Turning Twitter Sentiment into a Monthly S&P 500 Portfolio

At Context Analytics, we turn the conversation on Twitter into structured, tradable signals. Our S-Factor Feed scores that conversation for thousands of securities every minute of every day, giving traders and investors the building blocks for sentiment-enhanced strategies.

In this post, we revisit the Monthly Long-Only Strategy, built on the Raw-S factor and applied to S&P 500 constituents. We introduce the monthly sentiment metric behind it, walk through how the portfolio is selected and rebalanced, and look at what the signal rewards and how it shapes the holdings.

From Daily Tweets to a Monthly Signal

Monthly Weighted Sentiment aggregates a month of daily Twitter sentiment scores into a single value per stock, with the most weight on the final days of the month, closer to the rebalance date. It is built in two steps.

1. Raw-S factor: Each day, the S-Factor Feed aggregates the sentiment of the past 24 hours of Tweets about a security into its Raw-S value.
2. Exponential Monthly Weighting: Each day's Raw-S is multiplied by a decay factor (λ = 0.9) raised to the number of days left in the month, and the weighted values (w_raws) are summed over the month.

Here N_m is the number of days in month m. The last day of the month carries a weight of 1.0, a day one week earlier about 0.48, two weeks earlier about 0.23, and the first day of a 30-day month about 0.05. The score therefore reflects where sentiment stands as the month closes, while still counting how persistent it was.

A stock is only scored if the conversation around it is active. We require an average of at least one S-Volume per day over the month, which keeps thinly discussed names, whose scores rest on a handful of Tweets, out of the ranking.

Building the Portfolio Each Month

The portfolio is rebuilt once a month from the S&P 500 stocks with the highest Monthly Weighted Sentiment, weighted by market capitalization.

1. Universe: Only stocks that are S&P 500 members on the rebalance date are eligible.
2. Scoring: At each month-end, every eligible stock receives its Monthly Weighted Sentiment for that month, and stocks below the activity threshold of one S-Volume per day are set aside.
3. Selection: The 200 stocks with the highest scores make up the portfolio.
4. Weighting: Holdings are weighted by their month-end market capitalization, so the portfolio is directly comparable with a cap-weighted benchmark such as the S&P 500.
5. Execution: Positions are entered and exited at the Market on Close (MOC) price on the fourth market day after month-end, and held until the next rebalance.

The four-day gap between the end of the scoring month and the trade is deliberate. Every input to a rebalance (sentiment, market capitalization and index membership) is known days before the order is placed, so the backtest does not rely on information an investor would not yet have had.

What the Signal Rewards

Monthly Weighted Sentiment rewards stocks that are both widely discussed and discussed favorably, and it leans toward how that conversation looks at month-end.

• Tone and intensity together: Because daily Raw-S values are summed rather than averaged, a stock with a large, consistently positive conversation builds a higher score than one with a few very positive Tweets. The signal measures conviction across the crowd, not just its mood.
• Recency: With λ = 0.9, the final week of the month carries more weight than the first three weeks combined. A shift in tone late in the month moves a stock up or down the ranking quickly, which keeps the portfolio aligned with the current narrative.
• A broad, index-like portfolio with a tilt: After the activity filter, roughly 360 to 490 constituents are scored each month, so the portfolio holds about half of the eligible names. Combined with market-cap weighting, the portfolio stays close to the index in character while leaning away from the stocks the crowd views least favorably.

Portfolio Performance

Since 2018, the strategy has delivered a 249.7% cumulative return, ahead of SPY's 226.2%.

Monthly holding periods are from February 6, 2018 to September 4, 2026.

• Higher returns: The strategy's annualized return exceeded SPY's by about 1.0 percentage point (15.92% vs 14.95%).
• Comparable risk-adjusted performance: Volatility ran about one point above the index, so the Sharpe ratio matched SPY's (1.01 vs 1.01) and the Sortino ratio was slightly higher (1.84 vs 1.81).

What Drove the Results

The same features explain both where the strategy earned its edge and where it gave ground. Large, heavily discussed stocks that the conversation favored entered the portfolio at meaningful weights, and their strongest months drove much of the outperformance. The tilt cuts both ways: popular names can also fall sharply when the narrative turns.

The result is a portfolio that has kept pace with the S&P 500 in weaker markets and pulled ahead when sentiment-driven leaders rallied, at the cost of somewhat higher volatility than the index.

Putting Sentiment to Work

Sentiment moves prices, and Twitter remains one of the fastest real-time public forums into what investors think. Aggregating that conversation with Context Analytics' S-Factor Feed turns a noisy daily stream of messages into an actionable signal that captures both how strongly and how recently the crowd has formed a view.

This Monthly Long-Only Strategy shows how that signal can be put to work in a simple, transparent, index-like portfolio. To learn more about the S-Factor Feed and our other data-driven solutions, visit contextanalytics-ai.com.