Quick answer
A regular ranking of your portfolio positions by forward expected return, not past performance, reveals which holding competes least effectively for capital. The bottom-ranked position is the first sell candidate when new capital is needed or when a threshold review concludes the gap to the rest of the portfolio is material and has no near-term resolution. The ranking forces explicit, comparable forward-looking analysis across all holdings simultaneously.
What expected return means in practice
Expected return, in the context of ranking portfolio positions, is a forward-looking estimate of what a position is likely to return over a defined holding period, expressed as an annualized percentage. This is not a precise number. Equity returns are inherently uncertain, and any specific estimate will be wrong in its details. The purpose of computing expected return is not to predict the future with accuracy but to force each position's forward-looking case into a common, comparable format.
The most practical approach to computing expected return for a single position is scenario analysis. Define three plausible futures for the position: a bull case, a base case, and a bear case. For each case, estimate the return over your intended holding period. Then assign probabilities that the three cases represent how you expect the future to unfold, ensuring they sum to 100%. The expected return is the weighted average: probability of bull case times bull return, plus probability of base case times base return, plus probability of bear case times bear return.
As an illustration: suppose you hold a retail company and estimate that in the bull case (new store expansion succeeds, margins recover), the stock returns 40% over 18 months. In the base case (steady but unexciting), it returns 10%. In the bear case (credit cycle turns, consumer spending weakens), it returns negative 20%. You assign those scenarios 25%, 55%, and 20% probabilities respectively. The scenario-weighted expected return is (0.25 x 40%) + (0.55 x 10%) + (0.20 x -20%) = 10% + 5.5% + (-4%) = 11.5% over 18 months, or roughly 7.5% annualized.
The precision of this number is not the point. The value is that every other position in the portfolio is being described in the same terms. When you have ten positions with annualized expected returns of 14%, 13%, 11%, 10%, 9%, 8%, 8%, 7%, 5%, and 3%, the ranking is immediately visible and the comparison is direct.
Building a return-ranking table
A return-ranking table is a structured document that captures the key inputs and outputs for every position in the portfolio. Building it forces the analytical discipline that informal reviews skip. A useful table includes the following columns for each position.
Position and current weight. The ticker or name and the current percentage of the portfolio it represents. This is the baseline context for any allocation decision.
Current thesis summary. A one or two sentence description of the investment thesis as it stands today, based on current information. Not the thesis at purchase. The thesis you would defend in a review right now. If you cannot write a current thesis in one or two sentences, the position has become a comfort hold rather than an analytical conviction.
Expected return range. The scenario-weighted expected return as computed above, expressed as a range around the central estimate to reflect the inherent uncertainty. The range of outcomes in the bull and bear cases represents the width of this range. A position with a 7.5% central expected return and a bull-to-bear range of -20% to +40% has different risk characteristics than one with a 7.5% central expected return and a bull-to-bear range of -5% to +18%, even though the central estimate is the same.
Conviction level. A score from 1 to 5 that reflects how confident you are in the underlying analysis. A 5 represents a thesis built on proprietary research, multiple independent data points, and high-quality management diligence. A 2 represents a thesis built primarily on publicly available information with limited proprietary insight. Conviction should adjust the effective weight you give the expected return number: high expected return with low conviction is less valuable than it appears.
Time horizon. The intended holding period for the position. Expected returns are not comparable across wildly different time horizons without normalization. Annualizing each expected return puts them on the same basis.
With these columns populated, sorting the table by the conviction-adjusted annualized expected return produces a ranking that captures both the return opportunity and the quality of the analysis supporting it. The bottom of the ranking is the most natural starting point for sell review.
Why past performance is the wrong sort key
The intuitive approach to reviewing a portfolio is to look at what has gone up and what has gone down. Winners feel safe. Losers feel like problems. This intuition is backwards for sell discipline purposes and particularly dangerous in the opportunity-cost context.
