How technical problems can ruin the life of a trader, or why it is important to choose a reliable broker

What problems await those who decided to use automated trading strategies (robots) in trading? Advertising of trading robots speaks of their reliability and stability. Unfortunately, sometimes technical bugs and stupid work of a financial company can lead to significant cash losses.

I will tell about my experience with one Novosibirsk company, which has a branch in Moscow. As a result of a combination of software bugs, poor support services, and simply failures, my losses amounted to no less than 15% of the capital. In order to warn against mistakes of those who are interested in the topic of trading, I decided to talk about it. I tried to write as simple as possible, but I still need to be a little in the subject of trading.

To whom it is interesting to learn about my grief-experience - please, under cat.

General thoughts on using robots

Trading makes it possible to make good money on accumulated capital, but this is always a risk. Moreover, as a rule, the greater the risk, the more you can earn, but also lose more in case of failure. It is clear that not everyone has time to seriously engage in trading and develop their own trading strategies. Many want to find something ready. And since demand always creates supply, broker companies offer a wide variety of portfolios. Many of them have been tested and have been working for many years, you can easily see the dynamics over the past 3 years: how much the initial capital has increased / decreased, what were the maximum drawdowns (losses in the worst case). Here, for example, is a selection of trading robots , some of which are quite interesting.

By comparing the potential benefits and risks, you can choose something suitable. Then it remains only to connect the tariff with auto-execution of transactions and hope that capital will continue to grow further, as it has done on the chart for the last few years. And if the trend really continues, then we will make good money. And nothing needs to be done. So or not?

It turns out, not really. For example, you still need to monitor the trading system. If the virtual server (usually brokers themselves offer this service) that runs auto-execution of transactions falls, then something may not work, and you deviate from the strategy of the robot. Financial companies (at least the one that went to me) themselves do not monitor your server. If something falls, these are your problems. Which, in general, is wrong, in my opinion.

What can a server crash cause? For example, you are on vacation, and your trading robot bought shares of Sberbank (suppose you have an aggressive portfolio consisting of only one of these shares). The transaction was unsuccessful, the price went down, ideally, the robot should have closed it with a slight loss. But the server crashed and the position remained open. And then, as luck would have it, “Black Monday” (or Tuesday, or something else) happened, and Sberbank collapsed by 20%. And the server still lies. Such a technical bug can lead to more losses than choosing the most unsuccessful strategy.

This is unlikely, you say. Well, it depends on how often it will fall. My server, for example, crashed 5 times in one month due to the fact that on 1Gb affiliate hostingsomething has broken. True, SMS notifications came, although often with a delay. However, the fall of hosting, although they themselves are indignant, have never led me to any serious losses. Sometimes they can even work as a plus if they are lucky.

Well, suppose we have a very reliable server that never crashes. The selected portfolio shows good momentum. Is it possible to sleep peacefully? And again, no. It turns out that there are much worse things than server failures ...

A bit of background

For more than a year I used the services of the company. It was different. There were attempts to trade yourself, then it came to understand that it requires too much strength, energy, nerves. There were various ready-made portfolios, there were earnings and drawdowns. For the time being, the services of the company suited me quite well, but to be honest, there were simply no serious situations. That is, I connected the tariff, chose a portfolio, everything worked somehow. There were mentioned problems with the server, with the inoperability of installed applications. For example, once I was surprised to find that the robot does not execute applications. It turned out that the auto-executor, due to some kind of bug, cannot forward them. Everything was fixed, a working version was installed. But then again, and if I were on vacation, I didn’t notice and ... Well, okay, in the end,

As a result, I met the new year with an amount of about 450 thousand rubles. on the account. At the beginning of 2015, I decided to include an aggressive portfolio of Titan, which showed very good dynamics and was strongly advertised by the company.

For more than 3 years


Over the past few months


Prior to this, somewhere within six months I used the conservative Lira.



Strategy earned a little, but went to the plus. I always try to double-check everything as much as possible, even if everything should be fine anyway. But these 3 months very lulled my vigilance, and, switching to "Titan", I checked only the fact of the server and auto-fulfillment of applications. And he stopped checking the official schedule. But in vain.


For a while there was growth, and I was not at all worried. Then it started to fall, I was still not worried, without this in trading in any way. But the fall continued. At some point, the amount in the account went negative compared to the starting 450 thousand. Now pay attention to the “Titan” chart for this year, there are no points below zero, even with the largest commission.

I began to worry, checked the schedule on the site (it was the end of February). Imagine my surprise when, according to the schedule, I had to be in the black by more than 10% ( 12.22% on the chart ). I understand that there may be some technical deviations ( 1-2% ), but everything was much worse. I called the company.

