A daily crypto, finance, and AI vocabulary grouping puzzle for beginners.
Tokenized Collateral Yield
These concepts describe how tokenized assets can be tracked, constrained, or used as collateral without recommending any strategy.
Basis TradeCollateral YieldDuration RiskRedemption Gate
- Basis Trade: A basis trade is a position built around the difference between related spot and derivative prices or between similar instruments.
- Collateral Yield: Collateral yield is income or return associated with assets that are also used to support a financial obligation or on-chain position.
- Duration Risk: Duration risk is the sensitivity of a fixed-income asset's value to changes in interest rates over time.
- Redemption Gate: A redemption gate is a rule or mechanism that temporarily limits withdrawals or redemptions under specified conditions.
Derivatives Risk Systems
These terms explain risk-management mechanics in margin and perpetual futures systems at a high level.
- Cross Margin: Cross margin lets multiple positions draw on a shared collateral balance, which can change how gains, losses, and liquidation risk interact.
- Risk Backstop: A risk backstop is a reserve, rule, or participant group designed to absorb losses when normal risk controls are not enough.
- ADL Queue: An ADL queue ranks positions that may be automatically reduced when a venue needs to contain losses after liquidations.
- Maintenance Margin: Maintenance margin is the minimum collateral level required to keep a leveraged position open.
Agent Control Layer
These concepts describe how agent applications constrain behavior, choose actions, and test outputs before production use.
Function RoutingAgent PolicyBehavior BenchmarkPrompt Budget
- Function Routing: Function routing is the process an AI application uses to decide which tool, API, or structured function should handle a request.
- Agent Policy: An agent policy is a set of rules that guides what an AI agent may do, what it should avoid, and when it should ask for confirmation.
- Behavior Benchmark: A behavior benchmark is an evaluation that checks how consistently a model or agent follows expected conduct across tasks.
- Prompt Budget: A prompt budget is the available space or cost allowance for instructions, retrieved context, user input, and generated output.
AI Compute Settlement
These ideas connect AI infrastructure with workload accounting, availability, and verifiable service records.
Model ServingBatch InferenceAvailability LayerAttestation Log
- Model Serving: Model serving is the infrastructure process of hosting models so applications can send requests and receive predictions or generated outputs.
- Batch Inference: Batch inference groups many model requests for scheduled or bulk processing instead of serving each request immediately.
- Availability Layer: An availability layer provides redundancy, routing, or monitoring so a compute or data service remains reachable when demand or failures change.
- Attestation Log: An attestation log is a record of signed or verifiable claims about events such as workload execution, data handling, or runtime identity.