Success metrics
The basket of metrics the lab evaluates — because no single number is the objective.
The platform does not optimize a single metric. It continuously evaluates a basket of them — because maximizing any one number in isolation usually produces a worse system overall.
What the lab tracks
| Metric | What it tells us |
|---|---|
| Expected return | The edge a decision is expected to capture |
| Win rate | How often a decision is right — but not the goal in itself |
| Average reward / risk ratio | How much is won per unit of risk taken |
| Capital growth | The compounding outcome over time |
| Maximum drawdown | The worst peak-to-trough loss endured |
| Holding duration | How long capital stays committed per opportunity |
| Execution quality | Slippage and fees versus the intended fill |
Win rate is not always the right objective
The platform may discover that maximizing win rate is not optimal — a strategy that wins rarely but with a large reward/risk ratio can beat one that wins often and small. Every assumption about which metric matters must itself be validated through data.
Why a basket, not a single score
A high win rate with a terrible reward/risk ratio loses money; a great expected return with an unbearable drawdown is untradeable in practice. The metrics are evaluated together so the lab optimizes for a system that is actually survivable, not for a single flattering number.