Research guide

Delaware Statutory Trust Returns (2026): Why Averages Can Hide the Range of Outcomes

By Sequoia ResearchPublished Read as Markdown

What Is This Article About?

An average DST return can leave out large differences between programs and investors. This article examines a defined public record available in October 2026, then explains how to read the middle result, extremes and less successful outcomes. Original examples show how weights, missing records and timing can change what the same headline means. The aim is to describe the evidence more clearly, without turning a past range into a forecast.

Table of Contents

Ask What the Average DST Return Describes

An average compresses many observations into one number. That can help a reader see a broad pattern. It also removes details. Two sets of investments can share an average while exposing their investors to very different cash outcomes.

Start with the unit being counted. Is each observation one program, one property, one share class or one investor account? A property portfolio can contain many buildings within one DST. Counting each building as a separate investor outcome would change the question.

Next, identify the return measure. A distribution rate measures payments against a stated base over a period. A cumulative cash gain compares all cash received with the money paid in. An annualized measure accounts for time under a specified method. None should be mixed with the others just because each appears as a percentage.

A search for Delaware statutory trust returns may bring up all these figures together. Before comparing them, write the measure next to each number. If the method is unclear, leave the comparison open. A neat average of unlike figures does not repair the mismatch.

The SEC's September 2022 performance bulletin asks investors to examine methods, fees, market conditions and selection of past results. Those checks apply before calculating a median or chart. Better statistics cannot make incomplete or mismatched source data complete. SEC: Performance Claims.

Then define the population, meaning the group the analysis covers. “Completed programs listed in this report” is a usable boundary. “DSTs” is too broad when the evidence comes from one sponsor and one type of investment. Give the cutoff date and explain which programs remain outside the calculation.

The result should answer a sentence with a clear subject: “Among these eight completed programs, using this cash measure, the middle result was this amount.” Readers can check that statement. A claim that the number represents what a future investor should expect needs much more evidence.

A Defined 2026 Sample of Published Return Ranges

ExchangeRight's eight-page Investment Track Record, downloaded October 17, 2026, generally dates its statistics to August 31, 2026. We inspected all 22 net-leased full-cycle rows on pages 4–5: Portfolios 1–19, 21, 22 and 28. Other product groups were excluded.

The stated annual-average-return ranges reflect differing brokerage or advisory fees and include 721 exits valued using the acquiring REIT's net asset value. From the printed endpoints, our equal-row medians are 7.165% for lower endpoints and 10.315% for upper endpoints. The smallest lower endpoint is 6.37%; the largest upper endpoint is 12.12%, a 5.75-percentage-point span.

These describe published endpoints, not a median investor return, cash-only result or independently verified IRR. Investor counts and dated cash ledgers were not obtained. The listed full-cycle dates run through November 19, 2024; this is evidence available in 2026, not a set of 2026 exits. ExchangeRight: Investment Track Record.

All examples below are invented and separate from that named source. They explain the methods; they do not fill gaps in its records.

Compare the Mean With the Median

The mean is the sum of values divided by their count. The median is the middle value after sorting. With an even number of values, average the two in the middle. These are different ways of describing the center of a group, as explained in the NIST statistical handbook. NIST: Measures of Location.

Consider a fictional study called the Birch Study. It covers eight completed programs. For each program, assume an investor put in $100,000, made no later contribution and received all final cash within the same five-year period. The measure is cumulative cash gain as a percentage of that initial amount, before personal taxes and inflation.

Each modeled investor received $20,000 in operating payments during the period. The remaining receipts came at exit. All program costs and fees are already reflected in the cash figures. There are no additional investor charges in this base case. The letters below identify separate programs, not different years of one investment.

Program All cash received, including exit Cumulative cash gain or loss
A $70,000 −30%
B $95,000 −5%
C $105,000 5%
D $110,000 10%
E $115,000 15%
F $120,000 20%
G $125,000 25%
H $180,000 80%

The eight percentages add to 120%. Dividing by eight gives a mean gain of 15%. This is a cumulative five-year measure, not a 15% annual return. The middle two results are 10% and 15%, so the median is 12.5%.

