Research guide

Delaware Statutory Trust Reviews: Why Sample Size Matters in Performance Comparisons

By Sequoia ResearchPublished Read as Markdown

What Is This Article About?

This article explains how the number and selection of DST programs affect a performance review. It shows why a percentage needs a clear count, a defined group, and a list of missing or excluded records. Fictional examples reveal how a small sample or a change in the mix can alter a comparison. The focus is on which outcomes enter the analysis, rather than how to calculate an investment return.

Table of Contents

Start With the Count Behind the DST Performance Claim

A percentage is easier to read when the count appears beside it. “Nine of ten completed programs met the stated test” tells you more than “90% success.” The longer statement still needs scrutiny. Yet it identifies the number of programs, their completed status, and the fact that someone chose a test.

The denominator is the total count used as the base. In nine out of ten, it is ten. The numerator is the count that met the test, which is nine. If either count is unclear, the headline cannot be checked. Ask for both numbers before debating the second decimal place.

Then define the test. Did a program sell? Did investors receive cash at least equal to their initial equity? Did it meet a forecast after fees? Did a payment arrive on schedule? These events answer different questions. A sponsor can have a high percentage on one measure and a much lower percentage on another.

A review should also state its cutoff date. Ten completed programs as of one year-end may become twelve six months later. An added program can change the percentage even if none of the earlier results changed. A revised number is not automatically evidence that the old figure was wrong.

The SEC's performance guidance cautions against showing only favorable investments or periods. This matters before any arithmetic begins: a perfectly calculated percentage can still describe a selected group. A useful review explains how that group was chosen. SEC Performance Claims bulletin.

Treat a broad phrase such as “successful history” as an invitation to seek a definition. Do not invent one for the author. The reported record may support a narrow finding, while the headline suggests a much wider conclusion. Writing out the count and test often makes that difference clear.

Define What Counts as One Observation

One observation is one item in the statistic. For a DST program comparison, the intended item may be a distinct offering. Other tables may count properties, completed sales, investor accounts, or payment periods. Those units should not be swapped just because all can be counted.

Imagine one fictional DST owns six properties. It sells them in six separate closings. There are six property-sale events, but only one offering. A claim about the number of completed offerings cannot use all six sales as six independent program outcomes. A property-level study could use them, with that scope stated.

Investor accounts create another trap. Two hundred investors may hold interests in the same trust. Their account records can help verify cash delivery to those accounts. They do not turn one property portfolio into two hundred distinct portfolio tests. Some investors may also have different entry dates, share classes, or outside fees.

Repeated reports are useful for a different reason. Twenty-four quarterly statements from one program show a longer record than one statement. They may help trace changes and find gaps. But they remain repeated observations of that program. Calling them twenty-four completed programs changes the meaning of the evidence.

Make an identity list before building a summary. Give each offering a stable ID, its exact legal name, and the relationship to any property or prior name. Note mergers, renamed programs, and classes that refer to the same assets. A label change alone should not add a new economic outcome.

The correct unit depends on the question. A study of service responses may count inquiries. A study of capital outcomes may count completed investments. A study of property sales may count assets. State the unit in the table title so the reader knows what the count can support.

Separate the Full Program List From the Reviewed Sample

The full list is the group your question intends to cover. The reviewed sample is the part for which you performed the stated work. Sometimes those are the same. Often they differ because documents, dates, or final outcomes are missing. The review should show that difference openly.

Start by writing a selection rule. One example is: “All offerings launched by this named sponsor during the chosen years and completed in cash by the review cutoff.” This is only a sample definition, not a recommended rule for every study. It needs clear meanings for launched, sponsor, completed, and cash.

A rule should be set before you inspect which records look good. Changing it after seeing the results can alter the answer. There may be a sound reason to change the scope, but record that reason. Preserve the earlier version so readers can see what moved in and out.

Do not confuse a complete list with complete evidence. You may know the names of eight completed offerings but have enough final records for only six. The list is complete for that defined group; the performance evidence is not. A count of eight in a brochure does not fill the two missing files.

