Glossary
Exponential Moving Average
A statistical method that weights recent data points more heavily than older ones, commonly used to smooth agent performance metrics over time.
What is Exponential Moving Average?
Exponential moving averages (EMA) apply a decay factor to historical data, making recent performance more influential in current metrics while still considering longer-term trends. This approach balances responsiveness to real changes against stability in the face of noise. For trust scoring, EMA ensures recent agent behavior matters more than distant history, aligning scores with current capabilities.
The decay rate (often expressed as a half-life) controls how quickly old data loses influence. Short half-lives make scores highly responsive but potentially volatile, while long half-lives provide stability but lag behind real changes. Many systems use different EMAs for different metrics: fast-decaying for critical failures, slower for routine performance.
Example
An agent's reliability score uses a 30-day EMA. After two years of perfect performance, a week-long outage significantly impacts the current score. However, as performance recovers, the score rebounds within weeks rather than requiring months to rebuild, since recent uptime quickly outweighs the temporary issue.
How Signet addresses this
Signet uses exponential moving averages to weight recent performance more heavily in trust scores. This ensures scores reflect current agent reliability and quality rather than being dominated by distant history, while maintaining enough stability to prevent gaming through short-term manipulation.
Build trust into your agents
Register your agents with Signet to receive a permanent identity and trust score.