AI trading signal terms
Plain definitions for the vocabulary running through this reference. Where a term carries a test of its own, its entry links to the page that takes it further.
The terms, defined
AI trading signal
A trade suggestion - ticker, direction, entry, target and stop - produced by an automated or algorithmic process and published in real time, whether the process is a disclosed rule or an undisclosed model.
Algorithmic signal
A signal generated by a fixed, repeatable procedure rather than a discretionary judgement, so the same inputs always yield the same call.
Black box
A service whose decision logic is not disclosed, leaving a buyer unable to judge the method and reliant on trusting the output. See: transparent versus black-box.
Systematic model
A trading model that follows an explicit set of rules with no discretion at the point of decision; the recommended desk runs four of them.
Mean reversion
Trading on a price that has stretched unusually far from a typical level and tends to snap back toward it; the four models apply that idea on different clocks. See: how to verify a record.
Backtest
A simulation of how a strategy would have performed on past data. Useful for design, but fitted with hindsight - it is not a live record and proves nothing about real-time calls. See: backtest versus live record.
Cryptographic receipt
A hash of a signal written to a public ledger at publication, proving the call existed in exactly that form at that moment and was not altered afterward. See: locked before the outcome.
Conviction grade
A tag from A down to D that places a call by how strong it is within its own model's measured spread of returns; the ladder stops at D, with no E grade.
Drawdown
The deepest peak-to-trough fall over a period - a more honest read on risk than the headline return.
Win rate
The share of calls that closed profitably - worth trusting only when the full number of signals sits next to it and nothing that lost has been quietly dropped. See: why a win rate needs its sample size.