Private-offering research can face real disclosure limits. The SEC explains that private placements may provide less information than registered offerings. That supports being clear about the material actually reviewed. It does not justify treating unavailable figures as favorable or unfavorable. SEC private-placement bulletin, updated September 21, 2026.

A useful report gives three separate counts: programs on the identified list, programs eligible under the rule, and programs supported well enough for the stated calculation. When those counts differ, explain why. Readers can then judge coverage instead of having to guess it from a headline.

Work Through a Fictional Sponsor Review

Aster is an invented sponsor for this example. Its programs, records, amounts, and outcomes are fictional. The research file identifies twenty offerings by the chosen cutoff. Eight are marked completed in cash, ten remain active, and two have an unresolved status. The twenty are the identified universe for this exercise.

The review asks a narrow historical question about the eight completed programs. It checks whether total investor cash received, including earlier distributions and final proceeds, was at least equal to initial equity. All counted cash is after the modeled offering-level charges. Personal taxes and outside account costs are excluded from this chosen test.

This is a nominal cash-recovery threshold. It does not measure annual return, purchasing power, time spent holding the investment, or whether a forecast was met. Calling it “success” would hide those limits. The table therefore uses the phrase “met the stated cash threshold.”

Status in the identified list Program count Treatment in this review
Completed in cash; final records sufficient 6 Included in the threshold count
Completed in cash; final records incomplete 2 Outcome unresolved for the threshold
Active at the cutoff 10 Tracked separately; no final cash outcome assigned
Completion status unresolved 2 Kept on the list with an open status question
Total identified offerings 20 Full list for this fictional exercise

Of the six completed programs with sufficient records, four met the threshold and two did not. The observed proportion is four out of six, or about 66.7%. The evidence covers six of the eight programs known to be completed, or 75% by program count. These percentages measure different things.

A selected page showing only the four programs that met the threshold could display four out of four, or 100%. That number would describe its selection. It would not describe all six reviewed completed programs, all eight known completed programs, or all twenty identified offerings. The denominator is the reason the figures differ.

The narrow finding can be written in one clear sentence: “Four of six completed programs with sufficient final records met the defined cash threshold; two additional completed programs lacked enough evidence.” The sentence is less dramatic than a success badge. It is much easier for another person to verify.

Keep Active Programs and Missing Outcomes Separate

An active program has no final cash outcome yet. A completed program with missing records has an outcome that may exist but is unknown to the reviewer. Those are different limits. Put them in different columns, because more time may resolve one while more documents may resolve the other.

Aster's two completed but incomplete files allow a simple boundary exercise. If neither met the threshold, four of eight completed programs would meet it, or 50%. If both met it, six of eight would meet it, or 75%. These are the lowest and highest possible proportions given the stated known counts.

That 50% to 75% range is not a statistical confidence interval. It is not a forecast, and it assigns no probability to either missing outcome. It simply shows how much two unknown classifications could change this count. One of the two meeting the test would put the result between the endpoints.

The ten active programs should not be added to the denominator of a final-outcome rate as if they failed. They also should not be dropped from every discussion of the sponsor. A separate status table can show how much of the broader history remains unfinished. That is useful context without inventing results.

The two programs with unclear completion status need another label. Perhaps a final sale occurred but a payment remains pending; perhaps the reviewed records are old. Those are possible explanations to investigate, not findings to assume. The review should preserve the question until a source resolves it.

A changing sample can also create a distorted view if records disappear selectively. In its survey-method discussion, the Census Bureau explains that losing participants may bias results when those who leave differ from those who remain. This is a general data-quality lesson, not evidence about DST sponsors. Keep track of which records leave your file and why. Census Bureau, Nonsampling Error.

See How One Result Moves a Small Sample

A small denominator gives each program more influence. In a fixed group of five programs, one program represents 20 percentage points. In a group of fifty, one represents two points. This is simple counting. It does not tell you whether either group was fairly chosen or properly researched.

