ADDENDUM: Actionable Edge Mapping + Prediction Market Ladder

Stocks Desk — MERGED FINAL

Base: Claude Opus 4.6 (complete, unmodified) + Panel Additions from Gemini Pro, GPT-4.1, Grok-4-Fast

DeepSeek V3: reviewed — no unique additions included

This addendum supplements the original Stocks Desk spec. It does NOT replace the existing spec — it adds two critical layers: (1) mapping every edge to a specific trade on a specific platform, and (2) a systematic methodology for scanning prediction market ladders for stock-related contracts.

TOPIC 1: ACTIONABLE EDGE OUTPUT

1. Edge-to-Trade Mapping Table

2. Actionable Output Format

JSON output sent via Telegram and to other desks:

{
  "edge_id": "STOCKS-20260315-NVDA-EARNINGS-001",
  "desk": "stocks",
  "timestamp": "2026-03-15T10:30:00Z",
  "edge_description": "NVDA earnings in 2 days. Analyst revision momentum +8%, whisper number beats consensus by 12%. Kalshi and Polymarket post-earnings price contracts are underpriced.",
  "signal_type": "earnings_surprise_prediction",
  "ticker": "NVDA",
  "event": "Q1 2026 Earnings, reporting Mar 17 after close",
  "confidence_level": 72,
  "supporting_signals": {
    "analyst_revision_30d": "+8.2%",
    "whisper_vs_consensus": "+12%",
    "insider_activity": "No Form 4 sales in 90 days (neutral-positive)",
    "options_implied_move": "8.5%",
    "historical_earnings_surprise_rate": "85% beat in last 8 quarters"
  },
  "platforms": {
    "Kalshi": {
      "action": "Buy YES on 'NVDA above $950' (Mar 18, post-earnings settlement)",
      "current_price": 38,
      "fair_value": 52,
      "edge_cents": 14,
      "edge_after_fees": 12,
      "liquidity_available_usd": 600,
      "max_suggested_size_usd": 300
    },
    "Polymarket": {
      "action": "Buy YES on 'NVDA above $950' (equivalent contract, check daily for listing)",
      "current_price": 36,
      "fair_value": 52,
      "edge_cents": 16,
      "edge_after_fees": 15,
      "liquidity_available_usd": 200,
      "max_suggested_size_usd": 100
    },
    "Broker": {
      "action": "Buy 100 shares NVDA at market ($920), or buy Mar 21 $940 call",
      "current_stock_price": 920,
      "target_price": 980,
      "stop_loss": 880,
      "risk_reward": "1.5:1",
      "edge_pct": 3.5
    },
    "Options": {
      "action": "Buy NVDA Mar 21 $950 call at $15.50 (if implied move is underpriced)",
      "implied_move": "8.5%",
      "model_expected_move": "10.5%",
      "edge_pct": 2.0
    }
  },
  "time_sensitivity": "HIGH — earnings in 2 days. Prediction market prices will adjust as event approaches.",
  "edge_persistence": "Decreasing — sharpest edge now, will compress as earnings approach and volume increases.",
  "risk_factors": ["Guidance matters more than beat/miss for NVDA", "Sector rotation risk if macro deteriorates", "Options IV crush post-earnings reduces option value even if direction is right"]
}

Telegram summary format:

STOCKS EDGE: NVDA Earnings (Mar 17 after close)

Beat probability HIGH: revision momentum +8%, whisper +12% vs consensus
Historical: 85% beat rate in last 8 quarters

Kalshi: Buy YES 'NVDA above $950' @ 38c | Fair 52c | Edge 14c | $600 liq
Polymarket: Buy YES 'NVDA above $950' @ 36c | Fair 52c | Edge 16c | $200 liq
Broker: Buy shares @ $920 | Target $980 | Stop $880
Options: Buy Mar 21 $950C @ $15.50 | Implied move 8.5% vs model 10.5%

Confidence: 72% | Window: 2 days
Risks: Guidance > beat/miss, IV crush, macro

Table format for multi-platform edges (Telegram): [PANEL: GPT-4.1]

3. Platform Priority Ranking

TOPIC 2: PREDICTION MARKET LADDER STRUCTURE

1. Desk Data Mapping to Prediction Market Contracts

Note: Not all stocks or event types are listed on both platforms at all times. Pull Kalshi and Polymarket APIs daily to monitor available contracts. When a ladder exists on both, scan both and trade the better-priced side.

