Applying Machine Learning to Trading: A Practical Learning Path
Summary
The document is a discussion about turning formal statistical machine learning coursework into practical finance projects. The original questioner has studied regression, classification, neural networks, support vector machines, and principal component analysis, and seeks real data and finance context for applying those methods, possibly in less competitive markets. It does not present a trading strategy or evidence that machine learning can produce profitable signals.
Responses suggest using forecasting competitions that provide prepared datasets, and building a portfolio project with an automated trading platform, market data, and a public write-up or code repository. These are project-development suggestions rather than validated research guidance. The document gives no detailed advice on data quality, leakage, transaction costs, backtesting design, or deployment risk, so any resulting strategy would need careful evaluation before live use.
Key ideas
- Forecasting competitions can provide ready-to-use datasets for practice.
- A project can combine market data, a trading interface, and a documented algorithm.
- Public code and written explanations can help demonstrate applied skills to employers.
- The discussion offers learning resources and career suggestions, not evidence of trading profitability.
Tags
Full text
# Applying my Machine Learning class (possibly to small markets) # Applying my Machine Learning class (possibly to small markets) I've finished a university course in statistical machine learning that covered topics such as regression, classification, neural networks, SVM, PCA etc. The class was quite tough and rigorous (we had Bishop as a textbook) and we implemented the algorithms, but I'd like to solidify my knowledge by actually going out and applying the techniques to more real world material, preferably in Python. The goal of this is to have something to show employers aside from just a class. I have no illusions of spending a month working and being able to beat currency markets or whatever, but while I'm doing this, I'd like to test it on markets that aren't flooded with professionals and see what happens. I'm looking for resources/guides to get started - I'm not sure if I need data scrapers, need to set up a database etc. Basically just good resources to applying the stats I've learnt and possibly a primer on the finance I need to know. Totally open to different directions I should go in. Apologies if this is a duplicate, I did try searching but mostly saw people hoping for a 'magic bullet' to crush markets. I'd be super happy with (optimistic I know) setting up some sort of auto trader for crypto markets even if it lost a bit, just for the experience. edit: I should also add that I've done 5 computer science classes so am comfortable with programming in general, and have done a bit of stochastics through our stats department ## Answer by LazyCat (score 2) https://quant.stackexchange.com/a/35411 Try to find online forecasting challenges on sites like kaggle.com They would often provide you with the data, so you don't need to worry about it. Here's a good example (finished): https://www.kaggle.com/c/two-sigma-financial-modeling#description ## Answer by sen_saven (score 1) https://quant.stackexchange.com/a/35414 I would do the following: - Use a free autotrading platform (Interactive Brokers API, metatrader etc) - Get data from Quandl - Upload any trading algorithm to github. - Create a blog and write about the algorithm you have come up with. These will give you something to show to potential employers.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.