Building a Portfolio Around Alpha and Beta
Investors who tilt a portfolio toward a handful of favorite stocks typically size those positions by gut feeling, mixing up how much they like an idea with how confident they should actually be in it. The single-index framework offers a disciplined answer: weight each active bet by its expected edge relative to its own uncertainty, then size the whole active sleeve against a passive core.
Splitting a portfolio into active and passive pieces
A useful way to think about any portfolio that combines index exposure with individual stock convictions is as two distinct pieces: a passive piece that simply holds the market, and an active piece made up of positions where the investor believes they have a genuine edge, a nonzero alpha. The single-index framework gives a specific rule for how to weight positions within the active piece, and separately, how large the active piece should be relative to the passive core.
The weighting rule within the active sleeve is that a position's size should be proportional to alpha divided by firm-specific variance, not to alpha alone. A stock with a large expected edge but also large firm-specific uncertainty should receive a smaller position than its headline alpha suggests, because that uncertainty is not compensated the way market risk is, it is exactly the risk a passive index investor would have diversified away for free. Conviction has to be weighed against precision, not treated as a single number.
The math: weighting two active positions
Take two candidate active positions. Stock A has an estimated alpha of 2% and a firm-specific variance of 0.09 (a 30% firm-specific standard deviation). Stock B has a lower estimated alpha of 1% but a much smaller firm-specific variance of 0.04 (a 20% firm-specific standard deviation). The raw weighting scores are 2 / 0.09 = 22.22 for Stock A and 1 / 0.04 = 25.00 for Stock B. Normalizing these to sum to 100%, Stock A gets 22.22 / (22.22 + 25.00) = 47.1% of the active sleeve, and Stock B gets the remaining 52.9%.
Notice the result: Stock B, with only half the raw alpha of Stock A, still receives the larger share of the active sleeve, because its much lower firm-specific uncertainty makes that smaller edge more reliable. Sizing purely by expected alpha, giving Stock A the larger position simply because 2% beats 1%, would have gotten the ranking backward once uncertainty is properly accounted for.
Add a third candidate to see how the weighting rule scales. Stock C has an estimated alpha of 1.5% and a firm-specific variance of 0.0625 (a 25% firm-specific standard deviation), giving a raw score of 1.5 / 0.0625 = 24.0, almost identical to Stock B's 25.0. Renormalizing across all three, the total raw score is 22.22 + 25.00 + 24.00 = 71.22, giving weights of 22.22 / 71.22 = 31.2% for Stock A, 25.00 / 71.22 = 35.1% for Stock B, and 24.00 / 71.22 = 33.7% for Stock C. Adding a third, comparably attractive candidate did not just dilute the first two proportionally; because each stock's weight depends on its own score relative to the new total, the three end up nearly evenly split despite meaningfully different alpha estimates, precisely because their alpha-to-variance ratios happened to land close together.
A second example: sizing the active sleeve itself
Once the active sleeve's internal weights are set, a second, separate question is how large the entire active sleeve should be relative to the passive market position. The combined active portfolio's alpha is the weighted average of its components: 0.471 × 2% + 0.529 × 1% = 0.94% + 0.53% = 1.47%. Its firm-specific variance, assuming the two residuals are independent, is 0.471² × 0.09 + 0.529² × 0.04 = 0.0199 + 0.0112 = 0.0311, giving a residual standard deviation of √0.0311 = 17.6%.
The active sleeve's information ratio, its alpha divided by its residual standard deviation, is 1.47% / 17.6% = 0.083, a modest figure typical of realistic active bets. Comparing this to the market's own reward-to-risk profile, assume a market excess return of 7% and a market variance of 0.0256. The initial weight the framework assigns to the active sleeve, before final risk scaling, is (alpha_active / variance_active) / ((E(R_m) - R_f) / Var(R_m)) = (0.0147 / 0.0311) / (0.07 / 0.0256) = 0.472 / 2.734 = 17.3%.
In plain terms, given this specific alpha and confidence level, the active sleeve earns roughly 17% of the total risky allocation, with the remaining 83% left in the passive market position. This result is highly sensitive to the assumed firm-specific variances: had both stocks' firm-specific uncertainty been half as large, at 0.045 and 0.02 respectively, the raw scores would double to 44.4 and 50.0, but because both the numerator and the denominator of the active sleeve's own information ratio move together, the recalculated active weight rises to roughly 30%, nearly double the original 17.3%, from that single change in assumed confidence. This result is also separate from, and sits inside, the earlier decision about the overall risky-versus-bills weight y: an investor who separately decided to hold 70% of total wealth in the risky combination, following the capital allocation logic described elsewhere, would end up with the active sleeve representing 70% × 17.3% = 12.1% of their total portfolio, a useful reminder that the active tilt's real-dollar size is smaller than the 17.3% figure suggests once it is placed inside the full three-layer structure of active stocks, passive market exposure, and bills.