2. Sharp Reference Identification

3. Probability Curve Construction Method

For stock price thresholds:

• Distribution: Log-normal (baseline) with skew adjustment from the options IV smile.
• Inputs: Current stock price S, time to expiry T, risk-free rate r (SOFR from FRED), dividend yield q, and the implied volatility surface from the options chain.
• Method: For each Kalshi or Polymarket threshold K:
  1. Find the options IV at strike K for the matching expiry (interpolate from the chain if needed).
  2. Calculate d2 = [ln(S/K) + (r - q - sigma^2/2) * T] / (sigma * sqrt(T)).
  3. P(stock > K at expiry) = N(d2) where N is the standard normal CDF.
  4. Cross-check using the call spread: [C(K-delta) - C(K+delta)] / (2*delta) gives the model-free probability density around K.
• Earnings adjustment: If earnings occur before expiry, the options-implied move already includes the event. Use the pre-earnings IV (not post-IV-crush IV) for pre-earnings trades.
• Skew calibration: Inflate downside standard deviation by 10% when put IV is significantly elevated relative to call IV (negative skew environment). [PANEL: GPT-4.1]

For economic data (CPI, NFP, GDP):

• Distribution: Normal, centered on the consensus estimate, with SD derived from historical forecast errors.
• Historical SDs: CPI m/m: SD ~0.10% (actual minus consensus). NFP: SD ~70K jobs. GDP: SD ~0.5%.
• For count-based data (NFP): A Poisson distribution may be more appropriate than Normal when the count is small relative to its variance — evaluate on a case-by-case basis. [PANEL: GPT-4.1]
• Anchoring: If GDPNow or Cleveland Fed nowcast diverges from consensus, shift the mean toward the nowcast (weight nowcast 60%, consensus 40%, based on historical accuracy).
• Kalshi/Polymarket thresholds: Map each threshold to a Z-score. P(CPI > 3.4%) = 1 - Phi((3.4 - mean) / SD). Apply to both Kalshi and Polymarket ladders independently.

Recalculation frequency:

• Stock prices: Every 15 minutes during market hours (options chain updates continuously). Trigger an immediate recalc when the underlying spot price moves >0.25% or IV changes by >0.5%. [PANEL: Gemini Pro]
• Economic data: Daily until the week of the release, then every 6 hours in the final week as consensus estimates tighten.
• Fed rate: Weekly, or immediately after any Fed speaker event or major economic release that shifts rate expectations.

4. Ladder Scanning Methodology

Step-by-step process:

1. Pull all stock/index/econ contracts: Call Kalshi and Polymarket APIs. Group by underlying (AAPL, SPY, CPI, NFP, etc.) and expiry date. For each underlying, pull ALL available thresholds — scan the full ladder, not just the main line.

2. For each contract, determine the appropriate sharp reference:

   • Stock price contracts → options chain via Polygon.io.
   • Economic data contracts → nowcast models + consensus.
   • Fed rate contracts → CME FedWatch.
   • Vol/move contracts → options IV + historical realized vol.

3. Calculate fair probability at each threshold using the methods above.

4. Calculate raw edge: raw_edge = |fair_probability - contract_price|. Direction: if fair > price, buy YES. If fair < price, buy NO.

5. Net edge after fees: Subtract 1.5c for Kalshi, 0.5c for Polymarket.

6. Apply signal overlays:

   • If insider cluster buying detected for a stock → boost YES-side confidence by 10%.
   • If credit-equity divergence is active → boost NO-side confidence for index contracts.
   • If VRP is extreme → adjust probabilities (high VRP = overpriced options = prediction market likely overpricing tails too).
   • If earnings are imminent → flag and increase position size limits (high-conviction window).

