Trading the Crowd Before the Print

Inside the Earnings-Anchored Sentiment & Buzz Strategy

Earnings season is the loudest moment in the equity calendar. For a few days around each release, retail investors, professionals and financial media converge on a single ticker — and that convergence leaves a measurable trace. At Stockpulse we’ve spent years quantifying that trace. Our latest white paper asks a direct question: can the shape of the conversation before an earnings release tell you something about the move that follows?

Over a 3.5-year simulated backtest on the Nasdaq-100, the answer looks encouraging. Here’s the short version.

The idea in one paragraph

The strategy watches two signals for every covered security: buzz (message volume, normalised against that stock’s own historical baseline) and sentiment (the net positive-versus-negative tone of those messages). Neither is interesting alone. What matters is the joint state — an abnormal spike in attention combined with directional conviction. Two to three days before a scheduled earnings release, the engine checks whether a stock’s buzz/sentiment state matches a pattern that has historically preceded a tradable move in that specific name. If it does, a long or short position is opened in the underlying equity. If it doesn’t, the strategy sits the quarter out.

Why per-security calibration matters

The design choice we’d most defend is the refusal to use a single, universe-wide threshold. The media dynamics of a mega-cap semiconductor name look nothing like those of a consumer-staples constituent; different baseline volumes, different audiences, different reaction functions. So every stock gets its own optimisation: what counts as “unusually high buzz with strongly positive sentiment” is defined relative to that stock’s own history, not an index-wide constant.

This is also, candidly, the strategy’s biggest source of model risk. More on that below.

What the backtest showed

Over 16 January 2023 – 8 July 2026, a $1.0m notional book grew to roughly $2.87m in simulation.

The trade-level picture matters more to us than the headline. Across 531 closed trades in 94 distinct names, the win rate was 55.6% with a profit factor of 1.58 and a payoff ratio of 1.34. That combination is the one we want to see: edge coming from a modest hit-rate advantage plus slightly asymmetric outcomes, spread broadly — not from three lucky trades carrying the record.

Holding periods were short and event-shaped, as the design implies: a median of 8 calendar days, a mean of 10.7.

Genuinely two-sided

273 long trades, 258 short. Both sides profitable independently — 57.9% of longs and 53.1% of shorts were winners. The strategy isn’t a long book with a hedge stapled on, and it doesn’t need a rising market to work. That structural balance is what keeps it from quietly becoming an equity-beta bet.

Compared to the peer group

We benchmarked against the Barclay Equity Long/Short Index, which tracks a broad universe of equity-oriented hedge funds that trade both sides — a style benchmark rather than a market index, and the right comparison for a strategy that isn’t explicitly market-neutral either.

Over the common 42-month window (Jan 2023 – Jun 2026):

• +190.6% vs +44.6% cumulative
• +23.3% annualised alpha
• Correlation of just 0.25 (R² ≈ 6%)
• Information ratio 1.04

The low correlation is arguably the more important number. A return stream that’s largely orthogonal to what peer long/short managers deliver is additive inside an existing allocation rather than duplicative — which is a different and more useful property than simply being bigger.

One honest note on the comparison: the index posts a higher Sharpe (2.14 vs 1.48 on monthly data) despite producing a third of the return. That’s an artefact of index construction — averaging hundreds of funds compresses volatility to around 5% and mechanically lifts the ratio. It isn’t an investable programme. The meaningful comparisons are absolute return, alpha and correlation.

Where this goes next

The natural next steps are the ones the numbers themselves point to: walk-forward validation of the per-security optimisation, an explicit cost and capacity model, and benchmark-relative attribution over a longer and more varied set of market regimes.

The full white paper — including monthly return breakdowns, extended tail analytics (VaR, CVaR, Ulcer Index, skew and kurtosis), underwater curves and trade-level distributions — is available to professional and qualified investors on request.

Please contact us (use the contact form below) if you want to receive a copy of the full white paper.

All performance figures reflect a hypothetical backtest applied to historical data. Simulated results have inherent limitations: they are constructed with the benefit of hindsight, do not represent actual trading, may not reflect the impact of transaction costs, liquidity constraints, financing or taxes, and can differ materially from live results. Past and simulated performance is not a reliable indicator of future results. This article is for informational purposes only and does not constitute investment advice, a recommendation, or an offer to buy or sell any financial instrument. Any investment carries the risk of loss, including total loss of capital.