How to Estimate Your Expected Return Without Fooling Yourself

The easiest way to fool yourself about returns is to pick one number and call it a forecast. A 12% expected return sounds disciplined until you ask where it comes from. If the answer is “because the market has gone up over time,” you do not have an estimate. You have a mood.

Long-run U.S. stock returns are real, but they are not magic. From 1926 through 2023, the U.S. equity market returned about 10.0% a year nominally and about 6.8% after inflation, according to the annual Stocks, Bonds, Bills, and Inflation data series ^[1]. That is a useful anchor. It is also a terrible excuse to assume your next portfolio, your next factor tilt, or your next stock pick will earn the same thing. The job is to convert history into a range you can actually act on.

A single return number is usually a disguised sales pitch

Most return estimates fail for the same reason most stock pitches fail: they collapse uncertainty into a point estimate. A point estimate feels precise, but precision is not the same as accuracy. If you are evaluating a strategy, a fund, or a portfolio change, the first question is not “what will it return?” It is “what has to go right for that number to happen?”

History gives you a baseline, not a guarantee. The U.S. market’s long-run nominal return of roughly 10.0% and real return of roughly 6.8% from 1926 to 2023 is a broad market average across wars, inflation spikes, recessions, and valuation cycles ^[1]. That average is useful because it is hard to beat by accident. It is misleading because it hides the path. A 10% average can come from a smooth 10% every year or from a brutal sequence of flat years followed by a few huge rebounds. Investors care about the second version more than they admit.

If you want a cleaner framework, start with three numbers: a baseline return, a valuation adjustment, and a drag estimate. That is the same spirit behind AIBROKER’s three-numbers-that-matter framework, except here the numbers are not just for portfolio monitoring; they are for pre-commitment analysis. You are trying to decide whether the expected reward is large enough to justify the risk, the cost, and the time.

Direct judgment: most investors overestimate their edge because they use the market’s historical return as if it were their own expected return. It is not. If your strategy adds fees, taxes, turnover, or concentration, your expected return starts below the market, not beside it.

For readers building a portfolio from scratch, this is where compound growth and inflation and real returns belong in the same sentence. Compound growth is real. So is inflation. Ignoring either one is how people end up with fantasy wealth projections.

Historical baselines work only after you strip out survivorship bias and regime blindness

Historical averages are useful only if the history is clean. It usually is not. Survivorship bias makes dead strategies disappear from the sample, which flatters the survivors. That problem is not academic. It is one reason backtests and product brochures can look better than live results. AIBROKER’s survivorship-bias guide covers the mechanics; the point here is simpler: if your baseline comes from a cherry-picked index, a surviving fund family, or a hand-built sample of winners, your expected return is already too high.

Regime blindness is the second trap. A strategy that worked in falling rates, low inflation, and expanding multiples may not behave the same way when the macro backdrop changes. That is why a baseline should be segmented by regime, not just averaged across decades. AIBROKER’s regime-detection coverage is relevant because the return you should expect from a strategy depends on the market state you think you are in. A momentum tilt, for example, can look brilliant in persistent trends and mediocre in choppy reversals. The average hides the difference.

There is also a mechanical issue. If you are estimating returns from a backtest, you need point-in-time data, realistic costs, and a rule for delistings. A backtest that ignores those details is not a forecast. It is a fiction with charts. AIBROKER’s backtest-checklist and point-in-time-backtesting explain the hygiene. Use them before you trust any historical baseline.

Direct judgment: a backtest that does not survive a survivorship-bias check is not conservative. It is optimistic by construction.

If you want a practical filter, ask whether the historical baseline is broad enough to include bad years, bad regimes, and bad implementation. If not, it is not a baseline. It is a sales deck.

Reader warning
If the strategy’s historical return depends on a tiny sample of exceptional years, cut the estimate hard. Rare good years are not a stable source of edge.