The Importance of a Financial Advisor
Если вы когда-нибудь захотите работать с с другой финансовой компанией, то подходите внимательно к выбору вашего советника. Из-за его безответственности можно реально пострадать. Совершенно точно не стоит работать с финансовыми советниками Лебедевым и Черницким, с которыми мне пришлось столкнуться при решении этой проблемы.

I described the situation, the financial adviser said that he would try to figure it out and would soon tell me everything. This was my first tomorrow. The call was made on February 24th on Tuesday. As a result, I reinstalled the program - the executor of applications (called "Autopilot"). Allegedly there is a bug in it. In general, I’ll say right away that this measure is akin to the “restart Windows” advice. The probability of solving the problem is extremely small, but you can create the appearance of its solution, maybe the client will calm down. This is the first trick.

And they took the logs for analysis, but did not say anything. By the way, the logs were then asked to be sent every time I called, but I never made a single conclusion on them. This is another soporific measure in the style of “you send, but we will understand” or “already sent? Well, send again the freshest. " And the client sits and thinks: "Well, they took the logs, so they are doing something ...". This is the second trick. As a result, within a week no one answered me on the question.

And the problem was not worth a damn. To understand its source, it was enough to have a little command of school arithmetic. But first things first.


In the meantime, the money was lost. Lost every day. In itself, an expectation of 3 days, when the graph on the site successfully conquers new heights, and in fact there is a drop of 1.5% per day, pleasure, believe me, is not pleasant. Especially when you don’t understand why.

In the end, it was Friday, then Monday of next week. Since no one was going to help me, I started to figure it out myself and downloaded the file with transaction parameters from the official site.

Transactions of "Titan" on March 2, 2015.


Source (excel-table "Transactions" at the bottom of the page)

And checked

What happened in fact


As you can see, this is also the data at the end of the day on March 2, 2015. What is wrong in the screenshot? Let's look at the official dynamics:


That is, in fact, on Monday we grew by 0.28%. And here’s all- important -1.48% :

The first line does not count, it is associated with the outdated T0 mode.


We will calculate it ourselves to know for sure. To get started with the excel table above. For March 2, only 4 transactions:

  • Sberbank 1 : at the time of opening, there was a sale, the price at the beginning of the day was 53.82 , bought up (close short) at 54.59 , total (53.82-54.59) = - 1.43% . Divide by 3 , as the proportion of paper in the portfolio is 33% . Total we get: -1.43% / 3 = -0.477% . We also think for Surgut and Transneft.
  • Sberbank 2 : bought (open long) for 54.59 , price at the end of the day 54.10 , total -0.9% / 3 = -0.3%
  • Surgut : sold (open short) for 39.255 , bought (close short) for 39.595 , total -0.87% / 3 = -0.29%
  • Transneft 1 : at the time of opening, there was a purchase, the price at the beginning of the day was 137510 , sold (close long) at 137880 , total + 0.27 / 3 = + 0.09%
  • Transneft 2 : sold for 137880 , price at the end of the day 136630 , total + 0.915 / 3 = + 0.305%

Summarize, we get -0.672% , we add a commission: -0.1 * 5/3 = -0.166% , total we have about -0.838% . Where did 0.28% come from - a riddle ...

Everything is already so bad, the deception is obvious, but why did I get even less? Let's continue.

Bugs that cost real money


For those who do not understand what this table means
Синие строчки означают «исполнено» (куплено/продано), чёрные — заявка была выставлена, но потом снята, жёлтая строчка — заявка была исполнена частично (была куплена/продана только часть акций по заявке).

The table shows that applications are not always, and to be honest, they are almost never executed at the price from the “ideal” table. They were exhibited, then reset after a minute and exhibited less profitable. For Transneft, and she is the only one who earned that day, this happened as much as 2 times. We calculate the losses on this so-called “slippage”.

  • Sberbank : in table 54.59 , in reality 54.79 , the volume is 66% (closed the old + opened a new one in the amount of 33% ), total (54.59-54.79) * 2/3 = -0.244%
  • Transneft : in the table 137880 , in reality 137170 , the volume is also 66% , total (137170-137880) * 2/3 = -0.345% . Please note that since there was a sale for Transneft, a lower value will go to minus, and a larger one will be a plus, as opposed to a purchase.
  • Surgut : in table 39.255 , in reality 39.205 (take the average of 39.210 and 39.200 for simplicity), the volume is 33% , total (39.205-39.255) / 3 = -0.04%

Total: -0.593% . In total with the previous one it turns out approximately -1.4% . The resulting error is associated with Transneft shares, which I will discuss later.

