Project Record

Value-add Score

Investor behavior score measuring conviction, PnL quality, and portfolio stability.

portfoliobehaviorscoring

Idea

Evaluate how strongly investors hold positions and whether returns are delivered with disciplined turnover. The score supports IR/BD and portfolio oversight by distinguishing durable alpha from churn and luck.

Context

Investor behavior varies widely: some hold through drawdowns with high conviction; others trade frequently and may show volatile PnL. A single performance number does not capture quality of returns or risk discipline. This metric was designed to rank and segment investors for reporting and engagement.

Problem

We need a scalar that reflects (1) conviction (holding period, consistency of sizing), (2) PnL quality (risk-adjusted returns, drawdown control), and (3) portfolio stability (turnover, concentration). Raw metrics are on different scales and need to be normalized and weighted for interpretability.

Score

V = 0.45C + 0.35P + 0.20S

with conviction $C$ , PnL quality $P$ , and stability $S$ . Each component is normalized to [0,1] using historical or peer benchmarks.

Implementation

Inputs: position history, PnL series, and optional benchmark.
$C$ : derived from average holding period, rebalance frequency, and size stability.
$P$ : Sharpe or similar, plus penalty for drawdown depth and length.
$S$ : inverse of turnover and concentration (e.g. Herfindahl).
Output: score per investor/strategy, with optional drill-down by component for dashboards.

Trade-offs

Weights are heuristic; they can be tuned per strategy or client. Backtesting against known “good” and “bad” behavior helps calibrate. The score is not a substitute for full risk reporting but works well for screening and trend views.