Academy notes
Articles and updates from SoloQuant Academy.
Hugging Face Transformers for finance after work: classify a toy headline with pipeline()—not alpha. Educational only. Not investment advice.
Keras time series after work: window a series into supervised pairs and try a tiny MLP baseline. Educational only. Not investment advice.
TensorFlow vs PyTorch for trading research after work: pick one stack and avoid dual maintenance. Educational only. Not investment advice.
PyTorch 2 after work: try torch.compile on a tiny module and keep a correctness baseline. Educational only. Not investment advice.
scikit-learn trading-signal toys after work: TimeSeriesSplit classification with leakage controls. Educational only. Not investment advice.
PyTorch for finance beginners after work: tensors and a one-layer loop before any LSTM hype. Educational only. Not investment advice.
Statsmodels time series after work: fit a tiny ARIMA and read residual checks—not a trading bot. Educational only. Not investment advice.
SciPy signal tools for finance after work: detrend and Savitzky–Golay on toy prices you can audit. Educational only. Not investment advice.
NumPy for quantitative finance after work: vectorized returns, broadcasting, and reproducible RNG. Educational only. Not investment advice.
Polars LazyFrame for beginners after work: scan_parquet, filter early, collect once. Educational only. Not investment advice.
Polars vs pandas for trading data after work: when to stay, when to scan Parquet lazily, and when a hybrid wins. Educational only. Not investment advice.
Pandas 3.0 string dtype after work: cast tickers and notes off object columns so groupbys stay honest. Educational only. Not investment advice.
