Transfer Coefficient Explained: Why a 0.4 TC Destroys More Than Half Your Alpha

A manager can forecast returns brilliantly and still deliver mediocre performance. The transfer coefficient is the concept that explains why, and it is one of the more elegant ideas in the portfolio management curriculum — elegant because it isolates something real that practitioners argue about constantly and puts a number on it.

The fundamental law of active management

Start with the unconstrained form, sometimes called the basic fundamental law:

E(R_A) = IC × √BR × σ_A

Expected active return equals the information coefficient, multiplied by the square root of breadth, multiplied by active risk. In plain terms: how good your forecasts are, how many independent bets you make, and how much risk you are willing to run.

• IC — the information coefficient. The correlation between forecast active returns and realised active returns. This is raw forecasting skill. Real-world ICs are small: 0.05 is respectable, 0.10 is excellent. A manager with an IC of 0.05 is right slightly more often than a coin, and that slight edge is what the whole industry is built on.
• BR — breadth. The number of genuinely independent active decisions per year. Independence is the operative word and the most abused term in the framework.
• σ_A — active risk. The tracking error the manager is permitted to run.

This version assumes the manager can build precisely the portfolio their forecasts imply. Nobody can. Hence the full form:

E(R_A) = TC × IC × √BR × σ_A

What the transfer coefficient measures

The transfer coefficient is the correlation between the manager's risk-weighted forecast active returns and the actual risk-weighted active weights in the portfolio. It measures the fidelity of the translation from view to position.

It runs from 0 to 1:

• TC = 1: the portfolio is exactly what the forecasts imply. An unconstrained ideal that exists only in textbooks.
• TC ≈ 0.8–0.9: a lightly constrained mandate, typically a long-short strategy with few restrictions.
• TC ≈ 0.4–0.6: a typical constrained long-only institutional mandate.
• TC → 0: the portfolio bears no relationship to the manager's views. The constraints are running the strategy.

Because it multiplies through the entire expression, its effect is direct and brutal. A manager with genuine skill and a TC of 0.4 captures 40% of the active return their forecasting ability would otherwise justify. The other 60% was never lost to bad decisions — it was consumed by the mandate.

What pushes the transfer coefficient down

Nearly every real-world portfolio constraint reduces TC.

The long-only constraint

This is by far the largest single factor, and the reason deserves spelling out because exam questions test it directly.

A manager who forecasts that a stock will badly underperform can only express that view by holding none of it. If the stock represents 0.2% of the benchmark, the maximum possible underweight is 0.2% — nowhere near the conviction being expressed. Negative views are systematically truncated, while positive views can be sized freely.

This interacts badly with index concentration. In a benchmark where the top ten names carry 30% of the weight and the bottom five hundred carry a few basis points each, a long-only manager has enormous scope to express negative views on the largest names and almost none on the smallest. Skill at identifying small-cap losers is close to worthless within that structure — the manager can see it and cannot trade it.

Everything else

• Position size limits. A 5% cap per holding blocks the strongest positive views from being sized to conviction.
• Sector and country bands. These constrain groups of positions at once and can force the manager to hold names they have no positive view on, purely to stay inside a band.
• Turnover limits and transaction costs. A forecast that cannot be traded economically never reaches the portfolio. Short-horizon signals suffer most.
• Liquidity. Small-cap conviction that cannot be built at scale is capped by market depth rather than by the mandate.
• Risk model constraints. Factor-neutrality requirements can force offsetting trades that dilute the intended positions.

Worked example

A manager has an information coefficient of 0.06, makes 200 independent forecasts a year, and runs 5% active risk.

Unconstrained: 0.06 × √200 × 5% = 0.06 × 14.14 × 5% = 4.24% expected active return.

With a TC of 0.45 (typical constrained long-only): 0.45 × 4.24% = 1.91%

More than half the expected value of the manager's skill has been consumed before a single trade went wrong.

Note that the ex-ante information ratio scales identically, since IR ≈ TC × IC × √BR. A low transfer coefficient damages risk-adjusted performance, not just absolute active return — the manager is not simply taking less risk for less return, they are being less efficient with the risk they take.

Comparing two managers

Manager X: IC of 0.08, breadth 100, active risk 4%, TC of 0.35 (tightly constrained long-only).
Manager Y: IC of 0.05, breadth 150, active risk 4%, TC of 0.75 (long-short).

X: 0.35 × 0.08 × 10 × 4% = 1.12%
Y: 0.75 × 0.05 × 12.25 × 4% = 1.84%

The less skilled manager delivers more, because their mandate lets them use what skill they have. This is the exam's favourite illustration of the concept, and the point it makes is real: comparing managers on forecasting ability alone, without reference to the constraint set, is comparing the wrong thing.

The implications the exam wants

Relaxing constraints is a source of return. Moving from long-only to a 130/30 structure raises TC by allowing negative views to be sized properly. That is the actual argument for such structures — an argument about the transfer coefficient rather than about leverage, and questions test whether you can state it that way.

Breadth is not the number of holdings. Because breadth enters under a square root, doubling genuinely independent bets increases expected active return by only about 41%. And correlated bets do not count. A manager holding 300 positions all driven by one macro view has breadth closer to 1 than to 300. Nothing in the framework is more routinely overstated in marketing material than breadth.

The law is an ex-ante expectation. It describes expected active return, not realised. Any single year's outcome is dominated by noise. A manager who underperformed for a year has not disproved their IC, and one who outperformed has not proved it — the sample is far too small, which links directly to sample-size neglect in the behavioural material.

Skill without implementation is not skill in practice. Two managers with identical ICs deliver very different results under different mandates. When evaluating managers, the constraint set is part of the evaluation, not context around it.

How it is tested

The transfer coefficient appears in the active portfolio management material and recurs in manager selection and evaluation. Questions take three recognisable forms:

1. Compute expected active return given TC, IC, breadth and active risk. Straight arithmetic — watch that breadth goes under a square root.
2. Identify which constraint would most reduce TC. The answer is almost always the long-only constraint where it appears among the options, for the truncation reason above.
3. Explain why two managers with the same IC produce different results. The answer is some combination of breadth and transfer coefficient — and if any mandate constraint appears anywhere in the vignette, it is TC.

For essay-format answers, name the term, state the direction of the effect, and tie it to the specific constraint in the vignette. Stating that "constraints reduce active return" without identifying TC as the mechanism will not score full marks. The Level 3 essay guide covers how these answers are marked.