How We Created a Grail for News Trading and Lost Everything

This is the story of the search for, development, and testing of the perfect expert advisor for news trading. A spectacular rise, hundreds of percent in returns, and an encounter with harsh reality. We hope you will draw lessons from this story...

I think it is no secret to anyone why arbitrage systems attract so much attention. Making money without risk is a very appealing idea. That is why a beginner's work in Forex almost always starts with the search for the grail. On the one hand, there is nothing surprising about this, it is quite natural for a person to want to reach the goal by the shortest path possible. But by pure irony, in Forex this path turns out to be the longest.

The simplest definition of the grail is a money-printing machine. The trader may not participate in this process at all, most often acting as the trigger that sets the machine in motion. These days, no one is surprised by grails; almost everyone has a couple for themselves and a few more for friends. And it would seem that the task of finding the ideal trading system has exhausted itself: we choose a system we like from the first lines of Google results for the query " forex 1000% per month", test it, and put it on a live account to accumulate millions.

Unfortunately, the real market has not the slightest idea about the bacchanalia taking place in the minds of grail hunters and, without a shred of doubt, wipes out all the treasure seekers who covet its riches. At this stage enlightenment begins, and the earlier definition of the grail turns out to be fundamentally wrong and generally something semi-mythical.

Naturally, the idea of conquering such a creature as the market could not leave the minds of several enthusiast traders, and the collective mind began generating a wide variety of ideas, starting, of course, with excavating the skeleton of the Paleozoic era - the Ilanosaur (Ilan Expert Advisor) . The beast was X-rayed, the purpose of each bone was discovered, and all the shortcomings of its genetic structure were studied. As a result, the living creature took on the form of a machine, becoming faster, more powerful, and stronger than its predecessor, while still possessing all the same shortcomings.

Obviously, this development did not lead to high earnings, because although it had decent profitability, it imposed excessively large risks. Then it was decided to leave the idea behind and try another equally well-known trading method - trading the news. This time it was an unambiguous success. Orders were opened almost perfectly with the help of a simple expert advisor, and even manually. The feeling was as if a higher intelligence had shared a fraction of its intellect and you suddenly realized the meaning of the universe's existence.

So, the first stage was the formalization of the trading idea. The very idea of trading the news had long been in the air, but it seemed unrealizable. And, it must be said, not without reason. Judging reasonably, to trade the news on the real market, you need to descend to a fairly low level, far removed from the level of ordinary traders. But, oddly enough, it worked, and the positive slope of the balance line confirms it.

In general, the principle of news trading is very primitive. Some time before the news release (as little as possible), two opposite stop orders were placed at a sufficient distance from the price (depending on the instrument). When the news is released, the price breaks through one of the orders, after which in a very short time it can travel a distance of several hundred points. All that remains for us is to lock in the profit and rejoice that the deposit has literally grown by one McDonald's worker's salary in an instant.

A knowledgeable reader will most likely notice that not everything is so simple in the described scheme, and will be right. After all, there is slippage in the market and, in essence, it does not matter where you leave the orders - they can slip all the way to the very top, where they will eventually be executed. In most cases, with such an outcome, all profit is eaten up. So what is the secret?

Well, the thing was that the broker completely lacked a slippage mechanism. Naturally, it was an archaic market maker pretending to be an ECN broker, which of course it was not. However, this feature made it possible to trade the news perfectly, without spending a single point on slippage. The only costs left were a few points of fixed spread.

So, the idea was already working and bringing in money when trading manually. Of course, keeping up with the price flow during the news is not the easiest task, even under such royal conditions. Then the logical continuation becomes the creation of an automatic expert advisor trading according to a given algorithm, and its name is Pull.

No sooner said than done. To begin with, it was decided to modify an already finished expert advisor in order to test it in action right away. Therefore, at first the robot was fairly simple, with quite standard parameters for this type of expert advisors. Except that the functionality was somewhat expanded by adding a time filter and virtual trailing.

The algorithm itself is also quite simple. When the expert advisor starts, two opposite orders are placed at a fixed distance from the price. This is done some time before the news, usually no more than a few minutes. When one of the orders is activated, the second is deleted. To lock in profit, a virtual trailing stop and take profit are used. There are also timers to disable the trailing stop and delete untriggered orders if no spike occurred.

Even in this form, the work became much simpler. In fact, all that remained was to follow the news calendar and launch the expert advisor in time. But it was still far from that very printing press. Therefore, in the next version, the emphasis was placed on deep planning.

It was decided to write all news items into a separate text file, "news.txt". The robot treated the file as a scenario for action. Thus, trading could be planned in advance and the file supplemented with new data from time to time. The expert advisor now needed to be launched only once. Later, the idea with the file developed, and for each news item individually it became possible to specify its own extended set of parameters.

In the subsequent iteration, it was decided to automate everything that could possibly be automated. In essence, we set our sights on removing all routine processes from the work completely, leaving only the possibility of preliminary setup.

So, the first problem was downloading the news. Constant monitoring of the calendar is tiring, so naturally all of this had to be automated. Fortunately, there are many services with free access to the economic calendar. The choice was left with myfxbook, where there was an option to export in csv format, which was very easy for the expert advisor to parse (although now registration is required to obtain the data).

