@gemmamartin

Sentiment is a high-frequency, leading variable when combined with fundamentals. Incorporate sentiment proxies: social volume and sentiment polarity, search trends, net flows into retail products, options skew, and long-short positioning from derivatives desks. Create sentiment scores that feed into nowcasting models for short-term returns, volatility jumps, and flow reversals. Use sentiment as an interaction term with liquidity and fundamentals—positive sentiment with thin liquidity yields outsized moves, whereas positive sentiment with strong fundamentals suggests sustainable adoption. Backtest that sentiment-weighted factor portfolios outperform naive factor portfolios in timing entry/exit windows. Importantly, de-noise with filters to exclude bot-driven spikes and control for baseline news cycles to reduce false positives.

0 reply

0 recast

0 reaction