Again 6 (!) strategies stopped!
Forums › ProRealTime English forum › ProRealTime platform support › Again 6 (!) strategies stopped!
- This topic has 86 replies, 7 voices, and was last updated 1 year ago by PeterSt.
-
-
06/02/2018 at 8:23 PM #72074
I am trying to speed up backtesting in PRT
You must be trying to optimize too many variables at the same time either that or your strategies are way too complicated if you think back testing is too slow.
Most of my back tests complete within a minute or two but then I believe simple is best and then use a broad range of variables and hone in on the correct value step by step.
06/02/2018 at 8:30 PM #72075Thanks Vonasi, I’m backtesting with 100k points 1min, using a simple MACD strategy but overlaid with a position size money management system. Optimising 2 variables can take 30mins or more. Perhaps I need to start from the beginning and keep it simple.
06/02/2018 at 9:24 PM #72078I prefer longer time frames and I am not so fussed on tick by tick data which helps a lot. For me the minute time frame is just noise with a big spread in the middle of it.
Maybe rather than discussing it here you should start an alternative thread on how to best backtest although it has been discussed before in various threads. Maybe the search box could be your friend?
06/03/2018 at 9:41 AM #72101@Vonasi: If you want to use another programming language, you ahave to solve the following tasks:
1.) Find a supplier for the price data you want to use. I for example use quandl for futures and alpha vantage for US stocks and ETFs. So you have to connect to the suppliers API to download the data. This historical data you an use to run a backtest for example.
2.) Then you program you strategy in the language of your choice. You connect to IG via their web API. IG has 2 APIs, one rest API (this one you use mostly) and a streaming API if you need to stream data to make your strategy work.
Via the API you communicate with IG. Send your orders, get info about the execution of a pending order and so on.
3.) When you are done you will need a server you can run your strategy on. Theoretically you can do this with your computer at home but then you can’t turn it off and so on. Hiring a VPS is usually the best solution.
You have to literally program all your self. Also for example some program that shows you the performance of your strategy not to forget that you must first log all your trades to begin with.
All this makes quite some work as you can imagine.
06/03/2018 at 11:48 AM #72103@buffster76: I also agree very much with Vonasi that you should focus on longer timeframes, 1h and longer. I personally prefer the daily timeframe. Somehow it is very common that people that are new to also trading focus on short timeframes (I was the same) because you somehow feel like it is easier or safer but after some time everybody seems to end up using longer timeframes.
Like Vonasi points out there is a lot of noise on the lower timeframes and it is a real pain in the *** to backtest low timeframe strategies. It is very slow and also not unlikely that you will see results that don’t match reality because everything is much more sensitive to execution details.
And also for your questions regarding matlab. This is even worse than with PRT. Of course it is possible to feed tick data also to your strategy in matlab but this is really cumbersome to say it mildly. I never even tried this so far. All my matlab strategies use only daily bars so far.
06/03/2018 at 3:23 PM #72121thanks @vonasi and @despair your advice is very useful. I am new to this (4months now) and I started with 1min data as I thought (perhaps incorrectly) that this would help me catch changes in trends more quickly compared to using a longer time frame. I have gradually been realising that this is not true, and have been migrating to longer timeframes (5min, 15min) realising these are just as good at catching longer trends, so thanks for the advice, I will jump straight to 1hr and modify my strategies. On a related note, and this question will also help save me a lot of time, do you think it’s a realistic expectation (honestly) to make 30-40% profit annually after 1-2yrs of trading experience? The reason I ask is that you both seem to be experienced (and very helpful) and this would potentially save me a huge amount of time and help me set realistic expectations. Is it a realistic possibility or a dream? I am looking for additional income to help support a family move to a country where salaries are much lower than the UK, and if I could develop a reliable strategy (or set of strategies) and semi-reliable income from trading, then this would open up possibilities for me. Feel free to be as blunt as you like, it would save me a lot of time in the long run. thanks again for your help.
06/03/2018 at 4:44 PM #72125@buffster76: 30-40% annually is very good, more then the best CTAs (commodity trading advisors – full-time professionals) make on average. It happens in a good year but the average is significantly lower. Having this said, profitpotential is a function of the risk your are willing to take. CTAs have to play reasonably save because they can’t expose their clients to too much risk without losing them. Managing your own money you can take higher risks (if you want to) but of course also have to stomach the consequences when things go the wrong way.
So yes, it is definitely possible to make 30-40% annually but this is a lot and you will have to take some risks to accomplish this. Well and this risk can also unfold in the wrong direction. You can lower your risk to some extend through diversification but you can’t get rid of it (there is no free lunch). A setup that for example allows you to earn 100% per year has surely also the potential that you incur a drawdown during this year of at least 50%. So don’t “play” with money you need to pay your rent.
I for myself like to take higher risks BUT I do so only with a fraction of my money. So maybe my trading account makes say 200% in a year but this were 200% of just maybe 20% of my money. So after a year I have maybe just earned 40% (and not 200%) profit on my total account but I also never risked more than the 20% I put in the trading account to start with.
Another thing to keep in mind is that this goal will be harder to reach after 1st july because of the new ESMA rules restricting leverage to about 1:20 and not 1:200 or whatever we have now with IG.
06/03/2018 at 6:59 PM #72135wow, I switched to a 1hr timeframe and adjusted the averaging period and running with 5k instead of 100k points gives very similar results but runs MUCH faster (10secs instead of 3-4mins) – excellent, thank you both! I will be able to make much faster progress now. FYI I’m playing around with a double MACD strategy, using averaging periods of 1day, 20days, 40days. Seems to catch trends easily. And I’m using position sizing using the positionperf indicator which works ok. Target profit allows me to increase position size mid-trend. Position sizing makes a big difference to results. Makes more profit when things are going well, and keeps it low-risk when things are not. I’m also playing around with analysing the equity curve but not so successful yet. I will start reading about this on other parts of prorealcode. I think position sizing is the way forward.
@despair thanks for your advice about profit levels. I guess the new ESMA rules will mean that to make the same amount of profit, I will need to put more into my account (reduced / increased margins). I’ll need to backtest my strategies with reduced leverage to see what I might be able to achieve. And yes I won’t use more of my savings than I could live without. During backtesting there are times when the strategy loses significantly before it recovers. I guess in real life this would be a stressful and depressing time, so I want to develop position sizing to try to minimise these losing streaks.thanks to you both – I will be able to start making faster progress now 🙂 and I’ll park the Matlab project for later
1 user thanked author for this post.
02/18/2023 at 1:21 AM #209877Dear Ulrike,
I notice there were some discussions before about strategies getting stopped with IG being the broker. I am experiencing the same issues and would appreciate any assistance. I did approach Ig via the PRT Help center tab but the response I received is very general and not helping.
The following occurred.
I setup two different strategies – one each on USTech 100, US500 Cash. Minimum order quantities as per IG was inserted (2) and (10). Both strategies failed when the positions was attempted to be opened. The following errors was given. Date and time of failure – Friday 18/02/2023 – 09:30, 09:45, USTime
The ProOrder Autotrading screen refer me Order List for more details – Order list give following for both strategy failures – Order rejected by the broker or by the market(10 times in a row). No further information is given to assist with the problem. I rechecked all positions sizes against IG requirements (was correct). No stop losses were used in the code so minimum distance from market as required by IG was not relevant. I have tried to relaunch the strategies a few times after this with the same result.
I had the same issue the previous week with another strategy with the same reason for failure – Order rejected by the broker or by the market (10 times in a row). IG response to this as per below:
Questions:
- Is it common from your experience for IG or PRT to do testing on the same demo server as customers are using. I have to pay AU$ 40 p/m to use the PRT service via IG and then it cannot be used reliably? The concern for me is that if I start doing live trading that I will have the same issues with strategies just being stopped and rejected.
- The response from IG below is very general and intended to close the topic. Is there perhaps another reason for the failure as I note you cover the reason in your response above: – An order placed by the trading system is rejected by the Broker, or the broker’s servers do not respond to an order placement, modification or cancellation request 10 times in a row.
- Is it common for brokers or PRT to take a service down without notifying customers that a system will be down between certain hours?
Dear Pieter,
Thanks for your email.
Sorry to hear about your issues.
I am afraid we do demo tests from time to time on the demo server, during such times, there may be a slight impact on performance which could cause the issue that you have encountered.
Such tests are important for us to test bugs and make improvements before launching them into the live environment.
Typically these are short tests, which last for few minutes but could also extend depending on the type of tests that it was done.
Hope you can understand.
If you have any further queries, please do not hesitate to contact us.
