Page 1 of 2
Costly Robobroker Errors
Posted: Sat Jun 03, 2006 1:02 pm
On another thread I warned of the frustrations of using a systems assist broker.
As a result of my robobroker entering all contract months into CSI UA for Crude (instead of HMUZ as instructed) a false signal was taken. A loss was occurred when reversing the trade, which cost me over $3,000. The broker refused to accept that it was the fault of his firm and stated that such a loss was to be expected in the â€œnormal course of tradingâ€
Posted: Sat Jun 03, 2006 2:16 pm
My dad always taught me that if you want something done a certain way. Doing it yourself is the only certain way. Whether it is correct or ideal or whatnot... I have come to accept that no one, will look out for my best interest the same way as I can or will. So I manage my own order placement. It was a bear at first, but with some practice Ive worked the nightly game down to 90 min. Sometimes I have to stay up till the wee hours if a London or other order is very close to the market to babysit the process. I look at those times as an opportunity to perfect my espresso making skills and to watch some good films on the big screen.
Posted: Sat Jun 03, 2006 8:57 pm
Absolutely agreed with RedRock, never trust anyone to place your trades. From what I know you're a Long Term trend follower, placing an order once/twice a month isn't a big deal
Posted: Sun Jun 04, 2006 8:18 am
you should keep on posting your trading experineces with the broker, maybe other traders will thank you for preventing them from making the same experience... and of course the forum lives on postings, no matter whether everyone agrees
I also absolutely agree with RedRock
"no one will look out for my best interest the same way as I can or will"
Posted: Sun Jun 04, 2006 9:36 am
As a trader who input all trades myself, I would like to point out that it is inevitable for errors of some type to occur, even if you input all trades yourself. The main difference is I do not have to pay a higher fee for the robobrokers.
Some possible "errors" even if you input all trades yourself:
1. overlooking the first notice day etc because of public holidays (sometimes early closes) in foreign exchanges. Even in US alone, different exchanges have different holiday schedules for the same public holiday. I trade products from exchanges in Far East and US, and I do overlook this from time to time. As an example, until a few years ago, HKEX do not list half-day holidays (early closes) in their holiday schedule. At one time I made a complaint about this to them and their answer was " this half-day schedule was communicated to all brokers and your broker should aware of this". Now they include half-day trading in their holiday schedule calendar.
2. Untoward happening do occur - typhoons, rain storms that cause exchanges to close suddenly, and power black-out, computer faults, communication faults etc. As a person in a different continent you may not know this until after the event.
3. For some contracts - (bonds etc), CSI OI don't roll until after the first notice day. Even if you personally anticipate the roll and wish to do it yourself, you may have to stay up very late to wait for the close (if you are trading the electronic section) because you may have set the "Close Old Contract, Close new Contract" in CSI. If you trade exchanges round the clock, it is impossible for you to stay up until all closes of all exchanges you trade. (Note: for electronic trading, the broker "tried" to input your order as close to the closing time as possible but I don't think they could be held liable if the order is not filled)
These are only some of my personal experience. I would like to know how other traders work around these problems.
Posted: Sun Jun 04, 2006 10:41 am
I found this topic very useful and I would appreciate anyone sharing their personal trading experience with us because it clearly would help us on the road to be a better trader. Negative comments- well, I do get quite a lot of them from time to time. Sometimes they are constructive, other times they are not. I posted things when I think I should, even though I know they are controversial or may not be popular.
As for trying to roll earlier, sometimes it is not possible without unduely differing from the trading system. See the attached- which is an example of my keeping track of some of my contracts to find a better time to roll. As the number of trading vehicles getting more and more, it is getting more and more difficult to roll earlier without differing from my systems (at least for me). Adding day trading to all these and you could see why I sometimes make errors here and there.
Posted: Sun Jun 04, 2006 11:02 am
I have tested for fixed dates quite a lot in the past. It would have been absolutely no d**n problem for me what method the broker chose - he suggested OI, not me. I was relying on his diligence and expertise to keep me out of trouble.
In my case I have been able to satisfy myself that in the longer term different roll dates or algorithms do not make too much difference to my system's profitability. So I will be happy to use fixed dates.
What I was concerned about was to use in CSI the same roll methods the broker was using on the ground. Failure to do so leads to ridiculous and irritating distortions. On stops, entry and exit triggers and so on. Over the past two months comparing rolling on fixed dates and OI has produced the same trades in the same futures but differing entry dates and hence (at least in the short term ) different P&L.
Posted: Sun Jun 04, 2006 11:23 am
If for nothing else, I think sharing some of our frustrations with other traders would help reduce the pressure of trading, so keep posting.
Your experience with back testing for fixed dates is quite interesting. If possible, can you tell us more - do you use a certain days in advance of first notice day for all contracts or do you use different days in advance for different contracts etc? Personally, I find some contracts (e.g. Hang Seng Index future of HKEX) have little volume and large bid/ask spreads for non-front month contracts until very near expiration and very high premium/discount also.
