"Finally someone tackles the exploration-exploitation tradeoff in portfolio management. The results on the synthetic data look promising but I'd love to see this tested on more realistic transaction costs..."

"The section on data snooping bias is spot on. I've seen so many strategies that look amazing in backtest and then immediately fail in paper trading..."

"Great paper. The authors actually test on out-of-sample data and show that transformers aren't the magic bullet everyone thinks they are for financial time series..."

"This is exactly what I've been looking for. The documentation is solid and the benchmarks are impressive. Already testing it with my crypto strategies..."

"The alpha decay in alt data is real but this article misses the point. It's not about finding the one magic dataset, it's about systematic data discovery and rapid iteration..."

"Start with Fama-French and build from there. Don't try to implement everything at once. Also, make sure you understand the difference between cross-sectional and time-series models..."

"Good introduction but the author doesn't mention the lookback bias issues with HMMs. You need to be really careful about using future information when estimating states..."

"The MEV considerations are crucial. Traditional market making models break down when miners can reorder transactions. You need to think about this from first principles..."

"Calmar ratio and Sortino ratio are better but they're still not perfect. The real issue is that all these metrics assume normally distributed returns which is obviously wrong..."

"Love the transaction cost modeling and the slippage simulation. Finally someone who understands that backtesting is about finding what won't work, not what will..."