Hello everyone. I would like to offer my services in data science. I hope it's okay to offer my services on this platform. If it's against the rules, I apologize in advance to all forum members. The key services that might be of interest to you include price and volatility forecasting for primary assets and derivatives, portfolio optimization, risk management, and the development of complicated machine learning models (robots) for trading strategies. I have extensive experience in consulting services in the marketing and financial sectors. My portfolio includes a large number of completed projects using deep learning, classical machine learning methods, mathematical optimization, and much more. If any of the mentioned services or other data science services interest you, or if you're generally open to it, I'd be happy to discuss how you could enhance your business with data science. We could explore ways to define tasks and strategies in data science that align with your business goals. I am a private individual residing in Bulgaria. Ideally, I'm looking for a long-term collaboration. If interested, you could reach me here or drop me a message at Telegram @Ser496
If machine learning and AI etc. is good, then why not use it for own trading and success? Or does it imply that it's just some useless hollow marketing buzzwords?...
Nothing wrong with specializing in one aspect of trading and leaving other aspects to collaboration. What I am more skeptical about is the value of (traditional) ML for trading, as the data is too noisy.
You seem to have missed the point of my ad. I'm not offering a product that either works or doesn't. I'm offering machine learning services, both individually and as a package. It's up to you to decide which services you need and how to integrate them into your trading. Naturally, your earnings will depend on your own decisions. I'm not involved in trading myself; I have different interests. My focus is solely on providing services in this area. Yes, I'm also developing my own advanced trading platform based on deep learning for machine-generated trading strategies, but that's outside the scope of my services.
That's a good point, but in situations with small datasets, deep learning yields poor results. There's nothing sacred about deep learning. It's often useful, but not always. For example, gradient boosting on sequences delivers excellent results despite not being able to "remember" previous samples.
I use a fair amount of statistical learning techniques and I've had a positive experience with offshore researchers before (though it was in a more formal setup at a large fund). However, the above makes it impossible for me to consider your services. Someone who's a potential competitor can not be trusted with any type of sensitive alpha research.
Hello. It's up to you. By the way, have you tried reinforcement learning to make trading strategies? As for myself, it gains very promising results if properly trained.
Tinkered with it, but not enough to find anything conclusive. My prior is that is very prone to overfitting so it's hard to commit to it in earnest. For long-term research, I'm more excited about using DL to extract medium-term alpha from micro-structure signals. In all honesty, though, most of what I use in production is a set of relatively simple regression/classification models combined with smart feature engineering. That's where domain knowledge becomes a competitive advantage.
Got it. Just try RL and share your impression. Chinese guys published a few papers about the topic. There is also some code available in the GitHub with financial environments, though the trading concepts are usually totally wrong. In the near future, investments and trading will be done entirely by neural networks, without human involvement. It will be a battle of trained neural networks.
What is your own impression or practice experience of RL for trading? It already happens, but I doubt it will be useful for Average Joe, unless he is willing to pay and risk much... Yes, trading will be done by bots, and offered as a service to the masses, but on average the averages won't change much, IMO... Ie. a good trader will always beat any bot, IMO, simply b/c that good trader uses similar programs as an aid, but not only...