ゴートゥーエブリデイ

Background of Production

I always ask myself, "Is there any way to use the powerful weapon of FX leverage as an investment rather than a speculation?" I am always thinking "Is there any way to use the powerful weapon of FX leverage as an investment rather than a speculation?

After the first "Oceania Brothers," we developed 37 EAs in about 2 years to see if we could create a new logic, but none of them worked well when back-tested over a long period of time, or failed to generate profit when forward-tested even though back-tested well, so we put them in storage. I had to put them in storage.

And the 38th one is this second system trade, "Go to Everyday".

I thought this logic would be a level of system trading that I could sell in my mind, so I decided to release it to the public.

Point 1)
Trade at the middle market based on actual demand.

Just as the first "Oceania Brothers" falls under the genre of "repeat trades," "Go to Everyday" falls under the genre of "middle price trades.

The NAKANE is the daily exchange rate published by banks and other financial institutions at 10:00 a.m. every day. Based on this rate, companies doing business with foreign countries exchange yen into dollars for dollar-denominated settlements.

This middle price is determined by the rate at 9:55 AM.

Business transactions tend to be concentrated on the last day of the month. On the days falling on the 5th and 10th, companies request payment to their banks in advance, and these payments are often settled at the time when the middle market price is determined.

If there is a lot of dollar buying at that time, USD/JPY is expected to rise toward 9:55 AM. This is called an "intermediate price trade" to follow this trend.

While there are many indicators that are difficult to explain, the middle price trade, backed by actual demand for economic activity, is said to have certain advantages.

Point #2.
Align technical analysis.

If USD/JPY really does go up every day towards 9:55, why leave it to the EA in the first place?"

Why are there so many different methods of middle market trading? Is there a difference?"

In fact, the "middle price" phenomenon is relatively vague and weak for a well-known fundamental.

In a nutshell, a middle price trade is not simple and raises a number of questions.

We developed "Go to Everyday" thinking that we could win if we could find a significant law, back-test it, adjust parameters, confirm reproducibility by forward testing, and turn it into an EA.

In "Go to Everyday," technical analysis is also used to determine when it is appropriate to enter the market and when it is the right time to buy.

This resulted in a consistent 66% win rate overall and 74% on the fiftieth and Friday.

Although not a constant winner like the Oceania Brothers, repeated trades at this winning rate have resulted in a CAGR (compound annual growth rate) of 34.27% and a total return of 1,319% over 9 years, while keeping the maximum drawdown in the 6% range. (Back-tested with a domestic firm with 25x leverage)

Point (iii).
Tune in "every day of the week, every fifty days, and every month."

The phenomenon of the middle price has different characteristics for each day of the week, each fifty-day period, and each month.

In "Go to Everyday," the number of lots, entry conditions, and profit/loss conditions can be adjusted according to their nature.

First, let's look at what trends we see.

Trends by day of the week

First, let's look at the trends for each day of the week.

The columns of the table are, from left to right: CAGR (compound annual growth rate/annual interest rate), net profit (in yen), win rate, profit factor, maximum drawdown, total number of trades, and trend (graph).

If the trend is rightward, we intuitively know that the middle price trade is advantageous.

It is very clear in the trend graph. We found that there is a marked difference by day of the week.

We found that the logic favors Friday, the last business day of the week, as well as Wednesday, the middle of the week, for firms that have Saturdays and Sundays off, and conversely, it does not work well on Thursdays.

In "Go to Everyday," the number of lots can be adjusted to match this day-of-week trend.

Trends every 50 days

Next is the trend for every fifty days, which is generally considered favorable.

As expected for a fifty-day period, there are many straight lines that are easily righted compared to the days of the week. The average win rate is also 74%.

The day of the week comes around 4-5 times a month, while the date comes around only once a month, so the total number of transactions and net profit is lower, but it is still advantageous.

We found that the 5th, 15th, and 20th days were particularly significant. The number of lots should be adjusted for this 50-day trend as well.

By the way, the last day of the month refers to the 30th or 31st. In February, the last day of the month is the 28th, and in leap years, the last day of the month is the 29th.

monthly

As a further major indicator, differences exist from month to month.

After weighting the predominance of days of the week and fifty days of the week, let's look at the differences by month.

The first is from January to June.

Apparently, May is not a good month and the other months are favorable.

This is followed by July through December.

We found that an easy mid-price trade in August was largely negative, while the end of the year was advantageous.

