FX system trade theory. "Oishi style FX system trade theory"

FX system trade theory

System trade is the only way to make a profit with FX?


FX is not investment. It's a speculation or gamble.

However, unlike other kinds of gamble, FX is very simple. Up or down, is all there is to it. You have to spend some money on costs like FX rate spreads, but it's a very small cost. FX is a good gamble that makes it difficult to lose money.

Is there a winning strategy for FX? Unfortunately, it doesn't exist. There is only probability theory in which you analyze information many times and earn and lose many times.

Fundamental analysis, technical analysis, etc. there are many analyzing methods for FX. Many pro-traders make full use of these analyses and make a profit.

Does it mean that everyone can make a profit if we use the same things like pro-traders? No.

FX is currency trading, so someone loses money when someone makes a profit. Nobody deals with the trade to lose money. Everyone uses similar indicators, but both winners and losers exist. Besides, there are a handful of winners who earn a lot and so many losers who lose their money.

The result of this has made me believe the prospect theory of behavioral economics. This theory says how people evaluate money is bases on their situation. According to the theory, evaluating money is based on a pattern.

Prospect theory

In the investment world, people tend to care about loss more than profit.

If you care about loss too much, you are repulsed by losing money even if it's a small amount and loss cut can be difficult. You tend to think "Perhaps the rate return to the same as when I bought." or "Let's cut a loss when the rate is better."
And you can't cut a loss easily and eventually your loss grows.

The coefficient of this mental strain is considered as 0.35. For example, supposing that you got a 60% success rate method, but your loss cut is always bigger than profit because of psychological effects, your loss will be like below.

Profit: 0.6 X 0.35 = 0.21
Loss: 0.4 X 0.65 = 0.26

That's why you can't make a profit easily even if you get a method.

I think the people who keep making a profit have 2 elements.

[Winner's elements]
1. They have a special method based on their instinct or effort.
2. They have extraordinary mental strength and they can make a profit and cut a loss theoretically.

As I said, FX is currency trading and rate spreads are cheap; it's not so tough to find the special method above. It's possible that you get 60% success rate, so you might think everyone can increase wealth.

The problem is having mental strength. People with mental strength do a great job in any world, such as the sports world and the business world. It's a rare, special ability so only a handful people have it.

It means it's tough to keep making a profit just with a method. If you say there is a method that everyone can make a profit, I think it is system trade. A System trade doesn't have emotion and it makes a profit and cuts a loss mechanically. If you have a method, you don't need mental strength at system trade.

My conclusion based on my own experience is "FX system trade is the only way to make a profit with FX for ordinary people" This web site shows my research based on this conclusion.

Oishi style 5S of FX system trade


When I make a new EA, I consider some subjects in order every time. It has 5 steps and I use high level ones in them. I named these 5 steps "5S of FX system trade"

1. Supposition

Construct a hypothesis that you can make a profit, and then combine essential indicators. The logic of the device is completed here.

2. Safety

As for financial products, safety is the most important. Evening up logics, such as loss cut and clearance condition are completed here. You should implement error handling function and so on, too.

3. Stability

Coordinate the parameter during backtesting. Aim at stable equity curve that has robust stability and small drawdown.

4. Study

Backtesting is just a test. You should manage it for the final adjustment. Management then improvement, management then improvement... just repeat. If it's not good even if you repeat it again and again, you should cut it off.

5. Statistics

Even if the EA is great, FX doesn't have absolute method. Show risks, expected revenue and correlation of each EA. Add fundamental analysis and make a portfolio.

Here is the detail. It should be long passage, so please read it at your spare time.

1. Supposition


I often see the following comments by EA developer on FX magazines.

"1 or 0 out of 100 EA makes a profit." "I make one EX a week at least."

I don't disagree with them because I know they make an effort and I respect them. However, there are not a lot of indicators and every indicator is similar. It's impossible to make a new logic one after another.

In my case, I construct a hypothesis first. It's a story about how the market will change and what investors will do, and what I should do to make a profit.

The hypothesis is not science, it's a theory. It's not an absolute theory; it's one of the possibilities. However, it supports you to make a profit stronger than intuition.

My rate of developing a good EA is fifty-fifty. I prefer improving some EA to developing new EAs.

Here is example of my hypothesis.

