Feedback loop of prevailing market conditions

• 04/21/2022 at 10:12 AM #192009
  deletedaccount051022

  Hi,

  Within a strategy I wish to capture the prevailing market conditions at the time that a trade is open, such as volatility, volume and so on, with a view to feeding this back into the algo to identify the current prevailing market cylcle or change thereof.  From this an algo could recognise what cycle the market is entering into (eg strong bull, range bound) and use a different set of stored parameters for the next trade.

  For example, an algo goes long and after the position is opened the volatility begins to increase and the volume reduces.  The next trade the aglo should take would use a more cautious set of parameters (a tighter stop loss, a smaller target profit, a shorter linear regression lookback period etc.)

  Is this type of monitoring and feedback possible?   The code I am currently using monitors Standard Deviation up to the point a position is opened.  If volatility spikes during an open position then the same STD calculation used for the next trade should be on a shorter lookback period.

  Thanks very much

  

  deletedaccount051022

  Participant

  Topics: 26

  Replies: 139

  Been thanked: 99 times
  04/21/2022 at 3:46 PM #192024
  phoentzs

  What speaks against it, simply starting several algos with different settings such as: large/small SL, large/small TP, faster/slower trailing stop, etc., in parallel and thus dividing the risk?

  

  phoentzs

  Participant

  Topics: 86

  Replies: 982

  Been thanked: 210 times
  04/21/2022 at 5:14 PM #192028
  deletedaccount051022

  Hi phoentzs

  Thanks for your suggestion.   To do as you suggest I believe that I would need to run 3x the risk.  Might work but am ideally looking for a loopback.

  Alternatively I could manually run a re-optimise on a regular basis, but you need to stop and re-start.

  Thanks

   

  

  deletedaccount051022

  Participant

  Topics: 26

  Replies: 139

  Been thanked: 99 times
  04/21/2022 at 5:20 PM #192029
  GraHal

  samsampop wrote:

  need to stop and re-start.

  I do this and record  – in the ‘Notes for the System’ – a brief summary of performance up to when I stopped to reoptimise (over a weekend usually).

  1 user thanked author for this post.

  

  GraHal

  Participant

  Topics: 220

  Replies: 9005

  Been thanked: 2295 times
  04/21/2022 at 6:19 PM #192034
  phoentzs

  @samsampop

  why triple risk? Spread the position size across all systems. So it’s basically the same risk as having the full position size in just one system. That’s why I like to use SP500, small margin, small spread.

  1 user thanked author for this post.

  

  phoentzs

  Participant

  Topics: 86

  Replies: 982

  Been thanked: 210 times
  04/22/2022 at 1:02 PM #192065
  JS

  STD Loop

  DefParam CumulateOrders = False NormSTDPerc = Std[20](Close) / Close * 100 If NormSTDPerc > 0.2 and NormSTDPerc < 0.4 then LookBackPeriod = 20 ElsIf NormSTDPerc > 0.4 and NormSTDPerc < 0.8 then LookBackPeriod = 10 EndIf If Close > Average[LookBackPeriod](Close) then Buy 1 contract at Market EndIf If Close < Average[LookBackPeriod](Close) then SellShort 1 contract at Market EndIf

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

  DefParam CumulateOrders = False   NormSTDPerc = Std[20](Close) / Close * 100 If NormSTDPerc > 0.2 and NormSTDPerc < 0.4 then LookBackPeriod = 20 ElsIf NormSTDPerc > 0.4 and NormSTDPerc < 0.8 then LookBackPeriod = 10 EndIf   If Close > Average[LookBackPeriod](Close) then Buy 1 contract at Market EndIf   If Close < Average[LookBackPeriod](Close) then SellShort 1 contract at Market EndIf

  Hi @samsampop

  You can normalize the standard deviation (NormSTDPerc = Std[20](Close) / Close * 100) and when the normalized standard deviation falls between certain boundaries, link these boundaries to, for example, the LookBackPeriod…

   

  

  JS

  Participant

  Topics: 27

  Replies: 2043

  Been thanked: 1090 times
  04/22/2022 at 6:27 PM #192086
  deletedaccount051022

  Hi JS

  Thank you, it gives me an idea of solving this a different way.  Much appreciated.

  

  deletedaccount051022

  Participant

  Topics: 26

  Replies: 139

  Been thanked: 99 times
• Author
  Posts