System Trading Blog

Make Your Money Work Harder

Archive for October 2009

Zulu/Saxo Live Account Test

leave a comment »

I opened a test account with Saxo Bank to trade it through On Zulutrade there are probably thousands of trading signal providers who compete globally, they have all the motivation to perform and everyone with an eligible brokerage account at any time can add or remove their signals by just a few clicks. Saxo Bank is a licenced and regulated Danish bank, deposits are covered by deposit guarantee programm and as such it probably meets the requirements of even the most conservative investors.

The quick observations are given below.

There are major issues with reporting
The equity figures just don’t add up. Equity figure looked strange throughout the week. At the end of Friday I closed all the positions, but the equity value continued to fluctuate. I contacted Zulutrade, but they suggested to contact the broker as the equity updates they get from the broker. Zulutrade couldn’t offer any valid explanation why in an account with no open positions current equity is so different from what it logically should be (i.e. initial equity + P/L)..

Another telltale of reporting saga — a record in the history log shows 111 pips profit while the corresponding EUR value is 0.

Saxobank doesn’t respond to service request
Incidentally, I can’t perform first-time log-in into Saxotrader account with the password as provided by Saxo Bank. A few days ago I’ve sent two inquiries about the issue, but there have been no response so far.

Brutal slippage
The promotional e-mail said “2 pips spreads”. What I reasonable expect is 1.5 pips added per round-turn to cover rebate to That would make the spread at 3.5.
At closing of the trade, the broker adds 4.5-5.5 pips. This is systematic beyond shade of doubt. Subsequently, the actual spreads are 7-8 pips or about twice what could be reasonably expected.
As a real-life example: if there is a sell stop for EURUSD at 1.4845, you’ll get a fill at 1.48405.
The implication is — if you autotrade a system with, say, average 20 pips profit per trade, the real-life profit is likely to be 10-11 pips due to slippage (i.e. the expected performance will be just about half of Zulutrade reported performance).

Low Leverage Level
Zulutrade has a built-in risk evaluation tool called “Margin Call – o – meter”. This is a good thing as it helps to evaluate risks arising from excessive leverage and, as the name implies, helps to avoid margin call (i.e. liquidation of open positions by the broker).

For most of the signal providers, 3-4 times leverage can be applied before the “meter” goes into the danger zone. There is no reason to question the meter’s accuracy and, consequently, customer can expect to trade at about 3 times leverage.

Equity value not too informative
The equity value is constantly updated and displayed in Zulutrade “Positions” section. Unfortunately, value of funds tied-up as margin for open trades, are deducted from the equity value. This is a poor practice as it treats margin as a loss while the trade is open. A more correct name for the “Equity” line would be “free margin” or “unused margin”.

Does anything of the above came as a surprise? Not exactly. It was well below expectations. Technical risks in autotrading are still considerable and not only on Zulutrade. Does it somehow devalue the autotrading as an investing method? Not at all. The concept is very sound. Services like Zulutrade are very efficient in sourcing trading talent globally and, effectively, allocating capital to profitable traders and their trading systems. This is the future of investing and technical issues will be sorted out sooner rather than later.

Written by A.S.

October 30, 2009 at 10:37 pm Gets Facelift

leave a comment » (aka C2) gets redesigned. Yes, its more sleek and nice now. More importantly, its easy to use and very practical. Browse it and use it!

Written by A.S.

October 27, 2009 at 9:56 pm

How Smooth Is A Realistic Performance Curve

leave a comment »

Small account traders gravitate to smooth perf curve as mosquitos fly to a source of light at night.

Usually smooth performance is either short-lived and not sustainable in long-term.  LTCM is example of very smooth performance curve in its early years.

What is realistic? In my experience, in futures markets one can obtain something like 1:3 risk/reward ratio. In other words, to make 30% annually, one must be ready to be down at least 10%. In forex markets which are more competitive and recently act as a poster child of random walk, 1:2 risk/reward ratio is more likely.

The most risky proposition is to start to use some system agressively after it has had high, say 1:100, risk/reward ratio. The chances are that the customer will get burned once the system “adjusts” to more sustainable, say 1:3, risk/reward ratio.  My suggestion is — trade the smooth-curve system, but do not overleverage it by blindly trusting past performance data.

Written by A.S.

October 27, 2009 at 9:42 pm

Posted in trading systems

Java as Just Another Foreign Language

leave a comment »

Computers have become so pervasive in our life that what some time ago so different things as library (and need to go to it), newpapers (to read news), music player, televison, typerwriter  (hmm, once upon a time there was such a thing …), telephone, photo album, games ….  have basically replaced by one relatively small thing — computer.