Past performance measures what has already happened. It is history. Forward expected return measures what is likely to happen from here. The two are not the same variable, and they are not reliably correlated. A stock that rose 60% last year may now be priced at a premium that leaves a very low forward expected return. A stock that declined 30% last year may now trade at a valuation that implies strong forward returns if the business recovers as expected.
Sorting by past performance and selling the bottom performers is often described as "cutting losers," which sounds like discipline. In practice, it is frequently momentum chasing in reverse: you are selling positions that have recently moved against you and holding positions that have recently moved in your favor, without asking whether the fundamental forward case for each has changed. This produces a portfolio full of stocks that have already had their run and short of stocks whose recoveries are still ahead.
The correct sort key for sell-review purposes is forward expected return. The relevant question is not what has this position done but what is this position likely to do from here given its current price, its current thesis, and the current information available. That question is exactly what the return-ranking table is designed to answer.
How to handle uncertainty in the ranking
One objection to expected return ranking is that the estimates are imprecise and therefore the ranking is unreliable. This objection misunderstands the purpose of the exercise. The ranking is not expected to be precise in absolute terms. Its value is in revealing relative differences and forcing explicit forward-looking analysis where implicit assumptions were previously hidden.
Several techniques make uncertainty more visible and more useful in the ranking context.
Use ranges rather than point estimates. Rather than saying a position has a 10% expected return, say it has an expected return of 7% to 13% depending on how the key scenarios resolve. Positions whose ranges overlap substantially are not clearly distinguishable and should not be treated as though they are. Positions whose ranges do not overlap at all are clearly distinguishable and can be ranked with confidence.
Apply conviction discounting. As described above, a high expected return estimate from low-conviction analysis deserves less weight than the same number from high-conviction analysis. A simple approach: multiply the expected return by a conviction fraction, where a conviction score of 5 out of 5 produces no discount and a conviction score of 2 out of 5 produces a 40% discount. A 15% expected return at conviction 2 becomes an effective 9%, which may rank lower than a 12% expected return at conviction 5.
Flag positions with very high uncertainty. A position whose bull-to-bear range spans more than 80 percentage points is not comparable to a position whose bull-to-bear range spans 20 percentage points, even if the central expected return is the same. Wide-range positions represent a qualitatively different kind of bet. Mark them separately and evaluate the portfolio's aggregate exposure to wide-range positions as a risk dimension independent of the expected return ranking.
The quarterly ranking review
A return ranking is most valuable when done at regular intervals rather than only in response to news or price movements. Quarterly is the most common cadence for this type of review: frequent enough to catch meaningful drift, infrequent enough to avoid over-rotation on short-term noise.
The quarterly review process has a standard structure. First, update the thesis summary for each position based on the most recent earnings reports, management communications, and relevant macro developments since the last review. This step alone is valuable: positions that cannot produce an updated thesis have lost analytical currency.
Second, rebuild the scenario-weighted expected returns from scratch for each position using current price as the starting point. Do not carry forward last quarter's estimates with minor adjustments. Starting fresh forces you to re-justify each position's return profile at its current valuation rather than anchoring to what you thought three months ago.
Third, update conviction scores based on whether the evidence supporting each thesis has strengthened or weakened. A thesis that has been confirmed by several quarters of data deserves higher conviction than one that is still waiting for its first catalyst.
Fourth, sort the table and examine the bottom two or three positions. Ask: if new capital were available today, would any of these positions be funded over any currently unlisted opportunity? If the answer is no, they are at least provisional sell candidates.
Fifth, document the review output and the decisions made or deferred. A decision journal entry for each quarterly review creates a record that surfaces patterns over time, such as repeatedly deferring a sell decision on the same position quarter after quarter.
When a position stays despite ranking low
Bottom rank is not an automatic sell signal. Three types of legitimate reasons to hold a low-ranked position exist, but each must be explicit and time-bounded.