It turns out that in addition to the fact that the schedule is obviously false (we checked this), even compared with the “honest” table, very significant deviations are obtained. -0.593% - this is a lot for one trading day, the maximum allowable deviation should not be more than 0.1% . It is clear that if such a deviation occurs systematically, then in 20 trading days we will receive more than 10% loss only relative to the official schedule.

In total, instead of a small profit, in reality there are big losses. It is partially possible to guess why this happened on the existing data, but then I will explain everything.

If someone thought this was a bad coincidence, then here is another day.


2 “slippages” occurred during the first 2 hours of trading. Both are negative. Friday of the previous week, the screenshots of which I did not have, also deviated by more than 0.5% in minus. And the very fact of the huge deviation from which it all began says that this happened regularly.

What happened next

Having sorted everything out on my own, I again called the company. Then I still did not pay attention to the fundamental incorrectness of the schedule on the site (by the way, this can still be verified by checking the schedule and table). But I wanted to deal with slippage. I was very surprised that a serious financial company could not do such a simple arithmetic analysis in a whole week. I called, the conversation began with the fact that they started telling me a typical bike for the third time, trying to shift the responsibility onto me. It consists in the following.

How to explain to the client that he is to blame for everything

One share of Transneft, as can be seen in the screenshots, is quite expensive. Since it occupies only 33% in the portfolio , it turns out that if there is very little capital, it becomes impossible to buy and the portfolio does not work quite as it should. The minimum portfolio size is easy to calculate. This is somewhere around 140,000 (we round up the price of Transneft shares from the table above) * 3 = 420,000 rubles . I was not very far from the critical value ( 450 thousand.), but still had a certain margin of safety. Three times (!) The company informed me that the problems could be allegedly due to too little capital, saying that Transneft could not open. This was told to me by my financial adviser, his deputy said, said a technical specialist, who reinstalled the contractor. The guys rely on the fact that the client does not know how or will not count. But essentially it’s just a hoax to shift responsibility.

Having heard this story for the third time, I directly asked to stop lying to me and told about the results of my research. The financial adviser did not argue with this, said something like "yes, yes, you are right, really big slippages, we are dealing with this right now, we will soon tell you about the results." That is, he perfectly understood the real situation. But as a result, the conversation did not lead to anything. I was told another "soon we will figure it out, send the logs again." Later, I left complaints about the poor work of financial advisers Lebedev and Chernitsky, but they also remained unanswered. I think that there were no sanctions for them.

How did it end

At some point, the money became less than 420 thousand , and Transneft really stopped opening. This was the last straw, to use the "Titan" further and hoping that something will work out was already impossible. I switched to Lira again, yet it somehow grew. An additional bonus was that the analysis of the execution of her signals allowed me to better understand the cause of slippage. Another screenshot:


What is most interesting to us here. The plate at the top left is an analog of the table in Excel that we have already seen. With the difference that it contains only the last operation for a particular piece of paper. The table on the right is the actual performance. What is important to note here is the time difference between the appearance of the signal and the execution of the transaction. Calculate:
StockSignalExecutionDifference (sec.)
FGC UES15:27:1315:27:2512
TransneftI hope now it’s clear why the robot can’t buy it

On average, we get about 7-8 seconds. on the "reflection" of the performer. What does this lead to? Let's look at an example for the purchase of Raspadskaya at 11: 25-11: 30 . It is not in the signal table, but it is most indicative.

A bit about Raspadskaya
Для тех, кто не в теме — это одна из самых волатильных акций российского рынка, на которой можно как очень быстро разбогатеть, так и так же быстро обеднеть. С октября прошлого года она выросла с 15 руб. до 50. А когда-то она стоила 260 руб.

The first attempt to purchase was 42.98 , in reality, it was bought from the fifth attempt in 4 minutes for 43.79 . We consider the percentage difference without taking into account the share in the portfolio: (42.98 - 43.79) = -1.9%

This is what delays may be in trading ... Another thing is that in a portfolio like Lira, they can sometimes work as a plus, which On the whole, it will bring the real picture closer to the official schedule.


My first hypothesis is that the Titan is simply not designed for such long delays. Since the strategy is aggressive, the mathematical expectation will lead us at a loss if we delay the execution of the application. It turns out something like this:

“Titan” suddenly saw an opportunity to earn money and said: “It will grow! Come on, buy it fast! ” The contractor is slow, not designed to work with a similar strategy and is waiting for something, as a result, we almost always buy or sell worse, which means we earn less, lose more.

