AI Trading Glossary

The use of computer programs and mathematical models to execute trades at high speed and frequency based on pre-defined criteria such as price, volume, timing, or technical indicators. Algorithmic trading eliminates human emotion from execution decisions and enables strategies that operate 24/7 across multiple asset classes simultaneously.

A trading strategy that exploits price discrepancies of the same or equivalent asset across different markets, exchanges, or instruments. AI-powered arbitrage bots identify and execute these opportunities in milliseconds, capturing risk-free or near-risk-free profit before the market corrects the pricing inefficiency.

The process of evaluating a trading strategy by applying it to historical market data to assess how it would have performed in the past. Rigorous backtesting accounts for slippage, commissions, and survivorship bias, and is a fundamental step before deploying any AI trading model in live markets.

The difference between the highest price a buyer is willing to pay (bid) and the lowest price a seller is willing to accept (ask). Tighter spreads indicate higher liquidity. AI trading systems continuously monitor spreads to optimize order routing and minimize transaction costs across all traded instruments.

A subset of machine learning that uses multi-layered artificial neural networks to model complex, non-linear patterns in large datasets. In trading, deep learning models such as LSTM networks and Transformers are applied to price time series, order book data, and news feeds to generate predictive trading signals.

The peak-to-trough decline of an investment or trading account over a specified period, expressed as a percentage. Maximum drawdown is a key risk metric used to evaluate the worst-case loss scenario of a strategy. AI risk engines monitor drawdown in real time and can automatically reduce position sizes or halt trading when thresholds are breached.

A specialized algorithm designed to carry out large orders with minimal market impact by intelligently breaking them into smaller child orders. Common types include TWAP (Time-Weighted Average Price) and VWAP (Volume-Weighted Average Price). These algorithms are critical for institutional traders managing significant position sizes.

The process of selecting, transforming, and constructing relevant input variables (features) from raw market data to improve the predictive power of a machine learning model. Effective feature engineering in trading may include technical indicators, sentiment scores, macroeconomic data, and cross-asset correlations.

A form of algorithmic trading characterized by extremely high execution speeds, large order volumes, and very short holding periods — often milliseconds or microseconds. HFT firms use co-location services to place their servers physically close to exchange matching engines, reducing latency to gain competitive execution advantages.

The time delay between sending a trading order and its execution on the exchange. In algorithmic and high-frequency trading, lower latency translates directly to competitive advantage. Senvix's infrastructure is optimized for low-latency order routing, ensuring your strategies operate with minimal delay.

The use of borrowed capital to increase the potential return of a trading position. Expressed as a ratio (e.g., 1:100), leverage amplifies both gains and losses. AI-driven risk systems dynamically adjust allowable leverage based on market volatility, account equity, and open exposure to protect traders from excessive drawdowns.

A measure of how easily an asset can be bought or sold in the market without significantly affecting its price. Highly liquid markets, such as major forex pairs, have deep order books and tight spreads. AI trading models incorporate liquidity metrics to assess execution risk and size positions accordingly.

A branch of artificial intelligence in which algorithms learn from data to make predictions or decisions without being explicitly programmed. In financial markets, ML models are trained on historical price data, fundamentals, and alternative data to identify patterns and generate actionable trading signals.

The study of how markets operate at a granular level, including the mechanics of order matching, price formation, and the behavior of market participants. Understanding microstructure is essential for designing execution algorithms, managing slippage, and building effective AI models that operate in live order books.

A trading strategy based on the statistical tendency of asset prices to revert to their long-term historical average after significant deviations. AI models identify overbought or oversold conditions and generate contrarian signals, taking advantage of temporary price dislocations while managing the risk of trending markets.

A strategy that seeks to capitalize on the continuation of existing market trends by buying assets that are rising and selling those that are falling. AI momentum models process cross-asset signals, volume data, and sentiment indicators to identify high-probability trend continuation setups across multiple timeframes.

A field of AI that enables computers to understand, interpret, and generate human language. In trading, NLP models analyze financial news, earnings call transcripts, central bank communications, and social media to extract sentiment and detect market-moving information faster than human traders can read it.

A real-time electronic record of all outstanding buy and sell orders for a specific asset at an exchange. The order book reveals the depth of market liquidity at different price levels. AI systems perform order book analysis to predict short-term price movements and optimize order placement strategies.

A modeling error where a machine learning algorithm is trained too closely to historical data, capturing noise rather than genuine patterns. An overfit trading model performs well in backtesting but fails in live markets. Techniques such as cross-validation, regularization, and walk-forward testing are used to detect and prevent overfitting.

The process of selecting the best combination of assets and their weightings to maximize expected return for a given level of risk. Modern AI-driven portfolio optimizers go beyond classical mean-variance analysis, incorporating factor models, tail-risk measures, and dynamic rebalancing to maintain optimal risk-adjusted allocations.

The use of statistical algorithms, machine learning, and AI techniques to forecast future market prices, trends, or volatility based on historical and real-time data. Predictive models power the AI estimate feature on Senvix, providing traders with probability-weighted forward-looking insights for informed decision-making.

A systematic trading approach that relies on mathematical and statistical models to identify and execute trading opportunities. Quantitative strategies range from simple factor models to complex multi-strategy AI systems. All positions are sized and managed according to rigorous quantitative risk frameworks rather than subjective judgment.

A type of machine learning where an agent learns to make optimal decisions by interacting with an environment and receiving reward or penalty signals. Applied to trading, RL agents learn to buy, sell, or hold assets to maximize cumulative profit while managing risk, adapting their strategies as market conditions evolve.

A measure of investment performance that accounts for the amount of risk taken to achieve a given return. Common metrics include the Sharpe Ratio, Sortino Ratio, and Calmar Ratio. AI trading systems optimize strategies not just for raw profitability but for maximizing risk-adjusted returns over time.

The use of NLP and machine learning to identify and quantify the emotional tone of textual data — positive, negative, or neutral — from news articles, social media, forums, and financial reports. Sentiment scores are used as input features in AI trading models to gauge market psychology and anticipate price movements.

A widely used performance metric that measures the average return earned in excess of a risk-free rate per unit of volatility. A higher Sharpe Ratio indicates better risk-adjusted performance. AI trading systems on Senvix are continuously optimized to improve Sharpe Ratios by reducing unnecessary volatility while maintaining strong returns.

The difference between the expected execution price of a trade and the actual price at which it is filled. Slippage occurs due to market movement during order transmission, low liquidity, or large order sizes. AI execution algorithms minimize slippage by intelligently timing and sizing orders relative to available market depth.

The process of forecasting the degree of price variation of an asset over a given period. Models such as GARCH, EGARCH, and AI-based neural network approaches are used to estimate future volatility, which is critical for option pricing, position sizing, and risk management in automated trading systems.

A benchmark price calculated by dividing the total dollar value of all trades by total trade volume over a given period. VWAP is used by institutional traders as a performance benchmark and forms the basis of VWAP execution algorithms, which aim to trade at or better than this average to minimize market impact.

A rigorous method of strategy validation where a model is trained on one segment of historical data, tested on the subsequent out-of-sample period, then re-trained and re-tested repeatedly in rolling windows. Walk-forward testing provides a realistic simulation of live performance and helps guard against overfitting in AI trading models.