Glossary

Key terms used throughout TRADEOS.tech documentation.

Attestation — A daily cryptographic commitment of TRADEOS.tech signal activity to Base L2. The attestor service builds a SHA-256 Merkle tree of all signal outcomes for the previous UTC day and submits the root to the SignalAttestation smart contract on Base L2. See On-Chain Signal Attestation.

Almgren-Chriss model — A mathematical model for estimating the market impact of executing a large order. TRADEOS.tech uses it to determine whether a proposed position size is viable given current market depth.

ATR (Average True Range) — A volatility measure that captures the typical range of price movement over a window of bars. Used by the exit manager to set dynamic stop-loss distances.

Autonomous Agent — The component responsible for system-level optimization. It monitors performance drift, detects miscalibration, and proposes parameter adjustments on hourly and weekly cadences.

Bayesian Change Point Detection (BOCD) — A statistical algorithm that detects abrupt changes in time series data. Used by TRADEOS.tech to identify regime transitions in real time.

Circuit Breaker — An automatic mechanism that halts or constrains trading when predefined risk conditions are met. Activates without human intervention.

Confidence score — A value between 0 and 1 representing a signal generator's estimated probability that a trade in the given direction will be profitable. Higher is more certain.

Dead-man's switch — A safety mechanism that triggers when the trading loop stops publishing heartbeats. Closes open positions and halts new trading automatically.

Drawdown — The percentage decline from the peak portfolio value to the current value. A 10% drawdown means the portfolio is currently worth 10% less than its all-time high.

Fail-closed — A safety design principle where a system defaults to a safe/restrictive state when it cannot determine the correct answer. TRADEOS.tech feasibility is fail-closed.

Feasibility Service — The primary risk enforcement component. Every trade intent must pass feasibility checks before an order can be placed.

Hero signals — The top-performing signal types by information coefficient and win rate. They receive a scoring bonus and are prioritized in the signal pipeline.

HMM (Hidden Markov Model) — A statistical model used to classify market regimes (trending, mean-reverting, volatile) based on observable price and volume data.

IC (Information Coefficient) — The correlation between a signal's predicted direction and the actual subsequent price movement. A positive IC indicates the signal has predictive value.

Intent — A proposed trade that has been approved by the feasibility service and is waiting to be executed. Intents are published to the intents Redis Stream.

Kelly criterion — A formula for computing the optimal fraction of capital to allocate to a bet given known probabilities of winning and losing. TRADEOS.tech uses a fractional Kelly (capped at 0.1-0.5 depending on profile) to manage estimation risk.

Market impact — The price movement caused by executing an order. Larger orders in thinner markets have more impact, reducing the realized value of the trade.

Merkle tree — A binary hash tree where each leaf is the hash of a data element and each internal node is the hash of its two children. Used by the attestor service to commit all daily signals into a single 32-byte root that can be submitted on-chain.

Merkle proof — A list of sibling hashes that allows anyone to re-derive the Merkle root from a single leaf. Given a signal's five fields and its proof path, any party can verify the signal's inclusion in a given day's attestation without accessing the database.

Policy profile — The complete set of risk and execution parameters that governs system behavior (see Strategy Profiles).

Redis Streams — The messaging infrastructure used for all inter-service communication in the trade path. Provides durability, consumer groups, and replay capability.

Regime — A classification of the current market state: trending (directional price movement), mean-reverting (price oscillates around a stable level), volatile (high-amplitude moves without clear direction), or low-volatility (calm, tight-range conditions).

Signal — The output of a signal generator: a directional prediction (long, short, flat) for a specific instrument with an associated confidence score.

Signal generator — A component that analyzes market data and produces trading signals. Examples: momentum, mean-reversion, order flow imbalance, funding rate carry.

TCA (Transaction Cost Analysis) — Post-trade analysis that measures the quality of trade execution and the accuracy of signal predictions.

TWAP (Time-Weighted Average Price) — An execution algorithm that splits a large order into smaller child orders distributed over time to minimize market impact.

VPIN (Volume-Synchronized Probability of Informed Trading) — A microstructure indicator that measures the probability of informed order flow. Elevated VPIN can signal abnormal market conditions and triggers conservative adjustments in TRADEOS.tech.

Win rate — The percentage of completed trades that are profitable. Tracked per signal type by the TCA monitor.