In "Go to Everyday," significant monthly differences are converted into entry, gain, and loss conditions and handled with the aim of extending profits while reducing risk.

It also allows for compound interest to compensate for the low number of entries as it examines them.

Point (iv).
"You can set a day when you will not trade."

No matter how much we optimize conditions by day of the week, by fifty days, or by month, we cannot beat the fundamentals of a wave larger than the middle price.

For example, the day after the FOMC releases its monetary policy, the USD/JPY will fluctuate wildly, and the significance of the middle price will be drowned out.

Therefore, "Go to Everyday" allows you to set a day when you do not trade by writing it in advance in the settings file.

In addition to Japanese holidays, where no real demand is expected, the FOMC meeting dates are written in advance.

If other economic indicators or fundamentals of concern are expected, the trade can be avoided by adding the date to the file in advance.

Point (5).
Uncorrelated to Oceania Brothers.

The last point is that it is uncorrelated with the Oceania Brothers.

The key to building a portfolio using multiple EAs is the "diversification effect" and the "correlation effect.

dispersion effect

The first, "diversification effect," is the well-known theory that risk can be reduced by holding multiple assets, familiarly known as "don't put all your eggs in one basket.

However, if you diversify into assets that move in the same way, the risk reduction effect will be limited no matter how diversified you are.

correlation effect

The second "correlation effect" means that if you build a portfolio with logics that move differently, you can reduce risk (in the case of EAs, total drawdowns will be reduced).

Think of an individual investor using an EA as the president of a company. The CEO does not want to work, so he/she makes money and logic work for him. In other words, the EA is an employee.

You have hired Mr. Oceania Brothers as your first employee and Ms. GoToEveryday as your second employee.

The two are each given a budget and asked to earn money by performing their duties.

At this point, what if these two employees have similar skills and personalities and are assigned similar tasks?

That's right. They make a mistake at the same time, they lose money at the same time, and the company is out of business.

If we are going to have them do similar work in the first place, why not just have one employee and increase the budget (number of lots in EA)?

If you are going to hire two employees, it is more stable for the company to have them perform completely different tasks with completely different skills and personalities.

Of course, in the short term, there may be periods when both of you will lose money, but in the long run, the probability that both of you will make mistakes at the same time is low, reducing the risk.

The larger the number of these employees, the better the diversification effect, but it is a prerequisite that each employee, or EA, is trustworthy. The employees should be as clear as possible about their origins (or logic in the case of EA) and make as few mistakes as possible.

It took me more than two years to find a reliable method instead of unnecessarily increasing the number to build a portfolio!

There is no correlation between "Oceania Brothers," which relies on the AUD/NZD retroactivity for long-term trades, and "GoToEveryday," which relies on the economic activity of the MAC to make short-term trades. The diversification and correlation effect is high, so please try to put a portfolio together.

Back Tests / Domestic Contractors

The following are the results of a backtest using the domestic FX firm OANDA Japan with the default values of the parameters.

The following are only the results of back-testing against historical data, and the default values of the parameters do not guarantee safety against forced losses.

parameter

The parameters that can be specified for "Go to Everyday" are as follows

basis

The following are basic parameters.

Even when operating at the default values, adjust the compounding factor for compounding and the simple lot number for simple interest operation to match the initial margin and leverage.

time

Set the time to be used for entry and close. Check the server time of the vendor you are using and set the time.

Lot Factor / Day of Week

Specify the lot factor for each day of the week. The lot factors for the day of the week and fifty days are summed and multiplied by the number of lots.

If significance is expected for each day of the week, set a larger value.

Lot Factor, 50 days

Specify the lot factor for each fifty-day period. The lot factors for the day of the week and the fifty days are summed and multiplied by the number of lots.

If significance is expected by 50 days, set a larger value.

decision MA

Specify the MA to be used to determine the intersection with Bollinger Bands.

Class Number / Month

Specify the class number to which each month is assigned.

Class Details

Specify the deviation of the Bollinger band and the profit/loss pips to be used for entry and close for each class.

The smaller the number for the entry condition, the greater the risk/reward for the close condition.

A trailing stop is available that automatically updates the stop loss level to reduce the amount of loss in the event of a loss.

For Class 1, see below.

Classes 2 through 5 are specified in the same manner.

Supported Currency Pairs

Only USD/JPY is supported.

Corresponding vendor

We intend to design the system to work with any vendor that handles USD/JPY.

The author has confirmed the operation with the following vendors

Domestic Contractors