7. Filter: Minimum 3c net edge, minimum $50 liquidity, minimum 55% confidence. Additionally, require time_to_expiry > 6 hours — contracts expiring within 6 hours have execution risk that degrades edge reliability. [PANEL: Grok-4-Fast]

8. Rank: Sort by net_edge * confidence. Flag economic data contracts separately (they settle at a known time — time-to-settlement matters more). Explicitly flag "Tail Opportunities" where fair probability is <15% or >85% — these are often the most significant mispricings. [PANEL: Gemini Pro]

9. Track edge persistence: Log every detected edge. For stock contracts, edges near earnings tend to grow as the event approaches (mispricings widen when uncertainty is highest). For economic data, edges tend to shrink as consensus tightens. Re-scan at least hourly; more frequently near market close or event time.

10. Cross-platform comparison: For any edge found on Kalshi, immediately check Polymarket for the equivalent contract. Trade the platform with the better net price. When both have acceptable liquidity, split the position across both platforms.

5. Edge Examples with Real Numbers

Example 1: AAPL Post-Earnings Price Threshold

• Setup: AAPL reports earnings tonight. Current price $195. Options imply 5.2% move.
• Consensus: Revenue beat expected (analyst revisions +4% over 30 days). Guidance uncertain.
• Options-implied probability distribution: Using AAPL Mar 21 options chain (Polygon.io):
  • P(AAPL > $200 tomorrow) = N(d2) with IV at $200 strike = 32% vol → P = 34.5%
  • P(AAPL > $205 tomorrow) = N(d2) with IV at $205 strike = 35% vol → P = 21.8%
  • P(AAPL > $210 tomorrow) = N(d2) with IV at $210 strike = 38% vol → P = 12.1%
• Kalshi ladder:
  • "AAPL above $200" at 28c (implied 28%). Fair: 34.5%. Edge = 6.5c (5c net).
  • "AAPL above $205" at 15c (implied 15%). Fair: 21.8%. Edge = 6.8c (5.3c net).
  • "AAPL above $210" at 6c (implied 6%). Fair: 12.1%. Edge = 6.1c (4.6c net).
• Polymarket ladder (always check alongside Kalshi):
  • "AAPL above $200" at 25c (implied 25%). Fair: 34.5%. Edge = 9.5c (9c net) — better than Kalshi here.
  • "AAPL above $205" at 13c (implied 13%). Fair: 21.8%. Edge = 8.8c (8.3c net).
• Best trade: Buy YES on "AAPL above $205" on Polymarket at 13c (better price than Kalshi at 15c). Edge is 8.8c, risk is 13c, potential payout is 87c. If Polymarket has thin liquidity, fill remainder on Kalshi.

Example 2: CPI Data Release — Tail Threshold

• Setup: March CPI releasing tomorrow at 8:30 AM ET.
• Consensus: 3.3% headline CPI (Investing.com median of 55 forecasters).
• Cleveland Fed Nowcast: 3.4% (running slightly hot due to energy prices).
• Historical CPI forecast error SD: 0.12%
• Adjusted mean: 0.6 * 3.4% + 0.4 * 3.3% = 3.36% (weighted toward nowcast)
• Kalshi ladder rung: "CPI above 3.6%" at 8 cents (implied 8%)
• Fair probability: P(Normal(3.36, 0.12) > 3.6) = P(Z > (3.6-3.36)/0.12) = P(Z > 2.0) = 2.3%
• Raw edge: 8% - 2.3% = 5.7 cents
• Net edge: 5.7 - 1.5 = 4.2 cents (Kalshi)
• Polymarket check: "CPI above 3.6%" at 9 cents (implied 9%). Fair = 2.3%. Raw edge = 6.7c. Net edge = 6.2c (Polymarket has lower fees). Trade Polymarket first.
• Trade: Buy NO on "CPI above 3.6%" — Polymarket first, Kalshi for remainder. Fair value of NO = 97.7c. Net edge 4-6c depending on platform.
• Why tail: The 3.6% threshold is 2 standard deviations above the adjusted mean. Prediction market participants overestimate the probability of extreme CPI prints because recency bias (they remember the 2022 inflation spike).

