RushRice USDJPY

2019/12 We have heard that the number of visits to our product page has suddenly increased. This was a text that I wrote only on the spur of the moment without organizing the flow of the story and without checking for typos and omissions. I apologize that it is very difficult to read and there are many parts that do not make sense. I will try to organize it so that at least the meaning makes sense. This EA has been suspended due to poor performance. 2019/05 Postscript. The following is a brief description of the process leading up to the suspension of sales of this EA. It is like a half excuse, so we hope you can read it off as you see fit. This EA was the first EA I ever exhibited. The core logic is to limit entry only in the direction of a long-term trend, and then to use Bollinger Bands to reverse the trend in the short term. This is a scalping EA using the 5-minute time frame. We knew that the logic was superior, but we ran into a number of obstacles before we were able to commercialize and sell the product. The price quoted by each broker was different, so we could not reproduce the entry/exit as in the backtest. The spread has a large impact on the results, and even a single difference in the spread can make a big difference. The results were affected by the slippage at the time of execution. The above are the typical problems, but there are many others. I made a prototype and tested it myself before selling it, but there was a period of time when I could not win. This EA aims to solve such problems of scalping EA from the technical side as follows. The price quoted by each broker was different, so the entry/exit could not be reproduced as in the back test. →EA has derived its own standard deviation based on price movement and heating (an indicator known as RCI), rather than price itself. This means that instead of using the MT4 standard Bollinger Bands, we created new Bollinger Bands that show our original values. Then, we set the entry condition for short-term inverse entry as "touching the original Bollinger Band 2σ". This was quite effective, and we believe that we succeeded to a significant degree in leveling out the differences in the prices quoted by different brokers. The spread has a large impact, and one difference can make a big difference in the results. The price slips at the time of execution, and the effect of the slipping is felt. →We responded by dividing the entry points into several and leveling the entry itself. Although the expected profit has decreased a little, it seems that the idea of wanting to enter at this point, and to take profit/lose at this point, has been reproduced. *Personal opinion: ...... We often hear that an EA with only one position in one trade = EA with no more pinching = EA with strong logic. However, EAs with only one position in short timeframe EAs are strongly affected by various problems such as those described here and often cannot withstand actual operation. If you want to enter 1 lot in total, it is better to divide it into 10 lots of 0.1 each, which will definitely be more stable in actual operation. When I see an EA that sells only 1 position, I look at it with suspicion, wondering if the creator has ever operated it in a real account ....... (I think all EA makers are probably looking at it with suspicion) In addition to this, in order to improve any accuracy, the following verifications and logic were organized. ▪ Using real tick data, we increased the accuracy of back testing as much as possible. →At that time, FXCM Japan, before it was absorbed by Rakuten Securities, provided real tick data on condition that the account balance was above a certain level. In order to minimize the effects of slippage and spread widening, we made adjustments by moving the limit price in detail so as not to place simple market orders. →As many of you may know, there is a famous free overseas EA called XMT-Scalper that is quite powerful. The countermeasure taken there is to adjust the limit price and stop-loss price in detail. My EA was equipped with a version of this method arranged for market trades. It is natural to say, but I think that overseas EA creators had the same problems that EA creators working at Gogojan had. If you are planning to create an EA, I suggest you start by reading the source code of XMT-Scalper, as it will be very helpful. I limited the timing of trades because scalping EAs often get hit hard during volatility, such as when an important indicator is announced. →This is not to say that the time of day to trade is restricted. (This is a slightly different approach from the common "morning scaling.) We cannot disclose the details here, but this is a response to sudden changes in the market that we have already had. In order to minimize curve fitting, we minimize the room for parameter adjustment. →For example, if there is an EA that enters and exits at the crossing of moving averages of different periods, the results will naturally differ between an EA with 10 short term and 50 long term and an EA with 20 short term and 100 long term. Just because the latter had better results when compared to the two backtests, there is no guarantee that the latter will continue to have better results in the future. To be clear, backtesting to find the optimal value while changing parameters one after another is almost meaningless. We only use values that are purely derived from the design of the logic and do not use values for which we do not know "why use that time period? I did not use values for which I did not know "why I used that period of time. I still think this is very important. These efforts, or rather, the process of eliminating problems one by one, bore fruit, and the profit-and-loss curve of my real account began to rise steadily. After about a month of forward testing, the results were almost as we had expected, and we decided to sell the product. It was an investment of everything I had at the time. I honestly thought I could revolutionize the automated trading industry with EAs. The result is as you can see. Sales were suspended due to poor performance. The same is true for the forwards in my real account, but at about the same time as the start of sales, the profit/loss curve began to drop steadily. The results were astounding. I put so much effort into this EA that I thought I had solved all the problems. Now I am relieved that I did not cheat my users by putting it on the market when it was steadily rising. Because I could not give up, I frantically investigated what was causing the problem. After a while, I was able to pinpoint the cause in my own way. It was not something that could not be solved, but I also knew that the next lifetime would come soon, probably just as soon as it was solved. I could only find logic that had a short lifespan. I could see the conclusion