234 points by aiwhiz 1 year ago flag hide 16 comments
finance_whiz 1 year ago next
Great work! I've been following your previous projects and this one does not disappoint. Can you share more details about the exact factors the AI takes into account for predicting trends?
data_scientist 1 year ago next
The AI uses a combination of technical indicators and machine learning models. Some of the indicators include moving averages, relative strength index (RSI), and trading volume...
coding_enthusiast 1 year ago prev next
Really interesting, how do the machine learning models contribute to the final predictions?
data_scientist 1 year ago next
The machine learning models use the aforementioned technical indicators as features, and they learn the optimal relationships between these features and price trends using historical data...
skeptic123 1 year ago prev next
Too many AI predict and profit systems have failed. How do you prevent overfitting and ensure robustness on real-world data?
data_scientist 1 year ago next
To tackle overfitting, we adopted a hybrid approach using cross-validation, regularization methods and a final ensemble of multiple ML models. The out-of-sample R² score indicates its ability to generalize reasonably well...
machinelearning_maverick 1 year ago prev next
Brilliant! I'm curious about model interpretability, though. Is there a way to explain the AI's decisions intuitively for end-users and perhaps identify blind-spots?
data_scientist 1 year ago next
Some of the approaches we tried for model interpretability are utilizing SHAP (SHapley Additive exPlanations) values, and custom-built decision trees that incorporate the ML models’ knowledge...
backtestingbob 1 year ago prev next
How does this perform compared to common indicator-based trading strategies like MACD, Bollinger bands or simple moving averages?
data_scientist 1 year ago next
In our extensive backtests, this AI algorithm showed consistently better performance in terms of risk-adjusted returns and profitability than popular individual indicator-based trading strategies over the last 5 years...
investor_ian 1 year ago prev next
What about applying this to cryptocurrencies and other non-traditional markets?
data_scientist 1 year ago next
With enough historical data available, there's no reason the AI wouldn’t be able to adapt to other markets like digital currencies. We plan to explore those opportunities in the near future...
code_nerd 1 year ago prev next
What are the primary programming languages and libraries that you used to build this?
data_scientist 1 year ago next
Python was our language of choice due to its strong ecosystem and the vast amount of available libraries. Key libraries include NumPy, SciPy, Pandas, Scikit-learn and TensorFlow/Keras for machine learning...
helpfulharry 1 year ago prev next
Are there any plans to open-source your code to contribute to the AI and finance community?
data_scientist 1 year ago next
We recognize the importance of the open-source community and are considering how we can best share parts of this project with the public while balancing intellectual property-related concerns...