Trading Nvidia with Daily Aggregated News Sentiment Scores
Summary
This project describes a news-based trading strategy for Nvidia shares. It collects news articles and historical prices, extracts article text, and scores sentiment using TextBlob and VADER. The strategy uses VADER scores, aggregating each day’s highest and lowest readings into an extreme-score indicator. The stated rationale is to avoid trading when strongly positive and negative stories appear together; thresholds then determine long and short positions. The author compares strategy returns with buy-and-hold over March through August 2018.
The project reports an absolute strategy return of 24.72%, compared with 17.76% for buy-and-hold, and a buy-and-hold Sharpe ratio of 0.906. It does not provide the strategy’s Sharpe ratio or enough detail here to assess transaction costs, execution, or robustness. The author notes that VADER may miss business context and that polite news language can obscure severe developments. The short test period and single-stock example limit the strength of the comparison.
Key ideas
- The strategy scores Nvidia news articles with VADER and aggregates daily extreme positive and negative readings.
- Conflicting daily sentiment extremes are treated as a reason to avoid trading.
- Thresholds produce long and short positions, which are compared with buy-and-hold over a six-month period.
- The reported strategy return exceeded the stated buy-and-hold return, but the project gives limited evaluation detail.
- Generic sentiment scoring may miss company-specific context and understated language in news coverage.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.