Posted: Sun Jun 04, 2006 12:51 pm
I agree that a thread on robo-broker errors definitively can be useful for system traders. No question about it.
mike 168, why do you make your life so complicated by choosing a roll-over procedure that is so difficult to implement correctly?
Personally I don't trust a trading system when its edge lies (solely) in the particularities of the roll-over algorithm. I agree that some procedures can add a little extra (at least theoretically), but in practice this is very difficult to cash in.
(Again) personally I prefer to only consider roll-over routines that are simple and comfortable. I try to be pragmatic. A good night sleep maybe allows you the next day to research a new and promising trade system idea. And don't forget that you will have to enjoy trading your long-term system for the next 10 or 25 years!
Posted: Sun Jun 04, 2006 1:17 pm
The reason I chose OI as the rolling criteria is for historical reason. I used to trade Aberration and subsequently 3 black box commercial systems, all of them use OI as rolling criteria. Hence I follow this in order to be consistent with the system signals and results. For reasons I pointed out above, I think using a different rolling criteria would impact on system performance. I intend to back test this on my TBB systems in future. However, instead of re-inventing the wheel, I would like to know what others have tested and found to be a practical alternative.
Posted: Sun Jun 04, 2006 1:51 pm
For a start take a look at http://www.pinnacledata.com./clc.html
. Its a good starting point and the dates are designed to avoid the "nasties" - termination of trading, first notice days etc. Without exception all these dates are based on trading front months rather than anything more exotic. I'm probably using rather loose termination here.
Some traders have found advantage in trading short term interest rates further out - many months further out. See Chuck Branscomb's posts and, I think, Sluggo's for more details. Test trading months 4, 5, 6 even 10 months further out in the very liquid markets. Clearly in such a case you have to use fixed dates.
Some have also found trading non standard moths for the energy an advantage. It keeps the noise down and the stop further away. Sluggo posted on trading alternate months. For crude some have taken the route of trading December only for instance. I don't trade NG but it is interesting to look at backtested results on a very long term version of Aberration using December only. I seem to recall the volume was sufficient. Again, essential to use fixed dates for these non standard months.
Posted: Sun Jun 04, 2006 4:57 pm
I am trying to understand your problems, really I am, but I just don't get it.
You trade a long term system, so I guess you don't trade each market that often. I trade a shorter time frame and can happily go days without putting an order in, as there is no trade approaching a breakout, hence no signal.
When I do place the order(s) I phone them in, takes about 2 mins, with an "if done stoploss", job done.
Never had any problems with either the accuracy or the execution if they are triggered.
In fact I would go as far to say that my trading is incredibly boring, as there can be days without any markets breaking out. (I do only follow 18 markets though).
I reckon the time it takes you to create the above commentary, would equate to a whole trading week for my system (
Good luck, I hope you or your god can sort ot out!
Posted: Sun Jun 04, 2006 5:08 pm
It is very, very , very, very simple. I do not trade my own system. None of my time, I repeat NONE of my time has been taken up by trading. My time has been taken up in setting up the ROBOBROKER to trade my system FOR ME. So that I need not place an order, move a stop, even pick my nose or my butt if I do not feel like it. So I can go to hospital, do a job, run a brothel, be a don, become a monk.
My time has been spent in monitoring what the broker has been doing for the first few months, making sure that he receives the right signals, seeing that he has got the picture, the right contracts, the right darn months the right BPVs the right everything. My time has been taken up in getting him to understand CSI UA, Trading Blox.
It is simple for you, for me, for most of us. But not so for the said broker apparently.
Posted: Sun Jun 04, 2006 7:22 pm
learning from other's experiences is very helpful, so posting such an issue is not doubt only an another confirmation for me of what I used to think for my self:Do it all by yourself and check where you were filled.
As I see the picture, only if you have 0$ costs to trade and have asystem that trades on 5 min data then it's worth it, otherwise the one would become a "robot" to do it all by himself.
I was rereading New Market Wizards and Monroe Trout shines as an execution freak example, for him half a tick matters. I am biased no matter what size the one trades it should be taken as seriously as Trout does it.
Posted: Sun Jun 04, 2006 9:02 pm
Thanks AFJ for your link above. That's a good starting point for my back-testing.
Posted: Mon Jun 05, 2006 10:40 pm
AFJ Garner wrote:I have tested for fixed dates quite a lot in the past.
In my case I have been able to satisfy myself that in the longer term
different roll dates or algorithms do not make too much difference to my
system's profitability. So I will be happy to use fixed dates.
Today is a specially good day to talk about calendar based ("fixed date")
rollovers. Although Pinnacle Data Corp gives the impression that it's
reasonable to roll over all markets in the same group on the same day, it
isn't true. Today's example is the Currency group.