What the record shows about active tilts
Studies of realized information ratios among professional active managers have consistently found that ratios in the 0.05 to 0.15 range, similar in scale to the 0.083 computed above, are common among managers with genuine, persistent skill, while ratios much above roughly 0.5 to 1.0, sustained over long periods, are rare and typically concentrated among a small number of specialist strategies rather than broad-market stock pickers. This matters because it calibrates expectations: an individual investor who believes their own stock convictions carry an information ratio of 0.3 or higher, sustained across many positions and years, is implicitly claiming a level of skill that outperforms most professional managers with far more resources and information access.
The historical record on individual active positions also shows that firm-specific variance estimates drawn from calm periods routinely understate the true uncertainty realized during stress, when firm-specific news and market-wide dislocation can compound each other. A framework built on an underestimated firm-specific variance systematically overweights the active sleeve relative to what a more honest, stress-tested uncertainty figure would justify.
Applying this without institutional infrastructure
An individual investor rarely has the tools to estimate firm-specific variance and alpha with institutional-grade precision, but the logic still applies directionally. The practical translation is to keep the large majority of a portfolio in a broad, low-cost index core, and to treat any individual stock conviction as earning a position size that scales down, not up, with how uncertain the underlying thesis is, even when the potential upside looks large. A high-conviction idea in a volatile, hard-to-analyze small company should generally receive a smaller position than a similarly high-conviction idea in a large, well understood, lower-volatility company, holding the expected edge itself roughly constant.
A second practical translation is capping the total active sleeve at a modest share of the overall portfolio, informed by an honest information ratio estimate rather than by enthusiasm. Given how rarely realistic information ratios exceed roughly 0.15 to 0.2 even among skilled professional managers, an individual investor's active sleeve, calculated the same way the worked example above computed 17%, will usually come out smaller than instinct suggests once the uncertainty term is taken seriously rather than glossed over.
A third, easy to overlook translation concerns how many active positions to hold at once, not just how large each one should be. The three-stock example above, where fairly different alpha estimates ended up producing three similarly sized weights, illustrates a broader pattern: the more candidate positions are considered together, the more the weighting rule naturally spreads capital across them rather than concentrating heavily in a single top pick, provided the candidates have roughly comparable alpha-to-variance ratios. An investor with genuine conviction in eight or ten reasonably well-researched ideas, sized this way, ends up with a meaningfully different, and typically less concentrated, active sleeve than one built around a single high-conviction stock, even if that single stock's raw alpha estimate looked more impressive in isolation.
Actionable breakdown
- Weight active positions by alpha relative to firm-specific risk.
- Do not size a position by expected return alone.
- A smaller, more confident edge can outrank a larger, shakier one.
- Size the whole active sleeve against the passive core deliberately.
- Compute or estimate the active sleeve's overall information ratio.
- Compare it honestly to realistic professional benchmarks.
- Use stress-tested, not calm-period, uncertainty estimates.
- Firm-specific variance measured in quiet markets understates risk.
- Widen uncertainty assumptions before finalizing any position size.
- Keep the passive core as the default, large majority position.
- Reserve the active sleeve for a modest, deliberately capped share.
- Let the sizing logic, not enthusiasm, set that cap.
Common pitfalls
The most common pitfall is sizing a position by raw expected return, giving the biggest bet to whichever idea has the highest headline alpha, without discounting for how uncertain that alpha estimate actually is. The worked example above shows directly how this can rank two positions backward.
A second pitfall is estimating firm-specific variance from a short, calm historical window and carrying that low figure forward into a position-sizing decision, which overstates the confidence the framework should actually assign and results in an active sleeve that is larger, and riskier, than the honest math would support.
A third pitfall is letting the active sleeve grow past its calculated share through a series of individually reasonable-looking additions, until it dominates the portfolio and the passive core that was meant to absorb most of the risk has shrunk to an afterthought. A fourth pitfall is forgetting that the active sleeve's weight is calculated against the risky portion of the portfolio, not the whole account, so a 17% active weight inside a risky sleeve that itself represents 70% of total wealth, as in the example above, works out to roughly 12% of the full portfolio, a distinction that matters when explaining position sizes to anyone reviewing the account from the outside.
The bottom line
Size any active stock bet by its expected edge relative to its own uncertainty, not by the edge alone, and let that same logic set a deliberate, usually modest, cap on the active sleeve as a whole.
Related reading: stock analysis fundamentals, building an asset allocation, estimating alpha and beta from real data, the single-index model, practical frictions in running this in real accounts.