An educational crypto, finance, and AI vocabulary puzzle about execution mechanics, fixed income basics, agent context, and model throughput.
Order Flow
These terms cover liquidity-taking behavior, passive quoting, spread width, and price movement from execution.
- Maker: A maker is an order that adds liquidity by sitting on the book instead of crossing it immediately.
- Taker: A taker is an order that removes liquidity by matching against existing book quotes right away.
- Spread: The spread is the difference between the best available buy price and sell price.
- Slippage: Slippage is the difference between the expected price of a trade and the price that actually fills.
Yield Math
These terms cover coupon payments, present value timing, cash return, and rate sensitivity.
CouponDurationPrincipalYield
- Coupon: A coupon is the regular interest payment attached to a bond or similar debt instrument.
- Duration: Duration is a measure of how sensitive a bond's price is to changes in interest rates.
- Principal: Principal is the original amount of money invested or borrowed before interest is added.
- Yield: Yield is the return earned on an investment, usually expressed as an annual rate.
Agent State
These terms describe the instruction, running context, stored recall, and working notes an agent uses.
- Prompt: A prompt is the text given to a model or agent to define the task it should perform.
- Context: Context is the information a model can consider while generating its next response.
- Memory: Memory is retained information an agent can reuse across steps or conversations.
- Scratchpad: A scratchpad is a temporary space for intermediate reasoning or planning notes.
Serving Speed
These terms describe grouping requests, how long users wait, output rate, and model compression.
- Batching: Batching is the practice of combining multiple requests to improve throughput or efficiency.
- Latency: Latency is the time delay between sending a request and receiving a response.
- Throughput: Throughput is the amount of work a system can complete in a given amount of time.
- Quantization: Quantization is a model compression technique that uses lower-precision numbers to save compute and memory.