signal evaluation

Transfer Coefficient (TC)

The correlation between unconstrained optimal weights and actual implemented weights; measures how fully a signal is expressed in the portfolio.

The Transfer Coefficient (TC), introduced by Clarke, de Silva, and Thorley (2002), measures the fraction of a signal's predictive power that survives portfolio construction constraints and is actually expressed in portfolio weights:

TC = corr(w_optimal, w_actual)

where w_optimal are the unconstrained mean-variance optimal weights (derived directly from the signal) and w_actual are the weights actually held in the portfolio after applying real-world constraints.

TC enters the Fundamental Law as a direct multiplier on IC: IR ≈ IC × TC × √N. A TC of 1.0 means all signal information is expressed; constraints drag it below 1.0, reducing the effective IR.

Common sources of TC degradation

  • Long-only constraint — inability to short forces the portfolio away from the optimal short positions; TC for long-only strategies typically falls to 0.5–0.7.
  • Position size limits — maximum allocation constraints prevent fully expressing high-conviction signals.
  • Turnover constraints — limiting trading to control transaction costs prevents full rebalancing to the current optimal weights.
  • Benchmark tracking limits — active weight constraints within a mandate restrict deviation from market-cap weights.

Related terms

Related articles