looking for a partner with a high-frequency trading system - page 10
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Hello all.
Loved this thread, resting my brain.
Lots of stuff mixed in, I can't say it's interesting or important.
I'll start with an anecdote or a sentence. Looking for a programmer working in EasyLanguage?
What's the gimmick actually? Perhaps there is a customer who wants to get a program to work in the market? Desire is good.
Where to start? A programmer comes in, we give him a job, to make a profitable system?
But you need a person with experience as a trader, not as a programmer. Or at least two people?
What about all sorts of systems and robots . My opinion is that it is easy to do. The simplest one based on the Bollinger was developed 5 years ago. (Just in case, I should warn you, I have no experience and I'm not looking for work).
The question is? How much % do you need per year? If 30% then Bollinger as a base, will do.
Of course making it work on at least 5 forex pairs and working with data from different time intervals.
My point is that there are already ready working systems that work.
If we assume that someone on this thread has a working system and wants to sell it, it's unrealistic. It would take a long time to explain why.
What do you mean by high-frequency trading?
And why HFT?
Are other trading options not acceptable?
Sorry to add for partner83 I hope my advice helps.
Put up a dozen computers, open several accounts with reputable brokers.
Hire traders working 24 hours a day, each with a daily limit on the account and a percentage of profit. And change them when you trade badly, choosing the best ones.
:-). I need some reference material for interacting with Microsoft Excel via DDE communication.
but there are no requirements for a partner?
Regarding all sorts of systems and robots . My opinion is that it's easy to do.......
Naivety is when you believe that all you have to do to land in the sun
♪wearing sunglasses♪
Here are snippets of an interview with Anton Yereshko, a man who has some authority (earned it on the HDL)
- Tell us about your team.
- Of course, each participant has priority areas, but we are to a large extent "universal soldiers". I believe that such team organisation has a lot of advantages, as there is always a back-up, a backup plan, we can always switch over and, if there are not enough resources, solve a particular problem together. As far as I know, few private teams involved in trading on the Moscow Exchange can distribute and concentrate their efforts so effectively. Our initial goal was to spread our presence as widely as possible - America, Singapore and so on. We are trying to evolve, and mathematical approaches help us do that. Our strategies are not just some little technical inefficiency that we found and exploited for three years until it disappeared. Our team tries to be as versatile as possible in terms of forecasts and calculations. We have a good background: we come from the Academy of Sciences.
-Do you do research every day?
- Yes, every day. There are also organisational aspects.
-What kind of connection do you have to the stock exchange and how much does this infrastructure cost?
- There are servers and a direct connection. I can't say how much it costs, there are certain nuances.
- In what programming language are your robots written?
- The robots themselves are written in C++ while the hypotheses are tested in C#.
- What operating systems are on the servers?
- Windows, but that's already atavistic. We are currently switching to Linux: it's a faster and more reliable operating system. We used Windows because historically it was so; the exchange provided a Linux API comparatively recently. However, on the futures market there are still black boxes (routers), which sometimes negate all the work to find the best and fastest set-up.
- Have there been punctures in algotrading or bugs in the code?
- Sometimes. This is very destabilizing. Testing takes a very long time. It is a thorough process that is carried out starting with the virtual implementation of the strategy. Only after repeated virtual trades, after stress-tests, after finding optimal solutions, the strategy is coded into a trading robot for a comprehensive run-in in the real market. Everybody knows about technical failures, which are sometimes caused by the exchange. They are disruptive to everyone.
- How do you evaluate trading strategies?
- We have many criteria. Simple ones are income, income dispersion, number of deals and transactions, maximum volume of transactions.
- How many strategies do you have?
- We have several robots for each market. There are several accounts, and there are at least two machines for each instrument. They implement their own approaches and can change or switch within themselves. The result is a set of strategies. In addition to the high-frequency approaches there is arbitrage, pair trading.
- Are the strategies that you have used on the LPI dying?
- Strategies based on technical dominance are very much subject to competition, and they lose power. If you make something universal or introduce additional parameters, which evaluate technical state of the market, then chances to be afloat for a long time are higher. In most cases, people catch some small inefficiency, which worked because before there were no people purposefully destroying it. Over time, such inefficiency disappears.
-So your algorithms are working now?
- There are some that work, and some that do not. We are trying to make them more versatile. We are changing and constantly adapting. In order not to find yourself in a situation where all strategies stop working and there is nothing new, you have to come up with something special all the time. Every day we have to work for the future.
- How often, on average, do your strategies stop working?
- Strategies that are implemented based on technical dominance break most often. Models that contain parameters that include an assessment of the technical component die less often, or very rarely if they are implemented competently.
- How do you feel about thechanalysis programmes such as Wealth-Lab, TSLab and the like?
- I think it's a near-market industry around trading automation. From the clients' point of view, it's a no-brainer. From the authors' point of view, it is a niche where people pay for the product. Writing code is very easy, the hardest part is coming up with a strategy and properly assessing how workable it is. Everything else - writing APIs to the market and so on - is routine and simple. Any normal programmer can do that.
As for classical thechanalysis - it certainly does not stand up to criticism. I have not tried it, but I think it does not work. Some people have invented it and used it somehow, maybe it worked for them. And it all has to be customized, plus there are some timeframes that people put themselves into. So, I think there has to be a kind of an own seed, a puzzle and a way to evaluate the market, a kind of indicator to estimate supply and demand, to predict and adjust.
