A bundle—basket, combo, parlay-style package—trades several legs as one economic unit: all must win for an AND payoff, or a defined partial table, or a minted on-chain complete set. The combination-tree chapter mapped the graph; bundles are pre-cut paths through it. Marginals on screen are not enough; joint probability governs profit and loss.
Bundle types in practice
An AND basket pays only if every leg resolves YES—election-night “sweep” packages are the familiar shape. Pair trades express relative value between legs without a listed combo. Minted complete sets on decentralized categoricals let you lock collateral and sell the expensive leg—economically a basket with atomic mint.
Not a bundle: three manual tickets filled minutes apart. You bear leg risk, non-atomic slippage, and one leg moving before the others fill.
Independence is the marketing assumption, not the truth
Multiplying marginals—60% × 50% × 40% = 12%—assumes independence. Shared party shocks, Fed paths, and storm fronts push positive correlation: fair joint probability is often higher than the product. Fréchet bounds give sanity rails: joint P(A and B) lies between max(0, P(A)+P(B)−1) and min(P(A), P(B)).
A listed bundle at 30% with Pennsylvania 52% and Michigan 51% might look cheap versus 26.5% product—that gap can be correlation, not edge. Your joint model must beat the bundle price, not the calculator.
Sweep example in prose
Election night AND on “wins PA” and “wins MI” at 52% and 51% with bundle YES 30%: independence gives 26.5%, Fréchet max joint is 51%, min is 3%. 30% is plausible under positive correlation—not automatic arb. If your joint fair is 22%, sell the bundle; if 38%, buy.
Atomic vs stitched execution
Atomic fill—all legs or none—lowers leg slip and makes arb real. Stitched legs across minutes invite broken baskets when one market jumps. Fee stacks differ: three legs at one cent each way may dominate a two-cent combo ticket; model stitch risk if no listed combo exists.
Regulated venues may require discrete orders per leg unless a listed combo exists; crypto pools may route bundle economics through one swap—know which you have.
Bundles vs trees
Trees expose nodes for per-leg trading and graph inequalities. Bundles collapse the path to one SKU with one pool or one disclosure page. Prices should still cohere with legs when rules match; gaps after fees invite sell-bundle/buy-legs tactics only if you can short or buy NO on each component.
Void and dispute poison one leg
Combo rules may void the entire ticket if one leg is ambiguous—props and scalars are frequent troublemakers. Invalid parent markets on decentralized designs can strand redemption. Selection—avoiding ambiguous multi-leg tickets—often beats bond-heavy litigation.
Hedging with baskets
Long national YES against a weighted basket of swing-state YES legs is a custom hedge, not a listed product. Buy the AND bundle when your joint fair exceeds the market but sits below the Fréchet max; sell when the bundle embeds independence pricing in a correlated world.
Scalar and macro bundles
Macro baskets might combine CPI brackets with rate thresholds. Read whether payout is all-or-nothing or averaged across legs—AND versus partial settlement changes expected value entirely.
Kelly on bundles
Treat a three-leg bundle as one bet for sizing. Triple Kelly on correlated legs is how accounts die in one shared shock.
Fee arithmetic on three legs
Three legs at 1¢ fee per side round-trip may cost 6¢+ versus a 2¢ combo ticket—arb between bundle and legs must beat fee drag. If the platform does not list a combo, model stitch risk explicitly.
Complete sets on decentralized categoricals
Minting a full set of outcome tokens locks collateral; selling the expensive leg expresses a view while holding the rest. Dispute on the parent can strand merge—tie bundle risk to dispute chapters.
Selling rich bundles
When the market prices joint above your fair ρ, selling the bundle or buying NO-equivalents on legs may be the expression—if shorting exists.
Narrative parlays versus listed combos
A story that “all five props hit” is a joint thesis. The listed AND price embeds correlation; your job is to beat that price, not the product of mids on a napkin.
Capital efficiency angle
One bundle ticket may tie less operational attention than five legs—but only if atomicity holds. Stitched legs multiply slippage and dispute tails. Choose the expression with the lowest total friction for your thesis.
Core concepts to remember
Bundles price joints, not products of marginals. Use Fréchet bounds. Prefer atomic fills. Model void on any leg. Size the bundle as one bet.
When the bundle is fair but legs look wrong
Sometimes the bundle price is coherent while one leg looks rich in isolation—correlation explains the wedge. Decompose only when you have a joint model; otherwise you are trading a story against a package price.
Common mistakes
Assuming independence in elections or Fed paths. Ignoring void on one leg. Chasing listed “discount” without ρ and fees. Equating pool shares with regulated combos without reading payoff tables.
What comes next
Bundles are promises to pay at a clock event. Expiration and settlement separate trading close, determination, and cash in your wallet.
Next: Expiration and Settlement Mechanics