Suppose four of five programs met a test. The reported proportion is 80%. If one of those four was misclassified and should not have met it, the corrected count becomes three of five, or 60%. One record changes the percentage by twenty points because the group is small.

Now use a separate group of fifty. Forty meet the test, again 80%. Reclassifying one favorable record makes the result thirty-nine of fifty, or 78%. The shift is smaller. Both corrections matter, but readers should understand why a small record set can move sharply after a single update.

Adding a new program is different from correcting an existing one. Starting from four of five, a sixth program that meets the test gives five of six, or about 83.3%. A sixth that does not gives four of six, or about 66.7%. In each case the denominator grows as well.

Show raw counts beside rounded percentages. Rounding four of six to 67% can be fine for readability if the count remains visible. Reporting 66.666667% adds digits without adding evidence. The uncertainty often comes from the record selection, definitions, or missing facts, rather than the arithmetic precision.

There is no program count in this article that turns a record into a reliable forecast. Ten, thirty, or one hundred observations can all be useful for some questions and weak for others. The next task is to ask whether the observations match the question and cover the relevant conditions.

Compare Program Counts With Equity Coverage

Program-count coverage tells you how many offerings were reviewed. Equity coverage tells you how much of the defined initial capital those offerings represent. A large program and a small one each count as one program, but they do not represent the same amount of invested money.

Return to the fictional Aster file. The six reviewed completed programs represent $100 million of initial equity. The two completed programs with incomplete final records represent another $50 million. The review covers six of eight programs, or 75% by count, but only $100 million of $150 million, or about 66.7% by initial equity.

Both coverage figures are useful. Neither should replace the other without explanation. Count coverage shows the breadth of the program record. Equity coverage shows how much of the defined initial capital is represented. Neither measures how many investors were helped or harmed, which would require account-level evidence and its own rules.

The threshold result can also look different by weight. Suppose the four Aster programs that met it each began with $10 million of equity. The two that did not began with $20 million and $40 million. Four of six programs met the test, but those four account for only $40 million of the reviewed $100 million.

That is a 66.7% share by program count and a 40% share by initial equity. The latter is a weighted share of capital in programs that met the test. It is not a 40% return, a loss rate, or the fraction of dollars ultimately recovered. The full cash records are needed for those other questions.

Choose weights to match the question and preserve the underlying amounts. Weighting by initial equity differs from weighting by current value, property count, or investor count. A label such as “weighted performance” is incomplete until the reader knows which weights were used and how they were applied.

Control for Program Age and Market Period

Programs need time to reach later stages. Comparing the number of completed exits for an older group with a recently launched group may mostly compare age. A new program cannot provide a ten-year observed history. An active status alone does not establish that it is delayed or performing poorly.

Group programs by a relevant starting period, often called a vintage. Define the date: property purchase, first investor close, or final subscription close may differ. Also identify how long each program has been observed. A review of two sponsors should use the same date definition on both sides.

A mixed group can hide important patterns. Consider the separate fictional table below. Every entry describes completed programs assessed under the same hypothetical threshold. “Earlier” and “later” are defined purchase-period groups, not claims about any actual market cycle. The numbers demonstrate how the group mix can change an overall comparison.

Fictional sample Earlier group Later group All included programs
Sponsor A 4 of 5: 80% 18 of 30: 60% 22 of 35: about 62.9%
Sponsor B 21 of 30: 70% 2 of 5: 40% 23 of 35: about 65.7%

Sponsor A has a higher observed proportion in each period. Yet Sponsor B has the higher combined proportion. B has far more programs in the earlier group, where both sponsors show higher proportions. The overall figures give different weights to the two periods. No underlying program result changed.

This does not prove A is the stronger sponsor. The period groups may still differ in leverage, property type, tenant mix, fees, or other facts. It shows why the combined percentage alone cannot settle the comparison. Report the group counts, then investigate whether the groups are meaningfully comparable.

