Benchmarks & Performance Attribution

An investment process isn't judged in a vacuum — it's judged against a benchmark, and a real allocator wants to know why a manager beat or missed it, not just by how much. This chapter covers how that actually gets measured.

Alpha and beta — separating skill from market exposure

Beta measures how much a portfolio moves relative to its benchmark — a beta of 1.0 means it moves in line with the benchmark; a beta of 1.2 means it tends to move 20% more than the benchmark in either direction. Alpha is the return left over after accounting for what beta alone would predict — the actual, benchmark-adjusted measure of a manager's skill (or lack of it), and the real number a sophisticated allocator cares about far more than raw, unadjusted return.

Worked example: a portfolio has a beta of 1.2 versus its benchmark. The benchmark returns 10% this year. Beta alone would predict a portfolio return of 1.2 × 10% = 12%. The portfolio actually returned 15%. Alpha = actual − beta-predicted = 15% − 12% = +3% — the manager generated 3 percentage points of real, benchmark-adjusted outperformance, not just 5 points of raw outperformance (15% − 10%), which would have overstated the manager's actual skill by not accounting for the extra market exposure taken on.

Tracking error — how far a portfolio strays from its benchmark

Tracking error is the standard deviation of the difference between a portfolio's returns and its benchmark's returns over time — a real measure of how much a manager deviates from the benchmark, independent of whether that deviation has been rewarded. A manager who closely hugs the benchmark has low tracking error (and, almost by construction, limited ability to generate much alpha either way); a highly concentrated, benchmark-agnostic manager has high tracking error, with real potential for both significantly higher and significantly lower returns than the benchmark.

Worked example: two managers both generate 2% of alpha over a year. Manager A did it with 3% tracking error; Manager B did it with 9% tracking error. The information ratio (alpha ÷ tracking error) makes the comparison concrete: Manager A = 2% ÷ 3% = 0.67; Manager B = 2% ÷ 9% = 0.22. Manager A generated the same outperformance while taking on much less deviation risk to do it — a materially better, more efficient result by this measure, even though both managers show identical headline alpha.

Performance attribution — explaining where the return actually came from

Attribution breaks total portfolio performance down into its real components — how much came from sector allocation (over/underweighting entire sectors relative to the benchmark) versus security selection (picking better- or worse-performing individual stocks within each sector). A manager who outperformed purely through a lucky sector bet (being overweight a sector that happened to rally, regardless of which specific stocks they picked within it) is demonstrating something genuinely different from a manager who added value through security selection within sectors — and a real allocator wants to know which one actually happened before deciding whether that performance is likely to repeat.

Why this level of rigor exists at all

None of this is academic box-checking — a real allocator (a pension fund, an endowment, a wealth manager choosing between competing asset managers) is making a genuine, consequential decision about where to place client capital, and alpha, tracking error, information ratio, and attribution together give a far more complete, honest picture of manager skill than raw historical return alone ever could.