Comparing FX to 3C of business strategy,

Market (The total of loss of traders who lost money ? The stockbroker's fee)
? Concurrence (The total of profit of traders who make a profit) = Own profit

Let's see each detail.

-Market (The total of loss of traders who lost money ? The stockbroker's fee)

Market, that is say many traders who lose money should exist. Such traders can be classified to 5 categories.

1. Nation

It buys and sells currency not to make a profit; it's a nation policy. For example, Switzerland sticks on minimum exchange rate, and Japanese invention. They don't intend to make a profit, so they can be a big customer.

2. Companies

Therefore, they can make a big trend in closing month or the end of month. However, they often make an exchange contract, so it's difficult to read their movement.

3. Long-term investors

They are investors who are individuals or funds.

4. Pro-traders

They are short-term investors who are individuals or funds. They move exchange rates widely and grasp other trader's loss cut and make a profit. They move especially speculative markets that don't circulate frequently like a pound. Such markets are often moved by them even in a moth or time that many people don't trade in.

5. Ordinary traders

It's us. If we trade without any special method based on the prospect theory of behavioral economics, we lose money.

Who among good customers are good customers, who lose money a lot? In my hypothesis, I target nations who don't intend to make a profit and ordinary traders who probably lose money.

-Concurrence (The total of profit of traders who make a profit) vs. own profit

In the FX world, the segments of concurrence equal markets. Especially pro-traders are the biggest concurrence. They are sometimes called profit-hunters. They are natural enemies of technical analysis.

They have an enormous financial strength and knowledge and amount of information. How can we compete with them? According to Michel Porter, there are 3 strategies.

1. Cost leadership
2. Differentiation
3. Concentration

1 (cost leadership) is a strategy that top share companies in markets use. In the FX world, pro-traders who were defined as concurrence are categorized here.

3 (concentration) is a strategy that concentrate on a specific currency or timing to make a strong environment partly, but we don't have enough capital strength to move a rate.

2 (differentiation) is a strategy to avoid competing with them. It's basic strategy for small and medium-sized companies and only one strategy we can use.

There are 2 ways to avoid to competing with them.

1. Trading during the time that pro-traders don't usually trade.
2. Cutting loss largely and avoid their loss cut hunting.

-Reading the mass psychology of nations and individual traders to make your position. -Trading during the time that pro-traders don't usually trade, or cutting enough loss to withstand their trick. Let's use these 2 ways and construct 2 hypotheses.

Hypothesis1. If you follow a nation policy and make a big loss cut to withstand pro-traders' short-term trade, you can make a profit in medium and long term.

Hypothesis2. If you read individual traders' mind while trading during the time pro-traders don't usually trade, you can make a profit.

Science is reproducible and anyone can make the same result. Theory accepts individuals' feelings or other interpretation, so it's just one of possibilities. Hypothesis is just logic, everyone has each different one and there is no absolute one, so please read my hypothesis as a reference.

2. Safety


Safety is the most important for not only FX but also other financial products.

In recent years, we have seen incredible cases like big Japanese and American companies going bankrupt; the nuclear power plant had a big accident and geniuses in Wall Street made a mistake... We recognized that there is no absolute thing in this world.

I'm an engineer and I can't make a product that never has trouble. Everything breaks someday so it's just a probability theory.

Safety means the design that avoids fatal damages based on all possibility of error. FX is the same as tjos. Absolute method? Winning strategy? Everything is a lie. Any movement can happen. Perhaps a nation suddenly gives up on their policy.

1. Set loss cut and leverage regulation that you are never saddled with more debt than your capital.

2. The incidence rate of loss cut X The amount of loss of loss cut < The percentage of making a profit X The amount of profit

3. Prevent loss cut error, leverage setting error by system.

Suppose you to make a risk and try safe trade. My EA has time condition setting, trading stop function and other evening up condition settings, but at any rate I have set 100% loss cut.

Among the latest stockbrokers, "countdown system" that you can't set loss cut when you buy and sell currency is the mainstream, and of course, my EA can deal with it. It orders loss cut setting as soon as a product is ordered. Sometimes an error interrupts the settings, so I implemented a function that checks if loss cut is set properly and checked again and again until it is set properly.

My EA has a standard function that deals with various errors.

