SimpleFunctions Opportunity Discovery

When to use

• The user asks "what are the best trade opportunities right now?"
• An agent needs to surface candidate trades for portfolio consideration
• A morning routine wants a top-N opportunity list
• The user wants to filter by category (macro, geopolitics, sports, crypto)

Prerequisites

npm install -g @spfunctions/cli

No auth required for sf discover. sf cross-venue works without auth. sf edges requires sf login if you want personalized edges against your active theses.

Instructions

Step 1: Pull the quality-filtered opportunity list

sf discover --quality --json --limit 10

sf discover --quality is the agent default. It:

• Aggregates trade ideas, world opportunities, cross-venue pairs, new markets, and contagion gaps
• Filters noisy sports, esports, crypto micro-window, penny-pair, and empty-liquidity candidates
• Applies a diversity pass so one signal family doesn't fill the entire response

Without --quality, the raw output includes everything (useful for breadth, noisy for ranked selection).

Step 2: Filter by category if needed

sf discover --quality --category macro --json --limit 10
sf discover --quality --category geopolitics --json --limit 10
sf discover --quality --category crypto --json --limit 10
sf discover --quality --category elections --json --limit 10

Available categories: macro, geopolitics, crypto, elections, sports, entertainment, science.

Step 3: For top opportunities, pull cross-venue pairs

Cross-venue arbitrage is a high-confidence signal class:

sf cross-venue --preset arb --json --limit 5

--preset arb returns pairs with confidence ≥0.75 AND |Δt| ≤90d. Other presets:

• world — broader pairing for world-state context
• detail — per-pair detail with both venues' indicators
• browse — full unfiltered pair set

Step 4: For each top candidate, inspect

sf inspect <ticker> --json

Verify liquidity, spread, and regime before recommending. A 10¢ edge with a 5¢ spread is a 5¢ realized edge — adjust for that.

Step 5: Format the output

Produce a ranked table:

Always include the realized edge (raw edge minus spread), not just raw edge.

Optional: scan instead of discover

sf discover is the opportunity router. sf screen is the raw filter:

sf screen --iy-min 200 --las-max 0.1 --json

Use screen when you have specific indicator thresholds in mind (yield, spread, regime). Use discover when you want the ranker to surface things for you.

Expected output shape

{
  "status": "ok",
  "data": {
    "opportunities": [
      {
        "ticker": "...",
        "title": "...",
        "price": 72,
        "edge": 12,
        "confidence": 0.85,
        "liquidity": "high",
        "spread": 1,
        "category": "macro",
        "source": "trade_idea" | "world" | "cross_venue" | "new_market" | "contagion",
        "rationale": "...",
        "nextActions": [{ "command": "sf inspect ..." }]
      }
    ],
    "filtered": 15,
    "totalCandidates": 37,
    "regime": { "label": "...", "confidence": ... }
  }
}

Errors and recovery

• Empty `opportunities` array: market might be quiet. Try without --quality for raw breadth, or filter to a specific category.
• Outdated edges: the lastComputedAt field in each item tells you freshness. If it's >1h old, run sf world --json to confirm regime hasn't shifted.

Related skills

• sf-research — topic-specific research (when you know what to look at)
• sf-inspect — deep dossier on one ticker
• sf-pre-trade-check — adversarial check before recommending an entry

Reference docs

• <https://docs.simplefunctions.dev/cli/agentic-cli#opportunity-discovery>
• <https://docs.simplefunctions.dev/api-reference/public-market-data>