What’s more, programming languages have evolved in something probably a non-geek can use already. If one needs a non-standard programm such as automated trading system “plugged” into a broker platform through an API, I can imagine that soon no more than just and average IQ will be needed to develop such programm.

To code trading systems, one has to be competent user of at least one programming language. It is not easy to learn a programming language, but my guess is that it is not more difficult than to learn a foreign language.

I myself can comfortably comunicate in four languages and have a basic understanding of another three. The next “foreign language” I’m likely to learn is Java. I guess it’s more usefull use of my time than to improve my French or Swedish or German.

Why Java? C++ is too complex and slowly gets obsolete, C# is usable only in Windows environment (unsure for how long Windows will remain relevant), Cocoa for Mac OS (hardly anything trading-related runs on Mac).  Write your software in Java and run it on basically any OS.

I choose to put some extra effort in learning Java rather than limit myself to EasyLanguage, C# or Cocoa.


Written by A.S.

October 27, 2009 at 9:29 pm

Posted in Uncategorized

Recent Twist of Turtle Brand: EA for $299

with one comment

I’ve a confession to make: I don’t believe that so called “Turtle Trading Rules” work anymore. I’ve tested them myself. But the “brand” still gets promoted …

Recently I spotted an offer to buy Turtle Trader Expert Advisor (EA) for $299. This is so 80’s. I wouldn’t buy it and do not suggest anybody else to buy it. Not because the number of trades is too low to be statistically significant or because I’ve done my own back-testing and have all the reasons to question the performance claim.

But just because today none should trust “packaged” trading system product.

The way it’s done today is you put your system (if its any good) on or  or for independent verification and ranking.  Then prospective customers can evaluate it and, if they like it,  subscribe to it or buy EA.

Written by A.S.

October 26, 2009 at 11:34 pm

MetaTrader and TradeStation for Autotrading

leave a comment »

From all the back-testing software I’ve had chance to look at there are only two contesters for the 1st prize — TradeStation and MetaTrader.

Both have evolved as automated trading platforms. Both were somewhat inadequate at the very beginning. Online trading technology has progressed exponentially in the last decade. I remember placing the first trades online through Lynd-Waldock account — it was a big convenience for me as I didn’t enjoy the chore of calling the broker on the phone. At about the same time I started to use TradeStation 2000i and some software version from the same company (Omega Research, later Tradestation Securities) — for both versions I spent a couple of thousand dollars as far as I can remember.  Incidentally, made more than that from one trade with the vendor’s stock during internet stock boom.

Soon Tradestation changed its business model to platform subscriptions, introduced online trading functionality and started to offer brokerage services (in fact, mostly in capacity of introducing broker).  The first autotrading experience on Tradestation was with mixed emotions — too often I had to adjust positions manually and the platform rather than internet connection was to blame. However, Tradestation got is straight fairly soon and the last time I used the platform (Y2007, I guess), it worked just seemlessly. Nevertherless, I haven’t used it since and unlikely will use it again.

The reason? There several reasons: EasyLanguage, competition and evolution of programming. EasyLanguage was a great concept — it allowed to write trading systems in almost plain English with very few hard programming rules to follow. The problem is that EasyLanguage you can use only on Tradestation and it is somewhat limited in comparision to general programming languages. Competition has evolved and probably surpassed Tradestation Securities as once near-monopoly autotrading platform provider. Today every major broker supports autotrading by providing it’s own trading platform or by ensuring direct access interface.  Regular programming languages have become more easy to use and more traders have acquired coding skills in languages like C, Java or C#.  Besides, code written in regular programming language can be used with various brokers and platforms or integrated with other systems while EasyLanguage is restricted to one broker, one platform and is not directly compatible with anything else.

MetaTrader 4 platform has gained popularity over the years. It lacks the polish of Tradestation, support documentation is somewhat clumsy, but it is free to end-user (vs $250/mo for Tradestation), adopted by many brokerage companies (vs 1 for TS), programming is in MQL4 language which is similar to C programming language.  Meta Trader 5 will scale up the looks, ease of use and, more importantly, add some adequate trade execution  commands with MQL5.

I’m currently do my system coding work on MetaTrader. The next step (5 years from now?)  is likely to be that all the sytem coding and back-testing work will be done directly in a language like Java or F#, then directly plugged into broker system with no need for intermediary platform like MetaTrader or TradeStation.  And code in mql4 (or mql5) will be easier to translated into, say Java, than EasyLanguage …

Tradestation is developed by a Tradestation Securities in Florida, United States. MetaTrader is a product of Meta Qoutes Software, apparently a Russian company with offices in several countries.

Written by A.S.

October 26, 2009 at 11:00 pm