Pinnacle Data rolls over the US Dollar Index ("DX") futures on 8th DM,
i.e., on March 8, June 8, Sept 8, Dec 8. But this is a grievous error.
Unlike many other currencies (Aussie$, JapYen, SwissieFrank, Euro,
MPeso, British) that are traded on CME, the US Dollar Index is traded on
the NYBOT. Some months it has a very different FND and Expiry day,
than the CME currencies. This can hurt you.
Based on bitter experience (losing real money), I roll over the NYBOT
currencies on 6th DM rather than 8th DM. Tonight I am emailing my
rollover ("spread") orders to my broker, to be executed at market on
the open of tomorrow's (06/06/06) session. This includes CSI syms DX,
EJ, GB, RZ, SY, and other more exotic "cross" currencies. How would you
ever find this out? How did I find it out? By making an assumption,
trading that assumption, getting a bloody nose, and adapting afterwards.
Sorry but that's how it works. I guess you could hire a consultant who
claims to have gotten lots of bloody noses already, but guess what?
Pinnacle data hired a consultant who helped them choose their rollover
dates and they still got the NYBOT currencies wrong. Pause and consider.
Posted: Mon Jun 05, 2006 10:46 pm
Keep in mind that when using CSI data you can send your broker your file called portfile.adm This file is located in c:\ua\archives and when imported by your broker will exactly replicate your CSI environment on his computer. This eliminates symbol and or roll over differences etc.
Sharing this file with my broker has virtually eliminated differences due to data issues.
Posted: Tue Jun 06, 2006 3:02 am
Invaluable advice, really. Thank you. I have set aside today to comb through every avenue I can find (exchange websites, broker websites, holiday calendars etc etc) to TRY to get the best picture I can about such difficulties. I shall no doubt need many such days. I have absolutely no doubt whatsoever that just like anybody else and just like the hired consultants you talk about, I will still fail to get the full picture.
Thank you. Yes, quite agree. I thought of this one. In fact, I got the broker to send me his portfile.adm file instead. Unfortunately, I use CSI UA 2.8.0 while the broker uses a slightly later version. Neither his version or mine are still available for download.
I could not load his file into my CSI UA. I checked his text file line by grisly line. One thing I did not want to risk was upgrading to the latest version of CSI myself. I had enough irritation on my hands, and CSI themselves warned me that many people were having issues with the latest version. Maybe his version of UA would have had backward compatibility with my file.
The futuresinfo.txt file contains futures data and is used by Trading Blox to read BPVs, tick values, data file names etc â€“ in case you do not own TB.
I did at least send my futuresinfotxt.file to the broker. Clearly I should have just insisted he used my file.
But the broker was at first apparently attempting to use one CSI portfolio for a number of different clients I believe. I only discovered this fact over the course of several weeks. He was also therefore using 1 futuresinfo.txt file for several different clients. Again, I only discovered this in the fullness of time.
A wise forum member (let him be called â€œLukeâ€
Posted: Tue Jun 06, 2006 5:22 am
I would only take the Pinnable as a suggestion only and would advise studying the actual best rollover day ourselves. For example, it advise the Nikkei rollover to be 3rd DM. I have attached a 5 min chart for 2 days of SGX Nikkei (SGX equivalent) which shows that up to yesterday (5 June), the trading in the supposedly rollovered Nikkei is sporadic and the bid/ask spread wide. It's going to increase slippage if the Pinnacle rollover dates are followed without adjustment. A look at the volumes of the resulting contracts also showed some volumes are too low to be tradeable immediately after rollover, depending on what you have set for the "minimum volume" in TBB.
Posted: Tue Jun 06, 2006 5:52 am
c.f. or Tim noted somewhere how one can avoid making the screen unreadable by inserting graphics which are too large.............dunno whether you saw that? But the charts are making this thread a little difficult to follow.
Re roll dates, I do not approach this question from the backtesting angle alone.
I approach it first and foremost from the position of keeping away from the "nasties", last trading days, first notice days etc.
Only then do I consider whether there is any advantage to be gained (liquidity allowing) in trading non standard months or dates.
As I think I pointed out somewhere else, because of backwardation in Crude Light, historically at least, theorectical advantage may be gained in rolling on the very last day of trading, as the frequently existing discount on the expiring month has to come into line with the spot price.
Few would wish to trade this way in practice I suspect.
By way of example, try choosing non standard months in CL within CSI UA and set the roll for OI. One advantage of non standard months seems to be less noise - to add to the greater roll premium advantage. You will often find apparent benefit in this in backtesting. Inspect the CSI output however and you will find the roll very, very often occurs on the last trading day.
If you then use the same months but choose to roll on some fixed and earlier date (11th of the month prior to the delivery month by way of example) you find, in may cases, that some of the profitability and advantage has disappeared.
Of course, I may well be talking rubbish. My conclusions may be erroneous. But those are my thoughts.