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
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?
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
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).
thanked this post
@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.
thanked this post
JSParticipant
Senior
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…
Hi
JS
Thank you, it gives me an idea of solving this a different way. Much appreciated.