Valuation anchors beat narrative forecasts when you need a medium-term estimate

For broad markets and many individual stocks, valuation is the best anchor you have for medium-term return expectations. It is not perfect. It is better than vibes. Starting valuation does not tell you next quarter’s return, but it does tell you a lot about the odds over the next 5 to 10 years. Robert Shiller’s data show that high CAPE ratios have historically been followed by lower subsequent long-horizon real returns, while low CAPE starting points have tended to precede stronger returns ^[2]. The relationship is noisy, not mechanical. That is exactly why it is useful.

For a stock, the valuation anchor can be a P/E, EV/EBITDA, free-cash-flow yield, or a normalized earnings yield. For a fund or strategy, it may be the starting yield, the factor spread, or the discount/premium to intrinsic value. The point is to translate “cheap” or “expensive” into a return implication. If you buy a stock at 30 times earnings, your future return has to come from earnings growth, multiple expansion, or both. If the multiple compresses, growth has to work harder. That is arithmetic, not opinion.

Here is a simple way to think about it:

1. Estimate normalized earnings or cash flow.
2. Apply a reasonable exit multiple, not the one you hope for.
3. Add dividends or buybacks if they are real and sustainable.
4. Subtract dilution, taxes, and fees.

This is where many investors get sloppy. They use the best-case multiple as the base case and call the downside “temporary.” That is backwards. The base case should be the most defensible outcome, not the most flattering one.

For readers comparing factor tilts, this is also where factor investing and momentum vs. value become practical. A factor premium is not a free lunch. It is a compensation for something, and that something can disappear for years. Your expected return should reflect that possibility.

Scenario analysis works only if the bull case is harder to reach than the bear case is to avoid

Scenario analysis is where return estimates become decision-ready. A good scenario set does not ask, “What could happen?” Everything could happen. It asks, “What has to happen for this strategy to earn enough to justify the risk?” That means you need at least three cases: base, bull, and bear. Each case should change the same drivers: growth, valuation, payout, and costs. If you change only the upside assumptions, you are not doing scenario analysis. You are writing fan fiction.

Use explicit probabilities if you can defend them. If you cannot, use a weighted range and be honest about the uncertainty. The point is not to pretend you know the future. The point is to see whether the downside is survivable and the upside is worth the wait. A strategy with a 20% chance of a great outcome and an 80% chance of mediocre results may still be attractive. A strategy that needs everything to go right is usually not.

Here is a simple decision template you can reuse:

Now add a haircut rule. If your bull case depends on a valuation multiple above the last decade’s average, cut it. If your base case assumes zero fees or zero taxes, cut it. If your bear case still looks too kind, cut it again. That sounds harsh because it is. Markets are harsh.

For portfolio construction, this is where position sizing and risk-based sizing matter. A high expected return that comes with a wide left tail may deserve a smaller allocation than a lower expected return with a tighter range. Expected return alone is not enough.

Direct judgment: the bull case is usually too easy to write and too hard to realize. If your scenario table does not make the upside harder than the downside, you have not been honest with yourself.

Checklist: separating signal from optimism
1) Did I use a historical baseline with delistings and costs?
2) Did I anchor the estimate to a valuation or yield metric?
3) Did I write a bear case that includes multiple compression?
4) Did I haircut fees, taxes, and slippage?
5) Would I still buy if the base case were 25% lower?

Fees, taxes, and slippage are not small adjustments; they are the first haircut

Investors love to talk about gross returns because gross returns are flattering. Net returns are what you spend. The gap is often larger than people expect. Expense ratios, trading costs, bid-ask spreads, taxes, and turnover all drag on realized performance. A strategy that looks attractive before costs can become mediocre after them. AIBROKER’s coverage of fees and hidden costs and transaction costs and slippage is worth reading alongside any return estimate.

For long-term investors, taxes can matter as much as fees. A high-turnover strategy in a taxable account may have a strong gross edge and a weak after-tax edge. That is not a minor detail. It is the difference between a strategy that compounds and one that leaks. If you are comparing account types, the placement of assets matters too; see tax-efficient asset location and rebalancing without a tax bomb.