No program earned exactly 12.5%. A median can fall between actual observations. It describes a position in the sorted list. It does not identify a particular investment that an investor could have bought.

The mean exceeds the median because the high result pulls the arithmetic average upward. That does not make the mean false. It tells us that one summary is not enough. Reporting both helps readers see the effect of uneven outcomes.

Now compare an entirely separate invented group with cumulative gains of 8%, 10%, 12%, 14%, 16%, 18%, 20% and 22%. Its mean is also 15%, but its median is 15%. Every outcome is positive and close to the center. The shared mean conceals the wider spread and two losses in the Birch Study.

This comparison is deliberately controlled. Real program groups may differ in dates, fees and property types. The example isolates the effect of the distribution of results. It does not establish which real sponsor is safer or which future program will perform well.

Read the Range and the Middle Half Together

The range describes the distance from the smallest value to the largest. In the Birch Study, the endpoints are −30% and 80%. Their difference is 110 percentage points. In the second group, the endpoints are 8% and 22%, only 14 points apart.

Use percentage points when subtracting rates. Saying that the first range is “110%” could suggest a percentage change relative to a base. Here, we are simply measuring the distance between two percentage values.

The range depends on only two observations. It says little about where the other six sit. A record with one very high result and many modest results can have the same endpoints as a record spread more evenly between them. The middle portion deserves its own description.

One useful measure is the interquartile range, or the width of the middle half. NIST distinguishes this central spread from the full range, which uses the extremes. Different methods can place quartile cutoffs slightly differently, especially in a small group, so name the method. NIST: Measures of Scale.

For this article, split the eight sorted values into two halves of four. Find the median of each half. The lower half is −30%, −5%, 5% and 10%; its midpoint is 0%. The upper half is 15%, 20%, 25% and 80%; its midpoint is 22.5%. The distance between those cutoffs is 22.5 percentage points.

The 0% cutoff is not an observed result. It sits between −5% and 5%. Nor does the middle-half measure erase the loss of 30% or the gain of 80%. Keep those extremes beside it so that a central summary does not hide an important event.

A simple presentation can show the count, mean, median, minimum, maximum and middle-half method together. A graph can help, but with only eight rows, the full table may be easier to read. There is no need to hide a small dataset behind a complex statistical score.

Avoid reading an observed range as a safety boundary. The next investment could finish below the worst result in the file or above the best. Historical endpoints are observations, not contract limits. They also do not tell you how likely an outcome is.

The same caution applies to the middle half. It is not a forecast interval or a statement that a new investor has a fifty-fifty chance of landing within those cutoffs. It describes where values sit in this chosen historical example.

Count Losses and Shortfalls Without Hiding Them

A positive average can coexist with losses. In the Birch Study, Programs A and B returned less total cash than the original contribution. Two of eight programs, or 25%, have a negative cumulative cash result. That count should appear beside the positive 15% mean.

A loss count alone also leaves out severity. One investor is short $30,000; the other is short $5,000. Their combined shortfall is $35,000 on $200,000 invested in those two programs. That is 17.5% of their combined starting capital, before the excluded taxes and inflation.

The $35,000 is not the study's net loss. The other six programs produce $155,000 of gains, leaving $120,000 of net gain across all eight. Preserve both facts. The gain elsewhere does not mean that an investor who owned only Program A recovered the missing cash.

A chosen threshold can answer another question. Suppose the research file asks how many programs produced less than a 10% cumulative gain. A, B and C fall below that threshold. That is three of eight, or 37.5%. Program D equals 10% and is not included in a strict “below 10%” count.

The threshold needs a reason. It might come from a research question or an investor's stated comparison point. It is not automatically an industry standard or a minimum suitable return. Label it as chosen, and avoid moving it after seeing the results to create a more favorable count.

Also separate recovering nominal dollars from meeting spending needs. Program B returns $95,000 in total, including the $20,000 paid along the way. Its remaining exit payment is $75,000. An investor expecting the full original $100,000 back at exit would face a different shortfall than the $5,000 lifetime cash loss.

Specifically, the exit check is $25,000 below that $100,000 expectation. The lifetime shortfall is $5,000 because $20,000 had already arrived. Both calculations are valid, but they answer different questions. Calling the exit shortfall a total investment loss would ignore earlier cash.

This is why a return-range report should state what money is being measured. Whole-period cash, the final check and annual spending money can all matter. Their amounts should not be substituted for one another.

Distinguish Program Weights From Investor Dollars

An equal-program mean gives each program one vote. A capital-weighted measure gives more influence to programs with more money invested. Neither is automatically the right answer. Each describes a different experience.

In the base Birch Study, all eight investments start at $100,000. Total contributions are $800,000 and total receipts are $920,000. The $120,000 gain is 15% of the contributions. Equal starting amounts make the overall cash ratio agree with the equal-program mean.

Now use a separate allocation case. Keep each program's cumulative percentage result unchanged, but change how much a fictional investor places in each. Assume the cash result scales exactly with the contribution and the same costs are already included. This case does not change the original study's observed program results.

Program Separate allocation Gain or loss at the stated program percentage
A $400,000 −$120,000
B $200,000 −$10,000
C $100,000 $5,000
D $100,000 $10,000
E $50,000 $7,500
F $50,000 $10,000
G $50,000 $12,500
H $50,000 $40,000
Total $1,000,000 −$45,000

This allocation loses 4.5% in cumulative nominal cash terms. The equal-program mean remains a positive 15%. More money was placed in the two losing programs, so this investor's result differs sharply from the unweighted headline.

The calculation is a contribution-weighted cumulative cash measure. It is not an average annual return or a combined internal rate of return, called IRR. A combined IRR requires the relevant dated payments. Averaging program IRRs, even with capital weights, does not generally recreate that calculation.

If you report capital weights, identify the capital field. Original investor equity, total property cost and current estimated value are different bases. Debt is not investor equity. A large property purchase price should not silently become the weight in an investor-capital result.

Check whether every program uses the same field. If capital is unavailable for one program, the equal-program result may still be calculable. A complete capital-weighted result is not. Do not use an average-sized placeholder unless the output is clearly labeled as a sensitivity based on that assumption.

The practical question is what the reader wants to understand. A program-level summary describes the selected programs. An investor-level summary needs that investor's allocation, fees and dates. A broad sponsor mean cannot serve both purposes without further work.

Test How Much One Large Result Changes the Average

A sensitivity test can reveal how much a headline depends on an extreme value. It should supplement the complete result, not replace it with a more comfortable number.

In the Birch Study, the 80% result contributes $80,000 of the total $120,000 gain. It supplies two-thirds of the net gain across the equal $100,000 investments. That is a clear way to describe its influence without calling the result invalid.

Remove that one program only for a labeled test. The seven remaining percentage results add to 40%. Their mean is about 5.71%. This is not the mean of all eight programs. It answers what the selected average would look like without the highest result.

Removing both extremes produces another answer. The remaining six results total 70%, for a mean of about 11.67%. That procedure removes the −30% loss as well as the 80% gain. It cannot be described as a complete account of what investors experienced.

The median can be less sensitive to the size of an extreme. If the highest result changed from 80% to 180%, while all others stayed fixed, the median would still be 12.5%. The mean would rise from 15% to 27.5%. The rank of the highest result stayed the same, while its dollar size changed greatly.

That stability is useful, but it has a limit. An investor with a large allocation to the changed program would care about the dollar difference. A stable median is not a reason to ignore it. Center, spread and invested amounts belong together.

A source error requires a different response from an extreme result. First check whether the unusual figure is correctly transcribed, uses the same measure and belongs to the group. If it is wrong, correct it with a record of the source change. If it is valid, keep it in the main analysis.

Do not label every unusual value an outlier and discard it. A large loss may be the most relevant fact for a reader concerned about downside. A large gain may explain why a mean looks strong. Removing either without explanation can defeat the purpose of the review.

Keep Missing and Unfinished Programs Visible

The spread of observed results says nothing definite about results that were not obtained. A blank cell is not a zero, and an unfinished program does not yet have a final cash outcome.

