true. Dax made around 440 pips from 11th November 21.12  until today. My strategy made around 3 times more, so then it actually has overperfomed.

  

  Mattias

  Participant

  Topics: 9

  Replies: 76

  Been thanked: 19 times
• Author
  Posts