Second hypothesistrying to answer the question why applications already submitted so often are not executed. Strange effects are half-filled bids (remember the yellow line). One application makes up no more than a third of the capital, and this in my case is about 150 thousand.that by the standards of trading, nothing at all. And yellow lines appeared constantly with me. This is very unlikely in the normal case and therefore cannot be explained by chance. In my hypothesis, “Titan”, which is held by too many customers, simply cannot buy shares in them all. As a result, someone succeeds, while someone does not. For those who are unlucky, the performer takes a huge step in half a percent loss. And this can happen several times, as was seen in the example of Transneft in the screenshot. A similar situation, apparently, sometimes occurs with Raspadskaya in Lira. But there it is not so critical.

Based on these considerations, those numbers that are listed in the table on "Titanium" on the official website can happen with a probability tending to zero. And in real life there will always be a deviation to minus (we recall the law of large numbers from probability theory). If we take the average deviation of minus 0.5% , and somewhere it was in the interval that I followed, we get a deviation from the schedule of 2.5% in a week, and 10% in a month , which will correspond to the real situation.

But in any case, in the sense of a “client – ​​financial company” relationship, these hypotheses do not matter. The client should not have thought about this at all. The company advertises a portfolio, shows a graph of the dynamics of its work. The client agrees: “Yes, come on, I’m ready to pay for it. I want my capital to follow this schedule. " In reality, he gets:
  1. An advertising schedule is fundamentally different from what is happening in reality, and naturally for the worse.
  2. Even after clarifying the problem, the company does not solve it, ignores it, does not give intelligible answers, promises to solve it tomorrow and does not solve it.
  3. Also, the topic not covered in the article is outdated software in which nothing can be configured. For example, remove the same time delays of 7-8 seconds. or to limit the loss on slippage is impossible. There seems to be such a function, but it simply did not work. I was even somehow mentioned in a conversation that I can fix the code myself if I want to.


As a result, it turns out that on my account now should be somewhere in the 100 thousand rubles. more than there is. If there were 5 million rubles in my account, the difference would be a million. And it would be even more significant if I myself did not understand the problem. I do not think that there are any legal leverage to recover money in such cases. Usually, under a contract, it turns out that the company can work as it pleases, and customers themselves pay for its mistakes, if any.

Is this normal? I do not know. Maybe I was very unlucky, or maybe this is a completely normal situation in the exchange business. I would be glad if someone shares their opinion or, even better, experience on this topic.

If someone is interested in the question of whether I tried to obtain compensation for losses, then yes, I tried. This is a very fascinating and interesting story, which I will definitely tell if it turns out that the topic is interesting.

A brief summary of the written

  1. If you use trading robots, then a beautiful chart can only be a false marketing trick, in reality everything can be much worse.
  2. It is very important and necessary, especially at the initial stages of the algorithm, to monitor everything, not to be lazy to count and verify, this can help you find a problem at an early stage and avoid losses.
  3. It is important to choose a good financial company. It is especially bad when slowness and complete inability to quickly solve them are superimposed on obvious errors (if you can even talk about the ability to solve problems).
  4. It is also important to work with a good financial adviser.
  5. Poor financial advisors in bad companies have a number of methods for creating the appearance of solving a problem or shifting responsibility from oneself to a client. Among those noted in the article:
    • "Reinstall the software," although it is clear that this is not the case.
    • “Let's take the logs for analysis,” - it was easy to sort out my problem without them. Sending them 3 times in a row is definitely devoid of any sense.
    • "You have too little money for the portfolio to work as it should," - take a calculator and count. In this case, you can only trust him.
  6. If all else fails, take the calculator again and read. Trading sometimes seems to be something mysterious and complicated, but in reality, in order to often figure out something, enough knowledge of school arithmetic is enough.
  7. And do not forget that even if everything is technically good, the strategy can still work at a loss, always be on the lookout.

PS: There can be a lot of incomprehensible things in the sense of terminology and conclusions, so ask, I will explain where you need to. Any other spelling and punctuation errors are better in PM.

PPS: Do not take my article as a complaint. Of course, I was very angry when all this happened, but enough time passed to calm down. I have no profit from this publication. Such companies do not really care about such articles. People who come and put to them at the expense of 10 or 100 million rubles will still not read the “megamind”. And they do not take seriously clients with a capital of half a million. My main thought is that their advertising schedule is lying, though it is lying, it is lying a lot. And I don’t know how many people are sitting or sitting on the Titan with much more capital. And these people did not check anything. And how many more are planning to take advantage of the “wonderful” offer, admiring this constantly growing beauty ... Do not make my mistakes and be careful.

Also popular now: