When to stop a live automated system?
10/08/2019 at 5:29 PM #109636
I am running some system in prt since may this year. They are doing good/ok so far. I have found my way of choosing and “approve” my systems before going live (still wondering if its enough, and what I can improve 😁However, my question now is like the topic says “when to stop a system?”
All my systems are fully automated and I never interfere or change settings etc. I just let them run by themselvs. I of course keep track of all orders and stats. Looking at changes like % wins, losses in a row etc. BUT, my only real stop for a system is when it has a larger drawdown then tested in IS and OOS. I also add an extra 10% as I believe there is always a larger DD to come even though the system is not broken.
To stop a system is of course a must sometimes, but stopping a system that is not broken is really expensive! You end up with a drawdown and miss the future gain.
The robustness discussion is a very important topic, but I have seen very few discussions about shutting down systems.
What are your thoughts? What rules do you apply?
Happy trading!10/08/2019 at 7:56 PM #109639
stopping a system that is not broken is really expensive!
I consistently make more profit from my Systems when I intervene and stop / exit if I think the trade direction is going the wrong way compared with an overall view of market structure, notably fibonacci retrace and extensions.
I have done repeated trials, e.g 1 week intervening and I do well then I leave the Systems to run according to the code for say 1 month and they lose or make not enough to warrant all the effort.
I have done loads of manual trading so I guess this is a prerequisite as its no use stopping a System on a whim.
I hate it when I see Systems in profit and then it all dribbles away to end up in a loss!
EDIT / PS
I intervene when a System shows a good profit.
I also turn losing System Trades into manual trades via the IG Platform if I think the trade has a good chance of recovery. I have so often seen losing Autotrades exit just before reversal, e.g a Fib retrace reversal etc.10/08/2019 at 11:01 PM #109662
Be careful of automating the stopping of systems based on history – especially if they are index based systems. Range and volatility generally increase as time passes by on indices and so if you have an emergency quit filter then you have to take into account that emergency situations in the future are likely to be bigger than emergency situations in the past – but how much bigger is the 60 million dollar question?
I think I would rather work out my risk and decide that the risk is low but bad things might happen that have never happened in the past and then put some money in a strategy (money that I will not miss if I lose it when it all goes horribly wrong) – rather than say here is my money I hope the future is not as bad as the past because if it is I will lock in a huge loss!10/09/2019 at 7:43 AM #109674
Thank you for your answers. I do not automate a stop for my systems, I am just trying to get some rules around when and why I will stop a system that is no longer performing. So far I havent done so because they are all doing good . However, they will not do so forever and I will probably have to stop some in the future. I would like to have a mechanical way to look at it.
@vonasi , I have done my risk analysis and I can take a big Drawdown and thats ok. Just my ego that will be hurt 😉 . However, at some point you will finally have to stop something that is going wrong, the question is when? If one of your system is performing bad, how long will you wait until you stop it? what will make you take that decision?
I just want to be prepared when that day comes. We put a lot of work in analysing and test our system before we put them in live mode, but I believe we should also have some measurement for if/when they are obsolete.10/09/2019 at 9:33 AM #109706
I would like to have a mechanical way to look at it.
Here’s an idea that I recently started to trial …
- If Long, exit System when the 3D Indicator turns full red!
- If Short, exit System when the 3D Indicator turns full green!
Above to be checked on the same Timeframe as the System is set up to run on.
For me the logic of above is that when a Long shows full red on the 3D Indicator on System TF (or Short shows full green) then the System strategy didn’t work and the entry condition was not a good predictor of market direction this time?
From hereon we are living in the land of hope that price direction will reverse?
You could wait to see if you get 2 full reds for a Long … but losses would be increasing and stress levels rising? 🙁
For myself, I would monitor market structure and may even intervene / exit manually on the orange for Long (lime for Short) preceding the red bar! 🙂10/09/2019 at 10:28 AM #10971010/09/2019 at 11:18 AM #109713
I’ve had a few goes at converting to an Auto-System … to test out my theory described above, but as yet I’ve not been clever enough!
Might you consider having a go Vonasi please? I would open a new Topic for this.10/09/2019 at 11:27 AM #109714
Personally unless there is some unusual event then I tend not to intervene as when I have in the past the trade has just reversed on me, plus with the mix of systems I have they tend to edge each other.
I also have money management code that sizes them up and down depending on gains or losses
Again this is personal to me but I found that if I felt the need to intervene I probably had too much risk on, to counter this I devised a strategy weighting table (example attached)
Now I pretty much just check what the balance is and journal the trades at the end of the day
You must be logged in to access attached files.10/09/2019 at 12:43 PM #109723
Might you consider having a go Vonasi please? I would open a new Topic for this.
Sure – but in reality the strategy would be simply based on whichever of the following indicators that you choose to use.
- COMMODITY CHANNEL INDEX
- DIRECTIONAL INDEX
and full red or full green is just that indicator reaching the extremes of its bounded range.10/09/2019 at 1:06 PM #109726
and full red or full green is just that indicator reaching the extremes of its bounded range.
Haha I was reading too much into it!
I got a System working from the individual indicators, but thought there must be more to it!
I’ll go back and have another go and I’ll raise a Topic if needed.
Thank You10/09/2019 at 1:34 PM #109730
To intervene or not to intervene 🙂 it seems to be different from person to person and as with everything, it works for some and for some it doesnt. The discussion here is rather a question about intervene from case to case and also when in a trade. My original question was in the longer perpective, but you probably already understood that. Its a plan for when to put the system down permenently. When the edge is no longer in the market. It can take years or just a few months, but eventually it will probably stop working. Or do you take that decision as it comes?
I am sorry to repeat myself but I cannot see anyone really answer my question here. Maybe there is no really good answer, but maybe you have a thought on when you would stop trading a system if its not delivering.10/09/2019 at 2:22 PM #109734
Depending on your capital it may be that the system is not obsolete and gets stopped but that when you look at your list of systems, returns, volatility etc that you decide that the system is no longer part of the optimal mix, plus you may have an incling that it is not suitable ie if you had short bias on EUR/USD it may work great for a few years whilst EUR/USD is in a downward trend but then be next to useless for the next few years, at some point there will be systems that are a better fit for the environment that you would switch to. The phased out system can still run in demo and wait to make its come back if conditions change10/09/2019 at 2:46 PM #109740
maybe you have a thought on when you would stop trading a system if its not delivering.
If you use the performance stats (gains/ loss ratio, % of winning trades etc) as a measure to set a System going live then … when Live performance is less than these threshhold limits for a week. a month or a set number of trades … it would seem like it is time to review and re-optimise etc or file the System away in the archives??
I personally never delete anything in case I get future inspiration or so that I can see that that idea / strategy is rubbish! 🙂
I store Systems on my PC in folders … Good, Bad or Mid … and a few others which I won’t bore you with! 🙂10/09/2019 at 4:48 PM #10975010/11/2019 at 4:46 PM #109956
I created a code snippet for automating the stopping of an under performing strategy. It also includes some light money management too. You can find it in the following topic: