Slow order management automated trading 1 second timeframe

• 01/20/2024 at 10:13 PM #226633
  GraHal

  pointofbrew wrote:

  There is indeed a language “problem”, and it’s hard for me to express myself well in english

  From your written word, you are expressing yourself extremely well in english … I understood everything you wrote!

  Had you not mentioned a language problem, I would have thought you were a native english speaker pointofbrew

   

  

  GraHal

  Participant

  Topics: 204

  Replies: 8433

  Been thanked: 2048 times
  01/20/2024 at 10:35 PM #226635
  pointofbrew

  Really? Well thank you! I guess it might be more of an issue to me than everyone else. I just feel it’s hard to express nuances and specific technical details in a clear and concise way in a language I rearly get to practice. E.g. I see a few missing pieces in earlier posts that I wish I could have edited.

  Still struggling with brainfog from covid though, so maybe that’s to blaim 😀

  1 user thanked author for this post.

  GraHal

  

  pointofbrew

  Participant

  Topics: 4

  Replies: 22

  Been thanked: 5 times
  01/20/2024 at 10:59 PM #226636
  JohnScher

  I have up to 8 systems 1H, 4H that start a trade on Monday morning 09.00 in Dax40. Do not ask about the execution. Only rarely is there a system that starts trading at 09.00.00. 09.00.01-09.00.07 seems to be the norm. Spread is nothing compared to that.

  2 users thanked author for this post.

  GraHal, PeterSt

  

  JohnScher

  Participant

  Topics: 44

  Replies: 336

  Been thanked: 84 times
  01/21/2024 at 3:39 AM #226638
  PeterSt

  Dear pointofbrew,

  Every single screenshot I showed is from LIVE.
  I have 20-40 Auto Trades each day on IG. This means ~12000 since the time I started all, November 2022. None shows a delay that I can find, unless I do something wrong myself. Just in case, I am again showing a bunch of trades, the last which occurred. Nothing specially select to show a case or my righty of you will. I count 27 trades on this (last) Friday. So I will try again :

  All lines with black in the rightmost column, are entries. All the “Koop Bestens” are Market orders, and they were filed in the exact first second of the minute (:00).
  All lines with green in the rightmost column, are exits. All the “Positie sluiten” are Market orders, and they were filed in the exact first second of the minute (:00).

  This is always the case as far as I see happening. Unless in erroneous situations, like coincidentally happened Friday afternoon (broker issues) when our systems are kicked out and other things go wrong. But this is not what we are talking about here.

  All the lines with black in the rightmost column (which thus are entries) showing “Limit” are pending orders which fill randomly throughout the minute. You can see that happening. These are “correct” orders. Please verify for yourself that you can see this happening in my picture below, OK ?

  All the lines with black in the rightmost column (which thus are entries) which would show “Stop” are pending orders which would be wrong. You do not see these orders in my picture because I don’t place wrong orders. 😉 Please verify for yourself (Internet, his very forum perhaps) that pending Stop orders should not be used for Entry, or else they will act as Market orders. And Yes, you can use them, but if used, it would testify of wrong usage. N.b.: I can do that too, in more complex situations and it requires a lot of reasoning when it can be done and what would be the results in which exact situations. Further hint : Use Limit orders in your testing, and I would understand what you are doing. Now a couple of brain cells tell me “error”.

  All the lines with green in the rightmost column (which thus are exits) which show “Stop” are pending orders which normally will show on a random second throughout the minute. Please verify for yourself that you see this happening. While at this little task, notice the two orders at 01:45:00 which don’t comply to this “rule” and which is not a coincidence. In such a situation – a Pending Stop which goes out immediately – I virtually placed the Pending Stop at the wrong side of the price, which happens when the trailing narrows in combination with spread (notice this is at night) and I push the Stop price through the price of the instrument so it ends above the price. This is “wrong” between large quotes and the order will go out immediately, despite it is a Pending Order. It is not wrong at all for me in this case, because (I think) I know what I am doing, plus the order needed to exit anyway.

  To enable some visualisation for you, I added the second picture. The white line shows the trailing – the orange line are “due” Market Orders which in this case (this sample-picture) cause the exit as a “Positie sluiten” in my first picture. This is not really related to your subject, but just pictures what could be happening and it may make “speak” my texts a little better.
  But, it may show what I meant in my last paragraph above, how the trailing is pushed through the price. Hence, the white line ends above the price and would that be Pending Orders, then right at the whole minute (in my 1 minute TF case) the position would exit.

  ---

  Of course you know all this because you are a trader. Still in your examples – in text and in shown code – you don’t comply to this really. I too think that your English is perfect (mine is not !), but like I said, the entities / phenomena you point out, don’t fit in my books. I could try to tell you that you may approach things wrongly, but that does not come across as nice. 🙂

  I could also say that a perceived delay of IG could be between your ears, never mind you could reason the why of it. It just is not so, or else my :00 times would never show. But they do – always where they should.

