Reference
Glossary
Shared meanings for source, execution, experiments, and runtime evidence.
Configuration and runtime
| Term | Meaning | Guide |
|---|---|---|
| App | Saved strategy source, instrument, models, and settings | Apps |
| Stage session | In-memory instance processing observations and execution state | Engine |
| Server / sandbox | Hosting run with its own identity and stored results | Launching |
| Frame / bar | Observation supplied to a strategy invocation; not necessarily a closed live candle | Bars and intervals |
| Warmup | Preceding observations used to initialize state; live preload differs from a memory run's selected window | Warmup |
| Readiness | Evidence that the required inputs and initialization are usable for a decision | Readiness |
| SJS | Main JavaScript strategy source receiving the ordinary L2 frame scope | SJS |
| SQX | Line-based SQL-inspired declarations and conditions compiled into models | SQX |
| L1 primitive | Configured callback that can perform an action or update state | Execution primitives |
| L2 calculation/rule | Registered calculation or configured rule before the applicable L1 branch | Algorithms |
| L3 | Optional advanced frame hook and execution-intent review source | L3 |
| Policy | Module wired into an L3 policy stack; distinct from an SJS helper | Policies |
Requests and outcomes
| Term | Meaning | Guide |
|---|---|---|
| Signal | Evidence or a decision produced by a calculation; not itself a fill | Combining signals |
| Intent | A proposed action or an execution request being reviewed; inspect the specific policy/native/engine contract | Policy intents |
| Order request | Requested side, quantity, type, and other execution parameters | Placing orders |
| Accepted | A request outcome according to the API/adapter; not universally a confirmed fill | Pending orders |
| Fill | Execution of an order, under the simulator or broker's rules | Simulation versus broker |
| Position | Observed exposure after execution; read it rather than guessing from a submitted request | Positions |
| Trade record | Recorded execution information; a row need not equal a completed round trip | Reading results |
| Automation mode | In-memory control on new bot requests, not automatic cancellation or flattening | Automation modes |
| Source | Request provenance such as SJS, L3, or manual; affects some control paths | Execution flow |
Experiment records
A draft can contain unsaved editor changes. A revision is a saved source/settings snapshot. A run executes a specific revision. A baseline is a selected run for comparison. Publication saves a selected revision into an app and does not launch it.
Read Lab revisions, comparison, and publication. Stored server results are available historical snapshots and artifacts, not a fresh broker-account query.
Continue with the capability matrix or reference guide.