Conditional Tokens are a lower-layer contingent-claim settlement system, not a full prediction-market product. The docs separate condition preparation, outcome-subset encoding, position composition, and collateral settlement.
The most reusable mechanism is the four-part decomposition: conditions, outcome collections, collection identifiers, and positions. That is what lets the system express claims on subsets of outcomes and combinations across multiple conditions.
ERC-1155 matters here because positions are treated as transferable inventory slots in a broader claim graph rather than as one-off market-specific ERC-20s.
Oracle trust and question semantics are intentionally pushed outward. The contracts can prepare and resolve conditions, but they do not standardize how the question is authored, interpreted, or disputed.
The system is more general than binary or single-event markets. The docs emphasize scalar conditions, categorical conditions, and deep positions contingent on several conditions at once.
Split, merge, and redeem are the real accounting primitives. They control how collateral moves into root positions, how deeper exposures are composed, and how resolved claims unwind.
Specific source signals worth retaining
developer-guide.rst: the best source for the actual mechanism. It walks through condition preparation, index-set encoding, collection-ID derivation, position IDs, and split / merge / redeem flows in concrete terms.
glossary.rst: useful for keeping the vocabulary crisp, especially the distinctions among condition, outcome slot, position, collection identifier, root position, and non-root position.
Tutorial prediction-market use-case doc: useful because it shows that the contracts were explicitly framed as infrastructure for building prediction-market applications rather than as an end-user product themselves.
Why this entry belongs in the corpus
It provides a canonical missing base layer for prediction-market and contingent-claim infrastructure.
It creates a better comparison baseline for Azuro and future prediction-market entries by separating the settlement primitive from the market-making and distribution layers.
The durable insight is that many prediction market systems are really combinations of several layers: question authoring, oracle / dispute policy, settlement grammar, market making, and frontend distribution. Conditional Tokens isolate the settlement grammar cleanly.