  If I would apply your test it would be a completely different test without opportunity for misinterpretation. And I will apply your test, if I can find a means that won’t lose my money. At this moment I still don’t understand a hoot of your means, would that be for Live (thus not Demo). And in Demo I will not test anything, because the outcome can not tell a single thing. Not to me. But then I need the rest and focus to do it. This could be right now, but now no market is open and during the week I tend to have a normal job.

  Can we please turn things around … Can you tell me what is wrong with what I showed and told about in this post ? Where do I miss things when I show my :00 entries or exits ? Does this perception of life (Live) harm reality of what you see ? Should I look at something else in my list of trades ?

  

  PeterSt

  Participant

  Topics: 71

  Replies: 2322

  Been thanked: 700 times
  01/21/2024 at 3:40 AM #226641
  PeterSt

   

  Yes, of course IG needs to “manipulate” its derivative, there is no need to explain this. But this happens at a different level than you reasoned it;
  I don’t know everything and all, but from what I see this by means of a “buffer”, like the 10 points or more in price difference to begin with. Within such a 10 points “band” they can do what they want such as “stealing” the spread when and how they deem it right. This means, for example, that they can take the spread at your entry, at your exit, or both (say “half”). This is not what I think – this is what I see (and video-record).

  If *I* compare IG with a normal direct stream, I see no delays. Maybe they are there, but I can not see them. This is because all is different. It is so different, that no any average what works for IG works for e.g. Interactive Brokers – which I better put the other way around because IG is “fake”. Example of what I mean without digging up pictures of it : when a real data stream – for example PRT-IB – shows 3 ticks in a second, IG would show 20. This is where IG’s manipulation happens. This is where they manipulate their own market. Where they hedge.
  And so you can’t speak of delay because you can not see it. And if a delay would be there, I still would not know how that would let end up my entries or exits at an other time than the data tells me (tells my program code what to do when). This is harder to explain, so never mind. It could lead to a worse price all right. But that is a different story.

  Let me end this again too long post with telling that I am trying to help. I am not trying to be right.
  Don’t tell others that I am also trying to learn. 🙂 At least it is my idea that the more I know about all this (inside) stuff, the more I can money from the markets.

  

  PeterSt

  Participant

  Topics: 71

  Replies: 2322

  Been thanked: 700 times
  01/21/2024 at 4:16 AM #226642
  PeterSt

  JohnScher wrote:

  I have up to 8 systems 1H, 4H that start a trade on Monday morning 09.00 in Dax40. Do not ask about the execution. Only rarely is there a system that starts trading at 09.00.00. 09.00.01-09.00.07 seems to be the norm.

  (emphasis is mine)

  Of course we should know about the execution, because it is all about that, right ? So … normal Market orders ?

  But John, your post makes me think of something very different …

  This is Eurex, right ? well, Eurex has its own order book and it is not at the broker’s. E.g. Nasdaq, which I showed in this post, has its order book at the broker. Thus, my order is given directly to for example Interactive Brokers. Or, in the case of IG, to IG. Not so with Eurex …

  If I am talking from my neck, let me know.

  Somewhere last week I talked about the extra 0,25 I need to pay (to Eurex) when I move a pending order. But, 10 moves are for free per running position. Thus, if my trailing would last as long (xx bars) as I showed in the post above, I am charged xx minus 10 times 0,25. This is because of the extra effort involved around this virtual extra step in the order process.
  Please note that while I got this data from IBKR (my AutoTrading for DAX and ESTOXX is at PRT-IB), I take it that this counts for IG just the same. It is Eurex which maintains the orderbook(s) and I would not know how IG could go around it.

  What I am now suddenly thinking is that this just takes round trip time. Maybe more than one communication back and forth. This is about asking, confirming, executing, and all the more stuff I don’t know much about. I wouldn’t know how pointofbrew’s observation wouldn’t be logical. Never mind I don’t understand his means of testing yet.

  
  @pointofbrew, did you test this same on Nasdaq ? if not, can you ?
  And please ignore my well meant advices for now. Just do your thing in the same way.

  ---

  And John, despite that you may see the same thing as pointofbrew does, your 09:00:00 thing is exactly what I was talking about, but inverse : at that time the whole world *is* trading and move markets (make that the whole of Europe). All Eurex stock markets open, and that may cause some delay in all the communications (interfaces) to all directions. I mean, you may see a different result on 11:00, or even better 11:07:24 (etc.). You will see the same on 08:00, 10:00, 15:30, 17:30, 22:00, 23:00, because these are all market opening/closing times. I know from PRT-IB what all can go wrong at 22:00, just because the US closes its stock markets (me trading Futures which stay open till 23:00). I even have excluded that moment from trading because IB can’t cope there (for Autotrading and in communication with the outer world (IB)).

