Skip to content

Social Signal Scoring SOP

Purpose

Score the social_signal rubric (0-10) for watchlist symbols that have Discord social signal data. This data comes from ~97 Discord trading channels aggregated in a Cloudflare D1 database.

Data Source

Social signal data appears in scoring packets under Social Signal (Discord). It includes: - Signal counts (BUY/SELL/SHORT) and confidence breakdown (STRONG/MODERATE/WEAK) - Per-channel breakdown with win rates and signal counts - Options flow (calls vs puts) - Analysis sentiment and price targets from D1

Scoring Anchors

Score Label Criteria
0-1 Strong bearish consensus Multiple high-WR channels signaling SELL/SHORT. Cross-channel convergence on bearish view.
2-3 Bearish lean More SELL than BUY signals, or bearish signals from trusted channels.
4-5 Neutral / no signal No meaningful Discord signal, or mixed signals canceling out. Default when data is sparse.
6-7 Bullish lean Moderate BUY signals from multiple channels. Some cross-channel convergence.
8-9 Strong bullish consensus Multiple high-WR channels signaling BUY with STRONG confidence. Cross-channel convergence.
10 Euphoric / crowded Universal BUY across all channels. Flag as potential crowded trade.

Evaluation Steps

  1. Check channel quality: How many channels are signaling? What are their win rates? A single channel with 50% WR carries less weight than three channels with 80%+ WR.

  2. Assess convergence: Are independent channels agreeing (same direction)? Cross-channel convergence from high-WR channels is a strong signal.

  3. Read confidence distribution: Mostly STRONG confidence suggests conviction. Mostly WEAK suggests tentative/speculative.

  4. Check options flow: Do options flow confirm stock signals? Heavy call buying alongside BUY signals is confirming. Divergence (calls + SELL signals) warrants investigation.

  5. Consider recency: Are signals clustering recently (7d count high relative to 14d)? Accelerating activity may indicate developing catalyst.

  6. Flag extremes: At 0-1 or 9-10, explicitly note potential overreaction or crowded trade risk.

Recording

openfin review score <SYMBOL> -m social_signal --score <0-10> -r "<rationale>" --run-id <RUN_ID>

When No Social Data Exists

If a symbol has no social signal data in its scoring packet, skip the social_signal metric for that symbol. The composite score computation handles partial scores by redistributing weight.