News is downloaded automatically, simultaneously syncing with the broker's time. The expert advisor downloaded the news calendar for the coming week, selected the news by importance (the parameters indicated which news to consider), and wrote it into the already familiar "news.txt". This file could then be edited however you liked, adding or changing parameters. Thus, the robot now had an exact program for how to trade each news event during the week, what stops to set, and so on. It looked something like this:

Several symbols could be specified for one date and time. This is relevant for dollar news, for example, where all the majors participate in the rally. The letter D means opening a position with the minimum lot - purely for statistics. The letter S is the standard lot coefficient. You can divide and multiply using the corresponding sign. And the last numerical value is an optional take profit.

All additional parameters, of course, were intended for manual editing. Therefore, it was then decided to collect extensive statistics on all news events in order to find average divergence values so that the expert advisor could independently decide which parameters to set. All spreads were written out into a separate file and broken down by individual news events. As the news spread (the volatility of the movement), the size of the first one-minute candle was taken, which in most cases is the time of the greatest activity.

Then, broken down by major pairs, the average spread values were calculated.

As the expert advisor operated, new values were written to the file, and the moving window average was calculated. Thus, the robot self-adapted to trading. The spread values were then used when setting take profit, stop loss, and trailing stop. News events generating insignificant price divergence were ignored.

For some time, the mechanism worked perfectly. The machine printed money, little by little replenishing the internal wallet. Along the way, small withdrawals were made, totaling about 100% of the original deposit. But at some point all of this stopped, and the account was blocked.

This decision was explained by two clauses of the client agreement. The first pointed to a limit on the maximum profit from a single trade, the second limited the minimum holding time of a position. As it turned out later, you did not even have to break the rules - trades could simply be canceled without explanation. That is, the broker reserves the right to adjust the financial result of any trader at any convenient moment. Excess profits are already a very weighty reason.

That is, in essence, to fall under the ban it is not at all necessary to trade the news. But, due to the specifics of order execution, it was precisely news trading that turned out to be that very arbitrage - a money-printing machine, if you like.

What is interesting is that the very fact of violating the local rules remains hidden from you for as long as possible. You will receive no notification or warning about possible problems - nothing of the sort. When you withdraw funds, the financial department will simply analyze the number of zeros in the amount being withdrawn and remove the extra ones, at best, or completely block the account. You can trade profitably for months without even suspecting that you will never see your money again.

Despite the failure, the team's enthusiasm remained, so it was decided to approach the problem from another angle. Thus, the staff were sharply retrained as testers. Accounts were opened simultaneously with several brokers, and execution, spreads, and slippage were checked. The robot was actively tested on popular ECN/STP venues and in some places even remained profitable, but these pennies could not compare in any way with the previous results. In most cases, however, trading was conducted at a loss.

At times, light was seen in the form of brokers advertising news trading. Here, in fact, there were two options. Either these were ordinary marketing tricks, or the firm would be shut down tomorrow. It just so happened that banks got involved in retail Forex not to hand out free money. Toxic flow will be completely blocked by the liquidity provider at first detection. Therefore, trading the news on the real market at the retail Forex level is in principle impossible - it is not for nothing that LMAX threw out all private traders. Perhaps in the future the situation will change, with the emergence of a full-fledged ECN, for example.

Companies operating under the ECN/STP model (see How the Forex Market Works) will be glad to see news traders, scalpers, and any other traders generating large volumes. The more commission a trader generates, the better. In this case, the broker risks nothing, because it will never execute a trade without first receiving confirmation from the liquidity provider.

It is practically impossible to trade the news this way, for at least two reasons. First, the delay in sending the order is aggravated by the difficulty of execution during extreme market situations. If liquidity is zero, the canceled order will be sent to the next provider, then the next, and so on, until it is finally executed. Ideally, the order will be executed at the best current market price if the provider pool is large enough.

Second, in most cases, you will not even have time to send the order for execution. Some time before the news, the market depth is almost completely emptied, which in the terminal is reflected as a substantial spread widening (tens of points). Thus, there is no possibility of placing an order at the beginning of the impulse - the profitability of such a trade is zero.

In turn, brokers operating under the market maker model (internal clearing) earn from client losses, that is, they trade against traders. When an order is sent, the broker confirms it immediately, without coordinating with a third party. Accordingly, there is no possibility of hedging a news position, and all profit received is a direct loss for the company.

In fact, everything was fine with the robot - it completed its task 100%. The problem was that we relied on the vulnerability of a single broker, that is, we were dependent on it. Even if this is arbitrage, a single vulnerable point can destroy the whole system, which is what ultimately happened to Pull.

Another moral is that you should not joke with bucket shops. Such firms really dislike it when money is taken from them. Even if you managed to withdraw some capital, sooner or later it will be noticed, and a competently drafted contract may not play in your favor. The size of the company says nothing. The public offer agreement and reviews from real clients are the only criteria that come closest to reality.

There it is - the pinnacle of human genius. Like the captain of an autonomous ship, you watch as the names of destinations appear and disappear on the chart, while in the logbook the robot periodically reports on the work done. It would seem that nothing could knock the autopilot off course - the chart was rising, but as if gravity intervened and at one moment the chart lost its relevance.

Best regards, Alexey Vergunov
TradeLikeaPro.ru