Tax timing. A position with a large embedded short-term capital gain may be worth holding until the gain qualifies for long-term treatment. The after-tax expected return comparison changes when you factor in the difference between short-term and long-term capital gains tax rates. A position with a 6% forward expected return but a large short-term gain to realize may be more attractive on an after-tax basis than a 10% forward expected return position if exiting triggers a 37% federal tax on the gain. The calculation is position-specific, but the principle is that the ranking should ideally be done on an after-tax basis for taxable accounts.
Liquidity constraints. A position in a small or thinly traded company may not be exitable at a fair price in the near term. Forcing a sale through a thin market creates bid-ask costs and market impact that can significantly erode the realized proceeds. In this case, the exit is valid but the timing needs to be managed rather than immediate.
Near-term catalyst. A position with a clear, time-bounded upcoming catalyst, a regulatory decision expected in eight weeks, an earnings report that will resolve the key uncertainty, a strategic announcement expected in the current quarter, may deserve a conditional hold. The logic is that the current expected return estimate is highly sensitive to the catalyst outcome, and waiting for that resolution produces materially better information at a modest cost of continued exposure. This rationale is legitimate only when the catalyst is specific and near-term. "Waiting for the thesis to play out" with no specific event in view is not a catalyst; it is an indefinite hold justified by hope.
Frequently asked questions
What does expected return mean when ranking portfolio positions?
Expected return, in the context of ranking portfolio positions, is a forward-looking estimate of what a position is likely to return over a defined holding period, expressed as an annualized percentage. It is not a precise prediction. It is a scenario-weighted estimate built by assigning probabilities and outcomes to a set of plausible futures, typically a bull case, a base case, and a bear case. The point of the exercise is not precision but comparability: forcing every position to be described in the same currency of annualized expected return makes it possible to rank them meaningfully against each other.
How do you handle uncertainty when ranking positions by expected return?
Uncertainty is incorporated through scenario analysis and conviction scoring. For each position, assign probabilities to a bull, base, and bear scenario and estimate the return in each. The probability-weighted average is the expected return estimate. Separately, assign a conviction score from 1 to 5 that reflects how confident you are in the underlying analysis. A position with a 20% expected return but a conviction of 2 should rank differently from a position with a 15% expected return and a conviction of 5. One approach is to multiply the expected return by the conviction fraction to produce a conviction-adjusted expected return. This rewards analytical depth over headline return estimates.
Why is past performance the wrong sort key for ranking positions?
Past performance measures what a position has already done. Forward expected return measures what it will do from here. The two are not correlated in any reliable way. A stock that has risen 50% over the past year has a higher past performance but may now be expensive relative to its intrinsic value, leaving a lower forward expected return than a stock that has been flat. Sorting by past performance instinctively ranks recent winners at the top and recent losers at the bottom, which is the opposite of what a forward-looking sell discipline should produce. Selling the laggard and holding the recent winner is a reflexive response to price history, not an analysis of future returns.
When does the bottom-ranked position trigger a sell?
The bottom-ranked position triggers a sell review when two conditions are met: it ranks materially below the rest of the portfolio by expected return, and either new capital is needed for a higher-ranked opportunity or a threshold review date has been reached. The gap must be material, not marginal. A position ranking last by 1 or 2 percentage points may not justify the tax cost and friction of reallocation. A position ranking last by 8 to 10 percentage points with no near-term catalyst to improve its ranking is a strong candidate for exit or reduction.
When should a position stay in the portfolio despite ranking low?
Three situations justify holding a low-ranked position. First, tax timing: a position with a large embedded short-term gain may be better held until the gain qualifies for long-term treatment, changing the after-tax expected return comparison materially. Second, liquidity: a thinly traded position may not be exitable at a fair price in the near term, making a forced exit a worse outcome than continuing to hold. Third, a near-term catalyst: if a specific upcoming event, an earnings report, a product launch, a regulatory decision, is expected to resolve the current uncertainty in the bull direction, it may make sense to wait for the catalyst before deciding. In all three cases, the holding is conditional on a documented rationale, not an indefinite stay.