Tactical Asset Allocation is a bounded, shorter-horizon deviation from the SAA, taken on a capital-market view (a valuation call, a macro read, a momentum signal) within a stated band (a common institutional convention is ±5 percentage points per asset class) and evaluated on whether it earns its keep net of transaction costs and taxes, and on whether the tilt-taker had genuine forecasting skill rather than a story. The honest evidence on that last point deserves a hearing before any numbers: the aggregate record on tactical tilting, taken across the published literature, favors real skepticism, most tilts fail to earn back their frictions, and the few managers who persistently add value net of costs are rare and very hard to identify in advance.
Does asset allocation explain 90% of returns?
A famous, frequently mangled statistic belongs here. Brinson, Hood & Beebower's 1986 study is almost always summarized as "asset allocation policy explains 90%+ of investment returns", and that summary is a documented misreading. What their landmark R² actually measured was the share of the time-series variance in one fund's own returns, over time, explained by its policy mix, not the level of return, and not differences in return across funds. Ibbotson & Kaplan's 2000 paper, "Does Asset Allocation Policy Explain 40, 90, or 100 Percent of Performance?", untangles the popular misreading explicitly and is worth reading once precisely because the correct, narrower claim is still an important one: most of the month-to-month wobble in a diversified fund's returns comes from its policy mix moving with the market, which is exactly why the SAA decision deserves the care this node insists on, it is just not the same claim as "get the policy mix right and manager selection barely matters."
Advertisement
Worked example: a 5 percentage point tilt to equity
Apply a concrete tilt to Bharat Family Office's SAA: a tactical call to overweight Equity by 5 percentage points, funded by underweighting Debt by the same amount (within a stated ±5pp band), verified in Python:
TAA (+5pp Equity / −5pp Debt): Equity 59.99%, Debt 21.93%, Gold 18.07%, REIT 0.00%; E(R) = 10.09% (+25.0bp vs SAA), σ = 11.83% (+83.2bp vs SAA), Sharpe = 0.4725 (down from the SAA's 0.4855).
Read this result honestly, because it is the whole point of judging a tilt on risk-adjusted terms. The tactical call was directionally rewarded: return did rise. But risk rose disproportionately faster than return, so the Sharpe ratio actually fell. This is not a contradiction; it is the exact same lesson M9.01's Kelly framework teaches for a single position, recurring here at the asset-class level: being right about direction is necessary but not sufficient, the sizing of the bet has its own arithmetic, and a tilt that "worked" in the sense that the overweighted asset rose can still have been a poor risk-adjusted decision if it was sized without reference to how much incremental volatility it dragged in. An institutional TAA overlay is judged on Sharpe or information ratio, never on "did the tilted asset go up", precisely because the latter question, answered in isolation, would have called this tilt a success.
Northfield Endowment's TAA arc runs the identical logic in the other currency: an SAA at target σ ≈ 9.5% gives Equity 29.86%, Bonds 30.42%, Alts 39.72% (E(R) = 7.176%, Sharpe = 0.3133); a tactical tilt of +5pp Alts / −5pp Bonds (an illiquidity-premium call, in the spirit of the Swensen/Yale endowment model) moves E(R) up 16.0bp and σ up 58.3bp, landing the Sharpe ratio at 0.311, again, a modest, real, quantified give-up, not a free lunch.