System Trading Blog

Make Your Money Work Harder

Archive for January 2011

News Trading Profitability

leave a comment »

My first exposure to ¨news trading¨ was almost 20 years ago when a former London trader explained one of his systems to me. Generally, his heyday has been in 80´s and his systems were simple. The ¨news system´s¨ design was about like this: after the news come out, measure the price move, then wait for X percent retracement from the price peak and trade for further retracement to Y percent. The retracement percentage values probably were Fibonacci figures. It made sense then as the retracements after the intial price move were quite consistent price behaviour pattern back then.

A more recent ¨news trading¨ method is to trade the news (economic releases, in this case) is to trade only if there is deviance from the forecast. If the news come out at or close to the forecast, no trade is placed. If the news come out with strong enough deviance from the forecast, the trade is opened within seconds from the news release. There is a price target based on statistics about historic price action following the release of the data. Some limited performance stats are available here. According to the system owner, the stats since 9/2009 are:  419 total orders prepared, 99 orders met the trade entry criteria, out of those 82 win trades  and 17 loss trades, total pip gain at 996 (or about 10 pips on average per trade). The system doesn´t seem to be packaged as a service, but there is a free application named Klik to open the entry trade automatically (you still have to exit the trade yourself) and support through skype.

 

Advertisements

Written by A.S.

January 31, 2011 at 2:09 pm

Posted in Uncategorized

System Trading Championship Results

leave a comment »

This forex trading championship took place from 4th of Oct till 24th Dec last year. $80k prize fund. #1 turned $10k into $77k, #10 in $33k. #4 even made his trading system code publicly available (his result was impressive $44k) here – look for link “files”.

A good illustration of system trading potential!

Written by A.S.

January 9, 2011 at 7:43 pm

Posted in Uncategorized

Which System Development Platform

leave a comment »

It´s been more than a decade since it´s possible to place a trade online. And nearly a decade since it´s possible to autotrade virtually any trading system through a publicly available software. Technically, today anybody with limited effort can design a trading system, backtest it, turn automated trading on and let the computer to do the trades in live account.

What choice of trading system development platforms do we have? The following list is far from exhaustive, but some of the most popular options are:

Tradestation. Easy to use. The programming language is the proprietary Easy Language. You can describe the trading system in a language which is quite close to describing it in English language. All major stock, futures/options exchanges and forex supported. However, you can use it only with one broker – Tradestation Securities. The software is free to use for brokerage customers, but there is a subscription-only service for $250/month plus data fees.

Metatrader. Widely adopted by forex brokers and traders. There is a proprietary programming language, but it is based on C (Metatrader4) and C# (Metatrader5). The manual rather insufficiently covers even the basic operations and some programming background is helpful. Available to end-users free of charge.

DIY in a general-purpose programming like C, C# or Java. This is the choice of big operators like larger funds.

Other options include: MatLab, X_Trader (Trading Technologies), Wealth-Lab, TraderStudio or Ninja Trader, eSignal, FXtrek, MCFX, . These software platforms, except for X_Trader and Ninja Trader, are not built for automated trading.

Which one is the best? For an individual without programming background or a small organization, I guess that Tradestation would do the job. It’s easy to use. If the strategy is simple enough, the chances are you’ll be able to program it in the shortest possible time with Tradestation. However, no portfolio testing is not available in Tradestation. If you’re looking for a super-simple solution (with limited functionality, though), then nothing gets more simple than Ninja Trader.

Technically, general-purpose programming language provides the most flexibility and versatility. But it is not suitable to everyone. Imagine you’ve a business proposition (a trading strategy), but you ‘ve to express it in fluent Chinese (a language you don’t know). So to make the proposition, you’ve got to learn Chinese first or hire an interpreter. The same is about programming languages – it takes long time to learn and master it, it’s not worth, if you only want to test a few strategies.

Written by A.S.

January 9, 2011 at 7:21 pm

Zulutrade Adds New Features, Some Old Problems Addressed

leave a comment »

By coincidence, the very next day after I mentioned in my blog post that Zulutrade has gone for about half a year without any changes to their service offer, Zulutrade introduced some new features.

  • Customization parameters to demo accounts: base currency, leverage and initial balance.
  • Provider notifications (new max loss per trade, substantial change of average trade time, stopped trading live account).
  • Provider alerts (high slippage and reckless trading warning).

The changes are rather evolutionary than revolutionary, but still highly useful. Zulutrade has a dismal record in the field of performance representation (as noted in a blog post here). So far it had pips and notional USD value as the reference points for performance evaluation – obviously insufficient parameters. Adding initial balance, base currency and leverage makes a lot of sense and is a big step forward for Zulutrade. As a side-note, extending free trial beyond 30-day period would be a smart move to keep potential customers interested and get more business. Empirically, 30 days is too short period to gain confidence in the service, but confidence is needed to commit substantial capital rather than some small amount for live test run.

Provider notifications and alerts are rather secondary importance for performance evaluation and obviously geared to retail customers. New max loss per trade is useful, but new max drawdown would be more useful. High slippage alert is a highly valuable tool. We have noted brutal slippage in a live account test October, 2009. My compliments to Zulutrade for addressing the issue of slippage with the new alert tool.

Reckless trading warning is something of marginal value as the trigger points are at least 1000 pip profit and 100% drawdown. None who´s in it for investment purposes would wait for 100% drawdown to pull the system from the trading mix. 20 or 30% are more likely trigger-points for removing the system from trading in live account.

The latest features give me some confidence that the far-from-perfect Zulutrade performance reporting is not intentional at least. If they want to make the business grow they should cater to needs of bigger investors rather than quick punters who go the markets for the thrill of it rather than consistent profit. The new features is a considerable step of moving from competition with the online casino sites toward bankable online investment proposition.

 

Written by A.S.

January 5, 2011 at 1:56 pm

Posted in Uncategorized

What happened to our real-time experiment?

with one comment

On June 1, 2010 we started a real-time experiment to illustrate how the system trading works and the profit potential of this type of investing.

We followed through for about 7 weeks before we diverted the resources to an income-producing work at hand. During those 7 weeks the results were mixed.

Overall, pure currency systems (Alpari PAMM, Tradency, Zulutrade) performed better than stock (Covestor) or commodity systems (Collective2). Since then, Covestor has redesigned its web site and the service offer looks more compelling than ever. The performance of Covestor models (i.e. strategies or trading systems) have recovered noticeably from last June. As at the time of writing, top-performing Covestor models are back on track with up to 55% returns during the last 3-month period (apparently, in line with the general market upward move).

Tradency has redesigned web site and now have the product offer directly from their web site rather than kept it obscured somewhere in brokers´ web sites. Collective2 and Zulutrade have kept their service offer basically unchanged, save for some non-essential features (such as additional languages and too-simple-to-be-useful analytic tools).

The basic premise of system trading is that, theoretically, a good trading system should perform well irrespectively of general market direction. In practice, most trading systems depend on certain market conditions, their performance is limited in time and it is critical to recognize and scale out systems whose ¨production stage¨ is over.

One of the conclusions to draw from the experiment is clear: constant monitoring and adjustment of the system portfolio is a must. Hardly any of the trading systems, originally included in the portfolio, pass the test 5-6 months later.

Namely,

On Collective2, out of 11 systems, 9 are down, 2 are up. The average decline is about 10%.

On Zulutrade, out of 15 systems, 4 systems are not identifiable, 6 are down and 5 are up. Average up is about 8% (not counting the unidentifiable four).

On Tradency, out of 6 systems, 2 are not identifiable, 3 are down and 1 is marginally up.

So, out of 32 systems, only 8 or 25% are good half-a-year later. The other 24 systems would have been removed from the system mix in the due course of the ongoing selection process.