Suppose a separate fictional register contains twelve programs. Eight have ended and have complete cash records; they are the eight Birch Study results. Three remain active. One ended, but its final cash record is unavailable. The observed eight-program mean remains 15%, but it covers only eight of twelve listed programs.

By count, that is about 66.67% of the register and eight of nine completed programs, or about 88.89% of the completed group. These are evidence-coverage measures. They do not show that the missing result resembles the available ones.

To see the effect of the missing completed record, try explicit alternatives. If its cumulative gain were −40%, the nine completed results would sum to 80%, giving a mean of about 8.89%. If it were 0%, the mean would be about 13.33%. If it were 40%, the mean would be about 17.78%.

Those alternatives are chosen values, not estimates or boundaries. The missing outcome could be outside them. Do not describe the three answers as the likely range. Their purpose is to show that an unavailable result can matter to a headline.

The median also changes in this test. Adding −40% or 0% makes the fifth sorted value 10%. Adding 40% makes it 15%. A missing row can affect the middle as well as the mean, depending on where it belongs in the ordered list.

Leave the three active programs out of a final cash-return calculation unless a different, clearly labeled interim method is used. Assigning them zero would assert a result that has not occurred. Assigning their original projection would mix forecasts with completed outcomes.

Completion itself can affect selection. Programs that sell quickly may enter a completed table sooner than delayed programs. That observation does not prove intentional selection or that the delayed programs will lose money. It identifies why a completed-only record should not be treated as every program's eventual outcome.

Retain the full register beside the results table. Record each program's status, source date and reason for exclusion. If a later report resolves a missing row, update the count and statistics together. A revised mean without an updated population can be hard to interpret.

Separate Investor Ranges From Program Ranges

A range within one program and a range across programs are different objects. One might reflect fee choices, entry dates or exit forms. The other might reflect different properties and financing. Combining them without labels can hide what actually varies.

Imagine one completed fictional program with ten investor accounts. Nine accounts have a 10% cumulative cash gain; one has a 20% gain because its documented terms differ. The lowest and highest values are 10% and 20%, but the account median is 10% and the equal-account mean is 11%.

The midpoint of the endpoints is 15%. That midpoint is neither the mean nor the median of the accounts. Knowing only the endpoints would not reveal how many accounts sit near each end. The same 10%–20% range could describe a very different account mix.

For example, if one account earned 10% and nine earned 20%, the mean would be 19% and the median 20%. The endpoints would be unchanged. This is why a published range cannot be turned into a typical investor result by taking its midpoint.

Fee treatment can create differences even when the underlying property is shared. In a separate example, two investors each pay $100,000 and receive $130,000 of program cash over the same period. One has no outside charges; the other pays $3,000 in separate account fees. Their nominal cash surpluses are $30,000 and $27,000.

Relative to the original $100,000 contribution, those surpluses are 30% and 27%. If the second calculation instead uses all $103,000 paid as the denominator, its net surplus ratio is about 26.21%. State which convention is used. Neither percentage is a dated IRR.

Do not subtract a fee twice. If the $130,000 receipt figure already reflects a particular program charge, taking that charge off again understates the result. Outside fees need evidence that they are truly additional to the reported cash.

Exit form creates another distinction. Cash received and a continuing interest valued at an estimated price are not identical evidence. Keep the value method and remaining ownership visible. A source's range may include both, but that does not give each reader an immediate right to withdraw the higher amount.

Explain Timing and Market Differences

The same cumulative gain can arise over different lengths of time. In a separate example with no interim payments, $100,000 grows to $130,000 at final settlement. Over three years, the compound annual rate is about 9.14%. Over seven years, it is about 3.82%.

The cumulative gain is 30% in both cases. A table using only cumulative percentages would conceal the time difference. A table of annual rates would need a consistent calculation and the relevant dates. FINRA's December 2024 returns guide distinguishes annualized results from simply dividing a total gain by years. FINRA: Calculating Your Investment Returns.