As for leverage regulation, I have set the upper limit of loss cut as 25 times, Japanese regulation. It is the setting that my loss cut never exceed my deposit.

After these preventive measures, I adopted logic that can make a high percentage of profit. FX is a money game and you should enjoy it casually. You shouldn't have an FX debt. Please enjoy FX with safety.

3. Stability


There are 2 things to explain here.

(1) Make an equity curve as steady as possible

Compare these 2 equity curves.

In the first one, it increases almost steadily. You can make a profit whenever you start to trade.

Second one is jagged. If it's the worst timing when you start to trade and don't.

"Drawdown" that is showed at backtesting tells you the degree of risk. The number it shows is the worst point of the equity curve. If you see the percentage, it might influence your asset setting, so see "pips" that shows the absolute value. You should invest more money than this number. Try to make a steady equity curve with small drawdown.

It was kind of common sense. I want to explain the key points of the next topic.

(2) Regulate the parameter on the top of a gentle curve.

Here is a scene of optimizing a parameter.

If you see not the graph but the number and optimize it, you might choose the sharp point on the left. However, you shouldn't do it.

What is a parameter? For example, it's a time at the indicator term setting.

Do you think a price fluctuation happens regularly? Of course not.

If it happened regularly, anyone could make a profit. It happens based on many kinds of factors. What will happen if you regulate the parameter at an important point? A big difference between backtesting and real management can be showed because the cycle must change. One parameter was set on the sharp point and other parameters were optimized by it, so whole balance breaks and you must lose.

This is the mechanism of carve fitting.

Here is the model answer. You should choose the top of a gentle arc. It doesn't influence very much even if the cycle changes. In the first graph, the third parameter from the right is the best.

As an engineer, I can call it a robust design of quality engineering. Optimize doesn't mean making the best quality; making a discrepancy as small as possible.

It shows us that practical use is rarely equal to or better than backtesting. A problem is how much you will be suppressed by the degradation. Don't be misled by the optimized backtesting, you should anticipate the result at practical use.

4. Study


You made an EA anyway. It's an EA that you can make a profit with basktesting at least.

In my case, I use the data for 2000 to 2010 for backtesting. If you use only data of a few years, it's not enough and you should fall into the curve fitting in.

I use the data of 2011 to 2013 for forwardtesting. Forwardtesting is similar to backtesting, but it makes a result on the assumption that it's in the future. If the data that I didn't use when I develop the EA is good, I can expect a good result. If it suddenly makes a bad result, I may fall into curve fitting. I can put a filter on the program that judge like that before practical use.

The EA that passed such a process is still just on backtesting. It finally got a starting point. You don't know if you can make a profit until you use it at real trade.

Practical use & Improvement & Practical use & ...

You should repeat it and make your EA better. It's a study phase and never-ending days of study and improvement.

This phase should be used as a problem solving technique. Foreign affiliated consultants and Toyota are famous for problem solving, and basically the same as them here.

Set up a purpose & analyze and find problems & clear up the cause of them & work out a countermeasure

This is the brief step. The theory is a bit difficult to explain, so I'll show you an example.

This is the example of the process of buy on dip logic that I used to distribute for free.

-Set up a purpose

My EA user reported results of practical use to me and it was totally different from my own result. Basically it was a great one, but it sometimes suffers more losses than making a profit in recent years. I tried backtesting with my accounts, and I got the same result as the users.

The general direction was same and the number of losses was one more than making a profit. There are many factors like rate, time, starting timing of practical use and ability of bargain. The problem is the EA that can't allow such errors.

In addition, nowadays, there are only a few trades. Since loss cut is big, it takes months to recover the loss cut if there are not a lot of trades. Because of big loss cut? Or the number of trade decreases because the logic is imperfect? In this case, I set up the following purpose.

Purpose: Improve the buy on dip EA to standardize income in 2 months.

-Analyze and find problems

Use the backtesting data and analyze it.

I split it into Buy and Sell first. You can see that Buy is steady equity curve, but Sell is obviously bad that shows it hasn't made a profit from 2011. The mutual part can be the problem, but let's target Sell and resolve it.

Let's see position logic and clearance logic. As for the clearance logic, I optimized the profit and loss-cutting. Both of them describe a gentle arc and they don't have any problem. Including the balance of risk, the current level is ok. You can make the loss-cutting a little bit less at most.

