# Arbitrage Scanner > Detect and evaluate arbitrage opportunities across sportsbooks and prediction markets. Calculate guaranteed-profit scenarios, middle bets, and cross-platform price discrepancies. Use when comparing odds across books, calculating arb percentages, evaluating middle opportunities, or building odds-comparison workflows. Also trigger for 'arb bet', 'sure bet', 'arbitrage', 'odds comparison', 'middle bet', 'risk-free bet', or 'line shopping'. - Skill: `michaelschecht/arbitrage-scanner` (Agent Skill) - Install (CLI): `npx skillmds@latest add michaelschecht/arbitrage-scanner` - Raw SKILL.md: https://api.skillmd.com/api/skills/michaelschecht/arbitrage-scanner/raw - Safety review: PASS (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: michaelschecht (https://skillmd.com/u/michaelschecht) - Updated: 2026-08-19 - Page: https://skillmd.com/skills/michaelschecht/arbitrage-scanner --- # Arbitrage Scanner — Cross-Market Profit Opportunities Identify, calculate, and evaluate arbitrage opportunities where odds discrepancies across platforms guarantee profit regardless of outcome. ## When to Use This Skill **USE when:** - Comparing odds across multiple sportsbooks / prediction markets - Calculating whether an arbitrage exists and its profit margin - Evaluating middle bet opportunities - Building a systematic line-shopping workflow - Checking if a position can be hedged for guaranteed profit **DON'T USE when:** - Learning bet types → use `/sports-betting` - Building predictive models → use `/odds-modeling` - Sizing value bets (not arbs) → use `market-mechanics-betting` ## What Is Arbitrage? ``` ARBITRAGE: When odds across different platforms imply probabilities that sum to LESS than 100%, guaranteeing profit regardless of outcome. Example — Two-outcome event: Sportsbook A: Team X at +150 (implied 40.0%) Sportsbook B: Team Y at +130 (implied 43.5%) Total implied: 40.0% + 43.5% = 83.5% ← UNDER 100% = ARB EXISTS Arb margin: 100% - 83.5% = 16.5% guaranteed profit Typical reality: Most arbs are 1-5% margin (not 16.5%) After vig removal, many apparent arbs vanish Execution risk (odds change, limits, account restrictions) ``` ## Step 1: Arb Detection ### Two-Outcome Arb Formula ``` Given: Odds A (decimal) = best odds on Outcome 1 (from any book) Odds B (decimal) = best odds on Outcome 2 (from any book) Arb exists if: (1/Odds_A) + (1/Odds_B) < 1 Arb percentage = (1 - ((1/Odds_A) + (1/Odds_B))) × 100 Example: Best Outcome 1: 2.50 (decimal) at BookA Best Outcome 2: 1.80 (decimal) at BookB Check: (1/2.50) + (1/1.80) = 0.400 + 0.556 = 0.956 0.956 < 1.0 → ARB EXISTS Arb %: (1 - 0.956) × 100 = 4.4% guaranteed profit ``` ### Three-Outcome Arb (Soccer, etc.) ``` Odds A = best Home win odds (any book) Odds B = best Draw odds (any book) Odds C = best Away win odds (any book) Arb exists if: (1/A) + (1/B) + (1/C) < 1 Example: Home: 3.20 (BookA) | Draw: 3.60 (BookB) | Away: 2.40 (BookC) Check: 0.3125 + 0.2778 + 0.4167 = 1.007 1.007 > 1.0 → NO ARB (but close — monitor for movement) ``` ### Stake Calculation ```python def calculate_arb_stakes(total_stake, odds_a, odds_b): """Calculate how much to bet on each outcome for equal profit""" inv_a = 1 / odds_a inv_b = 1 / odds_b total_inv = inv_a + inv_b if total_inv >= 1: return None # No arb exists stake_a = total_stake * (inv_a / total_inv) stake_b = total_stake * (inv_b / total_inv) profit_if_a = (stake_a * odds_a) - total_stake profit_if_b = (stake_b * odds_b) - total_stake return { 'stake_a': round(stake_a, 2), 'stake_b': round(stake_b, 2), 'profit_if_a': round(profit_if_a, 2), 'profit_if_b': round(profit_if_b, 2), 'arb_pct': round((1 - total_inv) * 100, 2), 'roi': round(min(profit_if_a, profit_if_b) / total_stake * 100, 2) } # Example result = calculate_arb_stakes(1000, 2.50, 1.80) # stake_a: $418.60, stake_b: $581.40 # profit_if_a: $46.51, profit_if_b: $46.51 # arb_pct: 4.40%, roi: 4.65% ``` ## Step 2: Types of Arbitrage ### Pure Arbitrage (Sure Bets) ``` DEFINITION: Guaranteed profit with zero risk MARGIN: Typically 1-4% FREQUENCY: Rare on major markets, more common on: - Cross-sport props - Niche leagues (lower divisions, esports) - Newly posted lines (before market corrects) - Cross-platform (sportsbook vs. exchange) EXECUTION WINDOW: Seconds to minutes → Requires fast multi-account execution → Software tools are almost mandatory for consistent arbing ``` ### Middle Bets ``` DEFINITION: Bet both sides at different spreads — win both if result falls in the "middle" window. Guaranteed small loss if it doesn't. Example: BookA: Team X -3.5 at -110 BookB: Team X +4.5 at -110 (from opponent's perspective) If Team X wins by exactly 4: -3.5 bet wins (they won by more than 3.5) ✓ +4.5 bet wins (opponent lost by less than 4.5) ✓ → WIN BOTH SIDES If Team X wins by 5+: Win -3.5 bet, lose +4.5 bet → small loss If Team X wins by 3 or less: Lose -3.5 bet, win +4.5 bet → small loss COST: ~$9 in vig per $100 staked on each side MIDDLE HITS: ~8-12% of the time on 1-point middles in NFL EV: Positive if middle probability × payout > guaranteed vig cost ``` ### Closing Line Arbitrage ``` DEFINITION: Bet