  … But I rather want pointofbrew to be right … and that there *is* this delay.

   

  

  PeterSt

  Participant

  Topics: 71

  Replies: 2322

  Been thanked: 700 times
  01/21/2024 at 4:37 AM #226643
  PeterSt

  OK, not to testify anything, just thinking of what I can do for IG and Nasdaq, I can do too for DAX. Mind you, this is PRT-IB now.
  This is DAX Mini and DAX Micro respectively. Unfortunately the lists are short because I only started these on Jan 8.

  Notice that IB does not show “Positie sluiten” for Sell (Exit) Market. This is “Verkoop” here. “Koop” is Buy.
  Black lines are entries again. Green lines exits.

  In both small lists of 22 trades I see 2 trades not executed at the exact minute. So, tadaaa.

  Justin case I should add that the DAX Micro (DXS) is actually hardly liquid enough to autotrade with. But it still works fine. Anyway I say this because that could theoretically cause filling issues (no Pending orders to be found for my Market orders).

  ---

  And so I dare say on this too few data for justified statistics, that you have a point, pointofbrew.
  That it seems not really IG related, is something else.

  

  PeterSt

  Participant

  Topics: 71

  Replies: 2322

  Been thanked: 700 times
  01/21/2024 at 9:36 AM #226650
  GraHal

  PeterSt wrote:

  I mean, you may see a different result on 11:00, or even better 11:07:24 (etc.). You will see the same on 08:00, 10:00, 15:30, 17:30, 22:00, 23:00, because these are all market opening/closing times.

  I woke thinking the same as above, then read Peter’s post.

  It be good if pointofbrew could run his code, but with times not on the hour and also avoiding the times Peter states above. 

  Times could be set at exactly 2 minutes past the hour, or 13 minutes past the hour or similar, but no multiples of 5 mins, e.g. not 5, 10, 15, 30, 45 … etc mins past the hour, as these are all often used trading times.

  If times at exactly x minutes past the hour (excluding above) are used then it be easy to spot delays in the seconds figure.

   

   

  

  GraHal

  Participant

  Topics: 204

  Replies: 8433

  Been thanked: 2048 times
  01/21/2024 at 1:10 PM #226667
  JS

  For me, this is the “normal” operation of the order book…

  The order book depends on the supply of the buyers and sellers (liquidity) and sometimes it can take a few seconds for your order to be (completely) filled…

  1 user thanked author for this post.

  GraHal

  

  JS

  Participant

  Topics: 22

  Replies: 1434

  Been thanked: 561 times
  01/22/2024 at 9:05 PM #226737
  pointofbrew

  Hello again.

  I’ll have to come back to read all your posts and reply to them when I have the time. Probably towards the end of the week.

  It also seems I should let you know that I’m a novice. I only have 4 years of part time experience as a discretionary trader, and I’m completely new to full time automated trading. I have spent my last few weeks just learning the programming language, so that’s where I’m at. You all probably have much more knowledge of the markets and different platforms than I do. I had just began starting systems in LIVE when I noticed unwanted behaviour outside of my system conditions.

  Now that I’ve heard back from IG Trading Services, it may all have been a total coincidence and bad luck on my part. Without me being aware, I began starting systems last week at the same time IG/PRT was having technical issues. Let’s hope it’s nothing more. I’ll run my tests again later this week. In the meantime, have a good one.

  “Awaiting full  report,  but from the Technology team a certain server was confirmed to be reporting  some latency issues.”

  1 user thanked author for this post.

  GraHal

  

  pointofbrew

  Participant

  Topics: 4

  Replies: 22

  Been thanked: 5 times
  02/14/2024 at 9:15 PM #228202
  pointofbrew

  Hello again. A bit later than planned, but I’m too busy these days …

  I have spoken with IG and this was their response regarding latency:

  ” There’s usually no delay, the same execution on IG’s platform would apply in PRT, so should be with a second or 2 at most,  but it’s usually instant.”

  In my experience that’s way too optimistic. As I previously wrote, more than half of the time I suffer from 1 – 2 seconds latency, both when opening and closing orders. Sometimes much more. In most of my tests I’m not even executing orders. I just place random stop orders and pull them to test speed. Btw, I noticed users of cTrader (regardless of broker) complain about 200ms latency when executing orders. Their normal order execution latency (again, regardless of broker) is 50-150ms. So it’s quite interesting to see that no one in here complains about 1-8 seconds latency.

  Sad, because I was starting to really like PRT. But these latencies are nothing but crazy, and I’m leaning towards scalping, so … Plus, it puts IG in a rather bad light. I might as well call my orders in by phone.

  1 user thanked author for this post.

  GraHal

  

  pointofbrew

  Participant

  Topics: 4

  Replies: 22

  Been thanked: 5 times
• Author
  Posts