The Birch Study's eight-program model uses the same five-year horizon to keep that issue separate. It does not claim that their cash arrived on identical dates. Its cumulative measure ignores timing within the period. An annualized comparison would require more detail than the total receipts table provides.

Separate the year a program began from the year it exited. An exit in 2026 may reflect a purchase made years earlier under different prices, loans and leases. Grouping all such exits together can be useful, but it does not create a set of investments made under the same starting conditions.

A wider spread can have several possible explanations. Different property types, loan balances, lease events, costs or holding periods may contribute. The return table alone does not establish which factor caused a particular result. Test the explanation against relevant source records.

Keep small subgroups honest. Splitting eight programs into many categories can leave only one or two observations in each. A precise subgroup average may then describe almost nothing beyond those cases. Show the count beside it and avoid treating the difference as a proven strategy effect.

The choice of dates matters for personal experience too. A strong whole-period result does not show whether payments arrived when needed. An investor who had to cover a large bill before exit could face a cash problem that a lifetime average never displays.

Use Return Dispersion in an Investor Research File

A useful research file combines a clear population with a small set of understandable measures. Keep the source list, row-level values and calculation method available. Another reader should be able to reproduce the count and see why a program was included.

For the Birch Study, a fair conclusion would report eight completed examples, a 15% mean, a 12.5% median, two negative outcomes and a −30% to 80% spread. It would identify the five-year cumulative cash basis and equal starting contributions. It would also state that every case is invented.

A real report needs the same clarity about its evidence. Attribute sponsor-reported values to the sponsor. Describe any independent work actually performed. Reading a published table and reproducing its summary is narrower than verifying each investor's cash receipts and fees.

Separate rounding from a substantive difference. Calculate from the most precise supported inputs and round the displayed answer at the end. If only rounded source values are available, say so. A small mismatch with a headline can reflect hidden precision or a different method; obtain the calculation before asserting an error.

Treat the distribution as a set of questions for the current offering. What could produce a loss? How much cash could be delayed? Which costs fall on the investor? What exit assumptions drive the estimate? A historical median does not answer those questions, but it can prevent one attractive average from ending the research too early.

FAQ

Is the Median DST Return Better Than the Average?

It answers a different question. The median locates the middle of a sorted group; the mean reflects the sum of all values. A large result can move the mean sharply while leaving the median unchanged. Report both with the population and measure, rather than treating either as a complete investment assessment.

Can a Positive Average Include Investors Who Lost Money?

Yes. Gains in some programs can offset losses elsewhere in a combined calculation. That does not repay an investor who owned only a losing program. Show the number and severity of negative outcomes, and distinguish a program-level average from an actual investor's allocation.

Does a Published Return Range Show the Typical Investor Result?

Not by itself. Endpoints do not reveal how many investors earned values between them or how much each invested. Fee terms, dates and exit forms can differ. A midpoint is a mathematical convenience, not an established median or mean of the investor accounts.

Should Active DSTs Receive a Zero in a Full-Cycle Study?

No. Zero would be an asserted final outcome. Keep active programs in the population register and identify them as unfinished. If the research uses interim estimated values, label that method separately from final cash returns and explain the value assumptions.

Can the Worst Past Return Be Used as a Downside Limit?

No. The lowest observed result is only the lowest in that defined file. A later investment can do worse. Missing programs, different market conditions and a limited history further restrict the comparison. A historical range is not a guaranteed boundary or a probability forecast.

Why Can Capital Weighting Reverse the Conclusion?

Larger allocations give some results more influence. In the separate allocation example, more money sits in the losing programs, producing a loss despite a positive equal-program mean. Name the capital measure and the covered programs. A weighted average of annual rates is not automatically a combined dated return.

What Should I Ask for Alongside an Average Return?

Ask for the included program list, dates, calculation method, fee treatment and exit-value basis. Request the median, endpoints and negative-outcome count where the data support them. Also ask which programs remain active or lack records. Those details help explain the average without turning it into a prediction.

About Sequoia Research

Sequoia Research, LLC provides independent research on DSTs and other real estate investment offerings. This educational library helps readers examine source documents, compare defined facts, and identify questions for their investment, legal, and tax professionals.

Sources