early at one book, then bet the other side at closing when the line has moved in your favor. Step 1: Bet Team A -3 (-110) early in the week Step 2: Wait — line moves to Team A -5 Step 3: Bet Team B +5 (-110) at another book Middle window: Team A wins by 4 → both bets win Worst case: Small vig loss on one side REQUIRES: Ability to identify line movement direction → See /market-sentiment for sharp money indicators ``` ### Cross-Market Arbitrage ``` SPORTSBOOK vs PREDICTION MARKET: Sportsbook: Chiefs to win Super Bowl +500 (16.7% implied) Prediction Market: Chiefs Super Bowl "No" at $0.78 (22% implied "Yes") If the prediction market implies a higher probability of "Yes": Bet "Yes" at the sportsbook (+500) Bet "No" on the prediction market ($0.78) → Different margin structures create opportunities → Prediction markets often have lower vig than sportsbooks SPORTSBOOK vs EXCHANGE: Sportsbook: Team A at 2.10 Exchange: Lay Team A at 2.00 (bet against) Back Team A at 2.10 (book) Lay Team A at 2.00 (exchange) → Guaranteed profit from the odds gap → Exchange commission reduces margin (typically 2-5%) ``` ## Step 3: Line Shopping Workflow ### Without Arbing — Pure Value Capture ``` Even if no arb exists, line shopping saves money: SAME BET, DIFFERENT PRICES: BookA: Team X -3 at -115 BookB: Team X -3 at -110 ← SAVE $5 per $100 BookC: Team X -3 at -105 ← SAVE $10 per $100 IMPACT OVER TIME: 1,000 bets at $100 each: At -115: Need 53.5% win rate to break even At -105: Need 51.2% win rate to break even Difference: 2.3% fewer wins needed = ~23 fewer wins per 1,000 bets → LINE SHOPPING IS THE SINGLE EASIEST EDGE IN SPORTS BETTING ``` ### Multi-Book Strategy ``` TIER 1 — SHARP BOOKS (set the line, high limits): Pinnacle, Circa, Bookmaker → Use for: reading the market, identifying true odds → Rarely limited but vig is low TIER 2 — MAINSTREAM BOOKS (follow the line, medium limits): DraftKings, FanDuel, BetMGM, Caesars → Use for: best promos, boosts, occasional soft lines → Will limit winning accounts over time TIER 3 — SOFT BOOKS (slow to adjust, low limits): Smaller regional books, offshore books → Use for: arb opportunities, stale lines → Limit quickly, but softest odds STRATEGY: Track lines at Tier 1, bet at Tier 2/3 when discrepancy exists ``` ## Step 4: Risk & Limitations ### Why Arbs Fail | Risk | Description | Mitigation | |------|-------------|------------| | **Odds change** | One side moves before you place the second bet | Bet simultaneously; don't arb margins <1.5% | | **Account limits** | Sportsbooks limit or ban arb bettors | Round bet amounts, vary patterns, mix in recreational bets | | **Max bet limits** | Can't place full arb stake at one book | Check max before committing to the other side | | **Voided bets** | Book cancels one side (palpable error rule) | Never arb obviously wrong lines (likely to be voided) | | **Settlement differences** | Books settle the same event differently (OT rules, etc.) | Verify settlement rules match before arbing | | **Correlation** | Two "independent" outcomes are actually correlated | Verify true independence of arb legs | | **Capital lock** | Money tied up in futures arbs for months | Accounts for opportunity cost in ROI calculation | ### Account Longevity Tips ``` STAY UNDER THE RADAR: ✓ Round bets to natural amounts ($50, $100 — not $47.83) ✓ Mix arb bets with recreational bets ✓ Don't arb every market — be selective ✓ Deposit and withdraw at normal intervals ✓ Take promotions / boosts (looks recreational) ✗ Don't bet exclusively on obscure markets ✗ Don't max-bet every arb at the exact limit ✗ Don't withdraw immediately after every win REALITY: On soft books, profitable accounts get limited eventually → Diversify across many books → Focus on exchanges (Betfair, Smarkets) for unlimited arbing ``` ## Step 5: Arb Tracking Spreadsheet ```markdown | Date | Event | Book A | Odds A | Stake A | Book B | Odds B | Stake B | Total Stake | Arb % | Result | Profit | |------|-------|--------|--------|---------|--------|--------|---------|-------------|-------|--------|--------| | 3/30 | NYK vs LAL | DK | +145 | $420 | FD | -135 | $580 | $1000 | 2.1% | NYK W | $21.00 | MONTHLY SUMMARY: Total arbs placed: [count] Total capital deployed: [sum of stakes] Total profit: [sum of profits] Average arb margin: [avg %] Failed arbs (odds moved): [count] Capital utilization rate: [active capital / total bankroll] ``` ## Output Save arb analysis to `artifacts/prediction-markets/arb-[event/date].md` ## Guidelines - **Line shopping is mandatory** even if you never arb — always get the best price - Arbs under 1.5% aren't worth the execution risk for manual bettors - The best arb is the one where you also have a value opinion on one side - Prediction markets (Kalshi, Polymarket) vs. sportsbooks are an underexplored arb surface - Soft books will limit you — budget for account attrition as a cost of doing business - Use `/sports-betting` to understand the markets you're arbing - Use `market-mechanics-betting` for optimal stake calculation - Use `/market-sentiment` to predict which side of an arb will move (bet the moving side last)