Regards,
Michael Ye
Trading Services02/18/2023 at 7:22 AM #209883Hey there Pieter,
Forget it. Nothing will be solved ever. Btw, there’s a topic with annoying things in which I / we mention all the variations of Stopped Systems. There are more reasons then you see in there (I often don’t have the time to post them) and all I can say is : it just is so.
- Learn to live with it;
- Know that Demo is much much worse than Live (trust me);
- Invest your time (and some times little money) in finding out the cause, knowing that most can be avoided once you know what the cause is. It takes several months throughput.
The most annoying is the Stop Distance messages because they are plain and idiot IG “bugs”. You can’t do a thing about them. They often happen at similar times of the day (mind you, what you see below is Amsterdam time).
With these errors, as far as I found, this is IG going over to “manual mode” and what they do is deliberately set this stop distance that far, so nothing will happen in their systems (all is rejected) so they have more easy times with hedging their own stuff. The ridiculous part is that it is PRT which can not cover for it. I will admit : this is not easy to do (how to recognise the situation), but meanwhile they just don’t deal with it. The number of tickets opened for this (by me) is countless, but they just don’t do anything about it anyway. My last ticket is from I think two weeks ago.Tbh I had to laugh when reading the first couple of posts by the topic starter. Totally nothing has changed since, and all we can (better) do is smile and live with it. But key is indeed : the time investment is ridiculous and you can’t go back. Thus control it. It really is the best advice (me thinks).
And oh, you may have far away hopes for IBKR and Autotrading; assumed this will work out after all (re your other post elsewhere), you will have very similar problems there but worse. Why worse ? well, because IBKR is well “under control” of PRT and PRT can’t accomplish one single thing with them. Here too it is IBKR who messes up, but it is soooo easy to counteract the issues (again, trust me) that it is completely visible that just nobody does a thing. I would count my number of tickets regarding IBKR and this area of faults on … over 200.
The worse part of this (IBKR) is that now, with AutoTrading live the very same issues happen which also happen with Manual Trading (status of last order is unknown and that kind of sht). With Manual Trading I learned to live with it (easily), but a newcomer will lose money over it. With Autotrading it is different, because now the System is Stopped once again. Same thing as with IG but worse because it goes wrong more easily with IBKR.No Ulrikes need to write and no Elviras need to be blamed. They are only puppets who are actually in the same ballpark as we are. The difference could be that we pay for a service and that they are payed to keep us satisfied.
Despite all this I am still having fun. One day though, this will go wrong and backfire to them hugely. This is because one day someone won’t lose 1000 over such a situation, but a bit more. That will be a nice day for a lawsuit. Despaired people will do that.
Again : it is annoying but don’t be annoyed. Live with it.
02/18/2023 at 9:28 AM #20988502/18/2023 at 10:11 AM #209886I had 18 Systems stopped yesterday
I’ve just found another / additional 19 Systems stopped … across 2 x Demo Accounts!
All were for – – ‘exceeds the number of retries’ .
1 user thanked author for this post.
02/20/2023 at 3:58 PM #209986Ok, thanks all. IG contacted me today to inform me the server was down on Friday night into Saturday morning for some work on there side. As PeterSt comment, IG also advise that Live accounts will apparently be more stable. I did not know about it before but there is a web link that update you when IG servers is down. With this I can at least confirm or it is IG related if something go haywire again. Below for other users to use.
1 user thanked author for this post.
02/20/2023 at 7:33 PM #209995IG Server maintenance regularly occurs after market close on Friday and is complete well before market opens on Sunday at 23:00 UK time.
IG Server maintenance was not responsible for my 37 System stoppages as they all occurred when the market was open on Friday morning 18 Feb 23.
1 user thanked author for this post.
02/21/2023 at 8:15 AM #210013Another 9 Systems stopped late yesterday or overnight, 8 were for ‘exceeds the number of retries’ … this is the dominant Rejection reason for me by far and it seems to be getting more frequent recently!?
Soo annoying as it leaves me wondering, if the trade had executed … would it have been a loser or a gainer!? 🙁
It’s the ‘same old same old’ since 5 years ago when this Topic was started?
Is the source of the problem that IG Servers can’t cope?
If PRT were to extend the time between retries and / or extend the retry period for longer … would this help?
Isn’t this why we have Demo Systems … why not give above a try PRT PLEASE? Or at least try some fix or other on a trial basis PLEASE??
-
AuthorPosts
Find exclusive trading pro-tools on