Rubinfeld

Cross-Venue edges you can actually trade (longread) Why this works: @Polymarket still owns the lion's share of volume, but @MyriadMarkets turns over faster and runs thinner books. That combo creates short-lived price gaps on the same events. If rules and windows match, you can capture the spread. If they don't, you can still trade the skew around catalysts. Strategy 1 - same-rule cross-venue basis (Lock the spread) Use when: Two markets ask the same question, use the same resolution source, and close on the same (or near-same) deadline. What to do (checklist): • Open both markets -> read rules line-by-line (event, window, source, cancel conditions) • Confirm they're truly the same bet (identical "Yes" meaning) • If price gap ≥ 3–6 percentage points, enter: • Buy "Yes" where it's cheaper • Buy "No" where "Yes" is more expensive (this replicates short "Yes") • Size to your collateral: you're holding two legs until prices converge or resolve • Exit on convergence (prices meet) or hold to resolution - profit is the initial gap either way if rules match 100 simulation (transparent math): Assume two identical markets show 12.5% vs 8.0% for "Yes" (gap = 4.5 p.p.) Budget $100. Buy the same number of shares S on each side: Total cost per share pair = cheap_yes (0.08) + expensive_no (1 − 0.125 = 0.875) = 0.955 Shares S = 100 / 0.955 ≈ 104.712 If they converge to the midpoint (10.25%): • Long cheap "Yes": gain per share +0.0225 -> +0.0225 × 104.712 = +$2.356 • Long expensive "No": gain per share +0.0225 -> +0.0225 × 104.712 = +$2.356 • Total PnL ≈ +$4.712 (+4.7% on $100) If you hold to resolution (regardless of outcome): net profit = initial gap × S = 0.045 × 104.712 = $4.712 (+4.7%) That's the beauty of true same-rule arb: you've locked the gap, outcome-agnostic. Main risks (how to neutralize): • Not actually the same bet (different windows, sources, or wording) -> your spread is fake. Fix: only trade after rule-by-rule match • Cancellation/edge-case edits -> both legs might not settle as expected. Fix: avoid creator-edited or ambiguous markets • Fee/withdrawal friction -> spreads can vanish in fees. Fix: check current trading/exit fees and chain costs before entering Strategy 2 - News-lag convergence (Catch the slow venue) Use when: The same event is listed on both venues, but one venue lags on a fresh headline (earnings, policy hints, injuries, official announcements). What to do (checklist): • Track an official source (press release, X post, on-chain tx, schedule) • As soon as one venue moves ≥ 2–3 p.p. and the other hasn't, buy the lagging side if you agree with the news • Micro-batch entries (e.g., 4×$25) over 3–10 minutes; cancel remaining clips if the gap closes without you • Exit on catch-up or into the first backfill wick; don't marry it - this is a flow trade, not a thesis $100 simulation (one-leg catch-up): You buy "Yes" at 0.29 on the lagging venue; the other venue already implies 0.32 If your side catches up by +3 p.p. to 0.32: Shares = 100 / 0.29 ≈ 344.83; PnL = 0.03 × 344.83 ≈ $10.35 (+10.3%) Smaller catch-up +2 p.p. -> 0.02 × 344.83 ≈ $6.90 (+6.9%) Whipsaw risk −2 p.p. -> −$6.90 (set a time stop; exit if the "faster" venue reverses) Main risks (how to neutralize): • Headline gets walked back -> fast exit discipline (time stop: minutes, not hours) • Different rules despite similar titles -> still read both rule blocks • Thin book slippage -> micro-batch and avoid chasing >1–2 ticks from mid How to read rules in 60 seconds (so you don't fake-arb) • Outcome definition: The exact thing that triggers "Yes" • Window & deadline: Same time zone? Same last minute? • Resolution source: On-chain? Official announcement? "Consensus of credible reporting"? • Cancellation conditions: If either market can be canceled for reasons the other can't, it's not arb • Wording traps: "Any cut" vs "exactly 25 bps"; "by Sept 30" vs "before Sept 30, 23:59 UTC" What to track (so you keep edges repeatable) • Basis (pA − pB) across venues for identical markets • Depth per clip: how many ticks a $1k–$5k order moves each book • Resolution speed: shorter cycles -> more spreads per week • Fee/withdraw flow: fees change the breakeven; check them before trading Quick formulas (keep handy) Convergence/Resolution PnL (true same-rule arb): Let gap g = p_expensive_yes − p_cheap_yes With S matched shares on both legs, PnL = g × S (whether you close on convergence or hold to resolution) With a $100 budget using the "buy cheap Yes + buy expensive No" construction: S ≈ 100 / (p_cheap_yes + (1 − p_expensive_yes)) One-leg catch-up PnL (news-lag): For entry price p0, move Δp, budget $B, shares S = B / p0, PnL = Δp × S Footnote: NFA. This is research, not advice. If you actually understand risk, you don’t need me to tell you it’s real. If you don’t - stop, DYOR, and only then press buttons.

Gm @Hackatao