Example 3: S&P 500 Weekly Close — Credit-Equity Divergence

• Setup: HY credit spreads have widened 60bps over 2 weeks (from 340 to 400bps). SPY is flat.
• Historical pattern: When HY widens >50bps while SPY is flat, SPY declines 3-5% within 3 weeks (70% of the time).
• SPY current price: $520. Expected move if divergence resolves: -3.5% = $502.
• SPY options chain: Using the skew-adjusted IV surface (elevated skew confirms market makers are pricing downside risk):
  • P(SPY > $520 in 1 week) = 48% (market is pricing roughly 50/50 but skew says downside is more likely)
  • P(SPY > $510 in 1 week) = 62%
  • P(SPY > $530 in 1 week) = 35%
• Kalshi rung: "S&P above 5200" (1-week) at 55c (implied 55%). Fair value with credit divergence overlay: 42%. Edge = 13c (11.5c net).
• Polymarket rung: "S&P above 5200" (1-week) at 53c (implied 53%). Fair value: 42%. Edge = 11c (10.5c net).
• Trade: Buy NO on "S&P above 5200" (1-week) on both Kalshi and Polymarket, splitting position based on available liquidity.
• Confirmation: VIX term structure is inverted (front month > second month) — market makers are paying up for near-term protection. Consistent with the credit signal.

Example 4: TSLA 1-Day Volatility Contract [PANEL: GPT-4.1 / Gemini Pro]

• Setup: TSLA reports earnings tonight. Options imply a +/-9% move. Historical earnings move SD = 8%.
• Polymarket ladder (if listed): "TSLA moves >5%": 44c. "TSLA moves >7%": 25c. "TSLA moves >9%": 9c.
• Fair probability at >9%: Log-normal model anchored to 9% IV → P(|move| > 9%) = 18% (two-tailed).
• Polymarket price: 9c (implied 9%). Edge = 9c (8.5c net).
• Trade: Buy YES on "TSLA moves >9%" at 9c. The options market is a sharper reference than the prediction market here — the prediction market is substantially underpricing the tails.

IMPORTANT NOTES

• The Stocks Desk has the widest range of prediction market contracts to scan (stock prices, economic data, Fed decisions, earnings outcomes, volatility). Prioritize by liquidity — economic data contracts typically have the deepest books on both Kalshi and Polymarket.
• Kalshi and Polymarket are co-equal platforms for this desk. Both have full ladder structures at multiple price thresholds. Always check both. Trade the better price. When liquidity allows, split positions across both to diversify platform risk.
• Options-implied probabilities are the gold standard for pricing stock prediction market contracts. Without Polygon.io ($29/mo) or free CBOE data access, this desk's edge calculations on stock price contracts are significantly degraded.
• Economic data edges have a unique advantage: settlement time is KNOWN (8:30 AM ET for CPI/NFP, etc.). This means you know exactly when the edge will be resolved, enabling precise position timing.
• Stock/options broker is TBD. Do not hardcode any specific broker in outputs or code. All trade descriptions are broker-agnostic — describe what to buy/sell (e.g., "buy 100 shares of AAPL at market"), not which platform to use.
• Cross-desk with Futures: share the FRED macro data infrastructure, CFTC COT data, and the economic calendar.
• Cross-desk with Options: share the Polygon.io subscription, IV surface construction, and Breeden-Litzenberger methodology.
• The variance risk premium (VIX - RV) is a meta-signal that tells you whether ALL prediction market prices are too high or too low. When VRP is extremely high, the entire tail end of every ladder on both Kalshi and Polymarket is overpriced.
• ETF arbitrage and volatility crush trades (sell straddles/condors when IV is elevated) have no prediction market mapping — route exclusively to the broker.
• Tokenized stock arbitrage is rare and short-lived. Only pursue when the spread is >1% and sufficient liquidity exists on both sides to close the position.