Whenever you're ready, walk me through how you'd approach this.
Explain the trade-offs between using a random forest versus a gradient boosted decision tree (GBDT) on highly noisy, non-stationary financial time-series.