Next, I optimized each parameter on position logic. There are 3 trend parameters and 2 buy and sell decision functions. The buy and sell decision functions have a parameter that correlates with decline of the number of buy and sell and rise of profit. Standardizing monthly profit is the purpose this time, so ignore it here.

You can think there is a problem on the trend decision. If you optimize it, the number of buy and sell dramatically changes. Trend decision logic makes the number of buy and sell decrease.

Problem: Trend decision logic makes the number of buy and sell decrease.

- clear up the cause of the problem

Let's clear up the cause of the problem while optimizing these 3 parameters.

These parameters correlate with each other, so you should change them at the same time. Check them and optimize at the same time. After that, they are combined automatically and you can see a graph like this.

Don't choose the best one. Don't forget about stability. You should see dispersion here. Right click on the graph and change it into 2D Surface.

This is a mode that set different parameters on the vertical axis and the abscissas axis, and you can see their correlation. You are seeing the correlation of 2 parameters out of 3. The depth of color indicates profit.

There is a white cell above the darkest cell; it is not a gradation off color. While trying some things I noticed that the number of buy and sell sharply decrease because of a small trigger if the 2 parameters are too close, and you can't make a profit. This time, I target optimizing too much and it came to the singular point.

Cause: Optimize the parameters too much and closed to the singular point.

-Work out a countermeasure

As you see the result of optimization, probably you can increase the number of buy and sell with profit factors. However, you haven't done, of course. Check practical use test and brush up your EA. You should repeat improvement again and again.

I showed you an example this time, but I had done more things to improve it like changing the pair of currencies and narrowing the width of loss-cutting. It's important to take a lot of time to improve your EA.

5. Statistics


Finally, it's the last step. Even if you make an excellent EA, there is no "absolute" in the FX world. There is no solid investment, so diversification investment is the basic way for risk aversion. I think you know that people call a diversification investment a "portfolio". In concrete terms, examining expected revenue, risk and correlation and funding to multiple places in the apt ratio.

(1) Expected revenue

Generally, the percentage of expected revenue is represented like this.

(Incidence of the economic situation X The expected revenue in the situation) +...

Apart from binary option, it's normally tough to analyze the economic situation at FX. When you use an EA, you assume that it includes various economic situation based on enough backtesting.

It's not rare that a trend last long. So you need to use 10 years data for backtesting.

Let's say he monthly expected revenue that backtesting shows you is expected revenue.

(2) Risk

An EA that always makes a profit can sometimes lose. Even if you make an average profit on during 3 years, you might not feel like keeping with the same EA if you lose for a few months in row.

EAs display "Maximal drawdown" that means the biggest loss. You can refer to it. In short, risk means discrepancy for the average. Of course, the smaller number it is, the better your situation is.

On the other hand, there is a concept called risk premium. It means that investment with a big risk should be a big profit. As long as a risk in line with profit, you use it as a part of portfolio and can make a continuously growing equity curve.

When you make a portfolio, following order is the basics.

Small risk / Big profit > Small risk / Small profit > Big risk / Big profit

Needless to say, Big risk / Small profit is out of the question. And it's usually too got to be true to have a Small risk / Big profit.

Robert Toru Kiyosaki said insider trading is the only way to make a profit overwhelmingly by investment.

Combining Small risk / Small profit mainly and mixing Big risk / Big profit a little. It's a practical strategy.

(3) Correlation

In other investment cases, you calculate correlation function of the investment destination that you would add to a portfolio. It's an indicator that shows how much correlation the investment destination is.

Even if you diversify investment, a portfolio doesn't have anything if the stock price crash happens because of Yen appreciation due to Japanese exporters. You should choose the combinations that don't have correlation for a portfolio.

In FX cases, you can think of it more easily. Even if you choose currency pairs thoroughly, usually dollars and euro are mainstreams. The options can be very simple; buy or sell and trend or range.

After all, you should combine the same amount of trend and range (oscillator) I considered it so I distribute 4 or 5 sets of EA for free at this web site.



That's Oishi style 5S of FX system trade.

Thank you for reading this long article. If you are in agreement and understanding with me, use EA and try system trade.


Japanese / English


Link