Here is a simple haircut rule set I use when translating a raw return estimate into something investable:

• Subtract the fund’s expense ratio or estimated implementation cost.
• Subtract a slippage allowance if the strategy trades frequently or in thin names.
• Subtract an after-tax drag estimate if the account is taxable.
• Reduce any return assumption that depends on unusually favorable execution.

That last line matters. A strategy that only works with perfect fills is not robust. It is fragile. If you need a reminder of how quickly execution can matter, AIBROKER’s bid-ask spread and order types explain why the price you think you paid is often not the price you actually got.

Direct judgment: most investors understate implementation drag because it is boring. Boring is expensive.

A worked example: turning a vague 14% hope into a range you can defend

Suppose you are evaluating a U.S. equity strategy that claims a 14% expected annual return. That number sounds attractive. It also sounds suspicious. Here is how to pressure-test it.

Step 1: baseline. Use the long-run U.S. equity return as a starting point: about 10.0% nominal and 6.8% real from 1926 to 2023 ^[1]. If the strategy is just a broad market ETF, the baseline is already close to the answer. If it is a concentrated factor tilt or active strategy, the baseline should be lower after costs.

Step 2: valuation anchor. Ask whether the current starting valuation is above or below its own history. If the strategy buys expensive growth stocks, the expected return should be haircut for multiple compression risk. If it buys cheaper names, the expected return can be higher, but only if the valuation discount is real and not a trap. Shiller’s work suggests starting valuation matters over long horizons ^[2].

Step 3: scenario table. Build three cases. Base: 8% to 10% after costs. Bull: 12% to 14% if growth stays strong and valuation holds. Bear: 0% to 4% if multiples compress and earnings disappoint. If the strategy cannot survive the bear case, you do not have an investment thesis. You have a timing bet.

Step 4: haircut rules. Subtract 0.5% to 1.0% for fees and trading friction if the strategy is active. Subtract more if it is taxable and turnover is high. If the strategy requires leverage, concentration, or frequent rebalancing, widen the range again. Leverage can magnify returns, but it also magnifies the cost of being wrong; AIBROKER’s margin and leverage guide is the right companion piece.

After those adjustments, the 14% claim may shrink to something like 7% to 11% net, with a wide enough range that you can make a real decision. That is the point. A narrower, more honest range is more useful than a flattering point estimate.

This is not a forecast. It is a filter. If the lower end of the range is unacceptable, walk away or size down. If the upper end requires heroic assumptions, ignore it.

Decision tree
If the base case is below your hurdle rate, do not buy.
If the bull case is plausible but the bear case is intolerable, size smaller.
If the range is wide and the midpoint is still attractive, you may have a real edge.

The margin of safety is just a haircut rule with teeth

Margin of safety is one of the most abused phrases in investing. People use it to mean “cheap.” That is too vague. A real margin of safety is the gap between your conservative estimate of value and the price you pay, after you have already haircut the assumptions. It is not a feeling. It is arithmetic.

For expected return work, margin of safety shows up as a required spread between your base case and your hurdle rate. If your hurdle is 8% and your conservative estimate is 7%, the answer is no, even if the bull case looks exciting. If your conservative estimate is 11% and your bear case is still survivable, you may have enough cushion. The cushion matters because forecasts are wrong more often than they are right.

There is a hidden tradeoff here. The more conservative you are, the fewer opportunities you will accept. That can feel frustrating. Good. Frustration is cheaper than overconfidence. The goal is not to reject everything. The goal is to avoid paying full price for uncertain outcomes.

One useful rule: if your expected return estimate depends on a single heroic assumption, cut it in half. If it depends on two heroic assumptions, cut it to zero. That sounds severe because it is. Markets reward realism more reliably than optimism.

For investors who want to formalize this process, AIBROKER’s investment policy statement guide and when to sell framework help turn a return estimate into a rule you can live with. That is the real test. Not whether the estimate looks elegant on paper. Whether you can still defend it after a bad quarter.