- Tell us about this process.
- We have our own fairly complicated complexes, which we call tweakers. There we program all of our hypotheses and test them on historical data. This is a standard procedure. The history is represented by tick data, there is just a flow of numbers. Any high-frequency, and indeed all trading is based on information. High-frequency algorithm reacts to each market data update, so there is no breakdown into 15-minute intervals or any other. That is nonsense. It is possible to come up with a strategy that trades on minute intervals. But why not 14 minutes or 13? It's all very subjective. We have a virtual exchange, where there is a virtual stream of quotes and virtual matching, and we get the results, whether the strategy is profitable or not. This is the standard scheme to check on historical data. But there are a lot of nuances: how matching is done, why a bid is executed this way and not that way. Everything is very relative. There is a hypothesis at each stage that is used to build a strategy. So, in addition to trading robots that do not take up much space, get the data, work off, and that's it, there are algorithms that help to test the idea.
- How do algorithms work?
- You can create a strategy only after you understand the mechanics of what is going on in the market, operating with understandable data provided by the market. For instance, you have to clearly understand what the order book is, what the deal flow is, what the features of it all are and what the rules are for its distribution. Next, you must answer the question of how to profit from the permanently fluctuating price of an instrument. A forecasting hypothesis emerges. The hypothesis is formalized, parameterized, coded and then the virtual trading takes place. Each such step has an infinite number of nuances and ways of its implementation. If the approach turns out to be unpromising, everything is repeated again. It is clear that you should first evaluate the system and then write a robot.
- Do you raise money from investors?- We only trade on our own, we do not have any investors. We do not need them. The situation in the financial world right now is such that finding money is a much easier task than turning that money around. If you can't "digest" the sums raised, additional investors are an unnecessary burden, and often with unfortunate consequences.
- Inyour opinion, is it realistic for non-programmers and non-mathematicians to make money on the market?
- When there is real competition, there is nothing to do without these skills. Algorithmic trading is in one way or another connected with statistics, calculations, methods of models efficiency evaluation, quality of meshes and their parameters. It is all mathematics. However, it is not enough to have this knowledge without the craft of programming: the chances of succeeding are greatly reduced. Successful algofonds seek engineering graduates with excellent mathematical backgrounds.
- Hasit become harder to make money in recent years?
- Yes, investment market activity has declined smoothly in recent years. It is now characterised by rare spikes. You have to adapt to this. Recently, many foreigners have entered the futures market. They are making a big turnover even when the market is neither alive nor dead, and it can be felt. There are people who can make a big turnover, but do not earn anything. Those who are not versatile and cannot come up with new strategies leave. They are both loners and small teams.
- Do you think that the loners who do algorithmic trading will eventually leave the industry?
- Of course, this is a trend. Until a few years ago, it was possible to exist. My prediction is that if people don't unite and develop, such private trading will die very soon. Everything will be eaten up, they will simply be kicked out of the field. You have to develop, you have to scale strongly and put a lot of effort into it. You have to look for your own kind, but not to hang out at parties, but to do business. There is an arms race, and that is a fact. In Russia, to enter this market, you simply have to have your programme certified by the stock exchange, and the entry threshold is 30,000 roubles. In America, there are serious requirements for people who want to get into this business, and the starting threshold for the DMA is half a million dollars.
-Why are you rarely seen at market events?
- We are a fairly closed team, we try to devote more time to work. Sometimes we attend conferences, events and seminars organised by the stock exchange. But we try to keep this more as a job than as our main life. It is unprofessional to mix everything in one pile and it affects the results not for the better. For example, if you have a wish for the exchange, you don't have to write about it on Facebook, you just have to come and say what you need.Of course making it work on at least 5 forex pairs and working with data from different time intervals.
As for classical thechanalysis - it certainly does not stand up to criticism. I haven't tried it, but I don't think it works. Some people came up with it and used it somehow; maybe it worked for them. And it all has to be customized, plus some kind of timeframes that people put themselves into on their own. So I think there has to be some kind of own grain, a mystery and a way of evaluating the market, you could say your own indicator for evaluating supply and demand, predicting and adjusting.
Systems programming languages.
There has to be a lot of everything.
More than 5 brokers, more than 7 programs used, 4-5 programming languages.
Several components of data transfer programs and user programs,( if you distribute system robots) identifying user IP and receiving commands in real time.
Regarding // all pairs are correlated with each other//. I think this is a mistake, but if you wish, then yes.
The pairs should be selected according to their expected yield at the moment. I.e. Select according to market conditions.
:-). A complete list of Excel DDE commands can be found in the MSDN documentation provided by Microsoft.
I can make a large sum, but you can't work with large sums. tchk.
Why are you picking on hft in the first place? You want high interest with minimal risk? It will not work.)
Why don't you tell us your profit and drawdown requirements?
Acceptable. Offer all kinds of trading, working, profitable. I will consider.
I`ve been trading news for more than two years in semi-automatic mode. The result is positive, except that I often have to change brokers. "Toxic traffic, you know..."
I have monitored some signals just for the sake of sport.
I want access to serious liquidity, for example EBS system, but the entrance there is from 500k zeleni - I do not have such sum yet.
I already tried other types Kurenecks, Integral, Lmax and similar - very little liquidation on news stays in the market and slips heavily.
Our MMvb is not liquid at all...
The CME and the Americas in general are not suitable - all the infrastructure is in Europe.