A self-proclaimed expert forex trader whose seminar I once attended demonstrated his success with an impressive trading history. As a professional engineer with responsibility for a large electronics company's quality control and assurance, I should have been wary. I regularly applied statistical techniques where variables and their ranges are identified, simulations were run and predictions generated. Having this knowledge and experience, I should have questioned his technical analysis approach to forex trading. That would have saved me a lot of money, stress and family problems.
Applying his theory and other supposedly effective market predictors and losing money, I stepped back from the market and began devising an approach to forex trading that ignored technical or fundamental analysis. My idea was that, using historic trading data and applying certain statistical techniques, one could estimate the probability that a trade would be profitable. To summarize the idea – statistical probability analysis.
Ten years of development and testing ensued, followed by almost four years of live trading. In that time, I have learned a valuable lesson. There are no trends in forex – at least trends a successful trader can rely upon. Certainly, there are long-term economic trends that affect the future of a country's currency. But, in the short term, volatility in the market makes "buy and hold" an unlikely way to consistently profit.
To demonstrate that trend trading is a sure loser, one needs only observe the movements of the EUR/USD pair in four-hour increments for the last thirteen years. It becomes clear that no trends becomes apparent.
In studying market movements, my objective was to see how long an upward or downward trend lasted. In over 10,000 four-hour periods, eight-hour up or down turns occurred just 12 percent of the time, while four-hour upturns and downturns happened nearly 25 percent of the time. Twelve-hour up or down movements both occurred approximately six percent of the time. When an investor bets on twenty-hour consecutive up trends, the odds are about 150 to 1. Combining leverage, volatility and the absence of trends, the profit picture becomes gloomy.
Things have been worse than gloomy for the largest 103 currency funds. Barclay CTA Currency Trader’s Index recently reported their annual returns averaged just two percent per year for the last three years. Clearly, using only fundamental and technical analysis approaches to predict the market aren't working.
Fundamental and technical approaches to trading generally work better in the equities market where there is a greater diversity of instruments and market sectors, liquidity is less, volatility is not as great and leverage is substantially smaller. The sheer size of the forex market, the geographic span of its investors, its liquidity, and sensitivity to political events make forex unpredictable.
Further complicating the efforts of fundamental and technical based trading computer systems is a common practice of using a single market analysis and projection algorithm. The single model approach overlooks the fact that there are three distinctly different types of markets – up, down, and seesaw. Each situation has its own characteristics and different transitional behavior.
An approach to forex trading that is showing significant promise employs multiple statistical models to evaluate the probability that a trade will be profitable and automation to trade based upon the probability. Combining the statistical probability approach with a conservative risk management strategy and computer automated trading, returns, significantly above average, are being consistently generated, even in periods of high market volatility.
Thirteen years ago, after I lost a considerable amount of money using technical and fundamental trading methodologies that I had learned from “experts,” I understood that, to paraphrase the title of a classic trading book, the forex market was a random walk in the currency world. There had to be a better way to structure trading activity. Studying market movements, a group of mathematicians, statisticians, engineers and I devised analytical routines to evaluate these movements and project to what degree a profitable trade could be made.
Development and testing of the first algorithm made clear the fact that different evaluation models would be needed for various market conditions. New, independent models were developed. To evaluate their effectiveness, backtesting was followed by live, moneyless trading. If trading were to be automated, rules reflecting the user's risk tolerance should govern all transactions.
Using rule-enforced automated trading, early exits, late exits, and compounding, fear, greed and other emotional decisions would not have any impact. Volatility averse trading was chosen when rules were set in place requiring a maximum loss of approximately forty percent maximum gain. Lot sizes would not vary, and transactions would be conducted on a trade-by-trade basis, not FIFO.
Warren Buffett once set down his rules for trading:
Rule 1: Don't lose money.
Rule 2: Don't forget Rule 1.
Impossible? Yes. But, our takeaway from this made us consider risk management as the first priority.
Six independent models evaluating statistically probability of a profitable trade, relying upon the computer to immediately initiate trades, and imposing strict limits on loss proved successful. We were now in the position to rewrite the Buffett rules to meet our situation:
Rule 1: Win 45 percent of the time and break even. Win 50 percent of the time and provide double digit returns over time.
Rule 2: Back off and allow the computer do its job.
Since the software was introduced to the market in 2010, returns have been consistent. As an indication that the six independent up/down/seesaw market analysis algorithms are working, no software changes have been made, and the automated system has provided average double digit returns with very limited risk.
A question often arises if statistical probability trading is similar to high-speed trading. No. Average trade hold time is 24 hours, maximum hold time is two weeks.
In the past three years when geopolitical and economic events have roiled the currency markets, technical and fundamental trading tools have failed to produce returns. Statistical probability has proven itself by generating significant returns, thus ensuring that large, diversified funds can still allocate assets to forex and remain confident that the returns will contribute positively to the overall success of the fund.