Metrics Definitions

Metric Definitions

This page explains the performance metrics displayed in the AITA interface.

These metrics describe how an agent has behaved historically. They are provided to help users understand strategy behavior, risk characteristics, and consistency over time.

All metrics are descriptive and based on historical data.

Total Return

Total return shows the cumulative performance of an agent over the selected time period.

It represents the combined effect of gains and losses produced by the strategy. Total return does not account for risk, volatility, or drawdowns on its own and should never be interpreted in isolation.

Win Rate

Win rate represents the percentage of trades or periods that resulted in a positive outcome.

A higher win rate does not necessarily mean a better strategy. Some strategies are designed to win frequently with small gains, while others win less often but capture larger moves.

Maximum Drawdown

Maximum drawdown measures the largest peak-to-trough decline in performance over the observed period.

This metric helps users understand the worst historical loss the strategy experienced before recovering. Drawdown size and duration are critical for assessing risk tolerance.

Volatility

Volatility describes how much the agent’s performance fluctuates over time.

Higher volatility indicates larger swings in results, both positive and negative. Lower volatility indicates smoother, more stable performance but may come with lower growth.

Sharpe Ratio

The Sharpe ratio is a risk-adjusted metric that compares returns relative to volatility.

A higher Sharpe ratio suggests that returns were achieved with less variability. It does not guarantee future performance and is sensitive to the time period measured.

Time in Market

Time in market shows the percentage of time the agent was actively exposed to market positions.

Strategies with higher time in market may experience more continuous exposure to risk, while strategies with lower time in market may be more selective.

Number of Trades

This metric indicates how often the strategy generated actionable decisions over the measured period.

A higher number of trades does not imply higher quality. Trade frequency should be interpreted in the context of strategy type and timeframe.

Backtested vs Live Metrics

Some metrics may be shown for backtested periods, live execution, or both.

Backtested metrics simulate historical behavior using past data. Live metrics reflect real execution and may differ due to slippage, fees, liquidity, and capital constraints.

Differences between backtested and live metrics are normal.

Using metrics together

No single metric provides a complete picture.

Metrics should be evaluated together to understand:

• Consistency

• Risk exposure

• Drawdown behavior

• Stability across time

Metrics describe what happened, not what will happen.