as close as possible that the lifespan of the short-term EA was extremely shorter than expected to begin with. I could see a weasel-word composition of problem generation and solution ahead. It was a weasel game in which the company's own funds were being sucked out of the market while the weasel game was going on, and the money was being sucked out of the company. When you look at the many EAs sold by GogoJan, I think you have to agree. Of the EAs that are winning in the short-term EAs, very few of them can win for a long time. How many years would you expect a logic that has withstood back-testing for 10 or 20 years to remain superior? This is true for most short-term EAs, but the reality is that the time is surprisingly short. To put it bluntly, backtesting a short-term EA for a long period of time with a steady upward trend is itself next to curve fitting. Curve fitting is like "having a certain trading logic and choosing the best results among the parameters used in the logic. For example, a logic that uses moving averages would choose the period of the moving average with the best results. It's a little hard to express in words, but my point is that the logic itself can be curve fitting to the environment. With the environment changing on a daily basis, is there really such a thing as short-term logic that works over a long period of time? If there is, it is probably a logic that has been curve-fitted to the long-term past. If it is the same short-term EA, it may not be possible to back-test it, but an EA that says, "This EA was optimized based on the price movements of the last 5 days and the power of other currencies" is much more reliable. From here on, things get a little more chatty. Again, there are many EAs on display at GogoJan that are competing on the merits of their logic. It may be a logic derived from extensive back-testing, or it may be a replication of a long experience of discretionary trading. However, after a certain period of time, very many of them fall into a steady decline. Or, although it is steadily rising in GogoJan-sama's forwards, the results are not so good when you actually purchase it and use it in a real account. One thing I can say is that EAs are not easy to win. This is especially true for the 15-minute time frame and below. After I found out that RushRice was not a winner, I joined overseas EA-related forums to learn about "EA making trends" and joined a certain EA development community in Japan for several years to exchange information. Thanks to these efforts, I have solidified my foundation in EA development and have been able to keep up with the changes of the times. Characteristically, in 2017-2018, automated trading, known as mmbot, became popular in Bitcoin FX. Perhaps because of this, recent EAs of the type that set up trades on short-term legs use sophisticated logic that the average user would probably not expect. The "logic that looks at the divergence of each broker and enters the trade with the broker that shows the divergence," which is described as an example in the explanation of our EA "AMA_EURUSD," is just the beginning. I am not sure if the short-term EAs that are popular among EA creators can even be called "EAs" anymore. They are equipped with analysis tools created in programming languages such as Python and R, and use statistically processed data to identify moments of advantage. MQL (which is a programming language for MT4) only does the core entry/exit. My personal impression is that when I talk to them, I don't get the impression that they are programmers talking to each other. They suddenly show you the mapped data and the conversation proceeds around the validity of the data used and the statistical process applied. It seems that the implementation of things like machine learning and artificial intelligence (AI) has not yet yielded results, but this is just not the case in the community I am a part of. I suspect that more technologically advanced teams and projects involving professionals with large capital have long since put them to practical use. Now, this is sometimes the reality of winning with short term EAs. To win with short-term EAs, you need to change and adjust the statistics on a daily basis and grab the advantage (statistical anomaly) that emerges only for a moment before anyone else does. There is one thing you should know here. That is, even with such a sophisticated EA, it is a prerequisite to "change and adjust statistical values on a daily basis" in order to operate it. Short-term EAs are not designed to be operated in such a way that when you find a particular Holy Grail logic, you just set it up on MT4 and wait for the rest to ....... The market is a kind of complex system with too many overlapping speculations, so the assumption that a specific logic will work is wrong in the first place. To be clear, when it comes to scalping, it is probably easier to get better at discretionary trading. The human brain is more flexible. This is how I came to the conclusion that long-term legs are better. To change the subject a bit, the advantage of automated trading programs in trading is the accuracy and speed of their decisions. And now this industry is becoming more and more a "speed" game. If the logic is at the same level, the faster program wins. In fact, in the stock and currency markets, a statistical advantage can be found sooner than you might think. If you can trade well at the moment the advantage appears, you will profit. Therefore, if you create a trading program like the one described above, you will be amazed at the steady performance when you back-test it. This does not mean, of course, that it is easy to win as long as you have programming skills. The moment you see an advantage and decide to enter, other programs have already completed their trades. Even if you have the Holy Grail logic, you will lose fast. Returning to the EA, even if there is a Holy Grail EA that can break through various difficulties and keep winning, its lifespan will come suddenly one day when it loses due to speed. You can never win a game of speed in an environment created by a large account with an environment that can only be created by a single individual. When bitcoin mining became popular. I think it is similar to the case when bitcoin mining became popular and there was no way to beat the huge mining factories in China with an individual's PC. If there is an EA that can win in the long term, it is still possible for an individual to win. It is for this reason that I released the AMA series as a stepping stone. This is just a brief description of what I know and what I have seen. Finally, please understand that this is only my personal viewpoint. *The name of the community is withheld, but those who know will understand. Please do not complain to the community if you make a mistake.