Do not force a fair-looking comparison by deleting awkward periods without explanation. You can show a matched-period view alongside the full disclosed history. State which question each view answers. A shorter common period may improve comparability while excluding relevant experience outside that window.

Watch for Shared Assets and Repeated Evidence

Separate program names do not always mean fully separate economic exposures. Programs can share tenants, nearby markets, financing conditions, or parts of a property portfolio. A common shock can affect several of them at once. A review should describe those links instead of treating the names as proof of independence.

For example, ten offerings tied to one major tenant provide ten program records. They may show differences in lease terms and debt. They also share a tenant-related exposure. A history without that tenant's distress cannot show how each would behave during such distress. The number ten does not supply the missing event.

A reviewer can map these links without inventing an adjusted sample size. List common tenants, property overlap, shared loans, and timing clusters where the documents support them. Say which links were checked. Do not claim that ten offerings are “really three” unless a defined method supports that conclusion.

Copied evidence creates a separate counting problem. A sponsor announcement reproduced on five websites remains one originating claim. Five pages may make it easier to locate, but they do not provide five independent checks of the result. Follow each page to the original record before counting sources as corroboration.

The same issue appears when several research reports rely on one unaudited table. Their conclusions may differ, but the underlying numbers can have the same origin. Keep the source map beside the program map. This helps distinguish several independent analyses from several independent sets of outcome records.

FINRA's private-placement guidance discusses evaluating third-party research rather than relying on it blindly. That is member-firm guidance, not a rule assigning every blog reader the same duties. For a review reader, it supports a practical question: what did each source actually check? FINRA Regulatory Notice 23-08.

Read Averages Alongside the Spread of Outcomes

An average can hide whether results cluster closely or vary widely. Before comparing two averages, ask to see the program-level values, their definitions, and the count. A single high result can affect the mean, which is the sum divided by the number of observations. The median is the middle value after sorting, or the average of the two middle values when the count is even.

Neither statistic automatically describes what a new investor should expect. A simple mean gives each observation equal weight; a weighted mean uses stated weights. The median gives the middle position. Both depend on which observations were included. Neither repairs missing records, incompatible metrics, or a sample chosen because its results look attractive.

Ask for the range as well: the lowest and highest observed values. A range is sensitive to unusual cases, but those cases may be important. Removing them just because they are large or small can hide risk. First check for a data error; then decide whether the record belongs under the written rule.

Do not average figures that answer different questions. A cash-distribution rate, a final equity multiple, and a projected return are not interchangeable measures. A larger table of mixed measures is still an invalid comparison. Settle the definition before choosing the summary statistic, even if the arithmetic software will accept every entry.

When the data are suitable for a statistical estimate, both count and spread affect its precision. NIST's discussion of confidence limits for a mean makes that relationship explicit. This supports reporting variability alongside sample size. It does not make a selected list of DST outcomes a random sample of future offerings. NIST, Confidence Limits for the Mean.

For a small review, a clear table may serve readers better than a complex summary. Show each eligible program, the same metric, the observation period, and any unresolved record issue. The reader can see patterns and exceptions that a single average conceals.

Use Confidence Language With Care

“High confidence” can mean several things. A reviewer may be confident that a number was transcribed correctly, that a historical list is complete, or that an estimate will describe future outcomes. Those are different claims. A report should say which one it is making and what supports it.

A statistical confidence interval is a range built under a stated method and assumptions. It is not a quality badge. NIST explains that a 95% confidence level describes the long-run coverage of the method across repeated samples. It is not a statement that a particular calculated interval has a 95% chance of containing the fixed true mean. NIST confidence-limit explanation.

Before accepting an interval in a DST review, ask what population is being estimated and how the observations were selected. Ask whether the method addresses shared exposures and incomplete outcomes. A neat range cannot answer those questions on its own. Random-sample formulas require more than a count typed into a calculator.

Small counts also call for care in the calculation method. NIST discusses why common symmetric approximations for proportions can be poor when the sample or number of observed failures is very small. That is a reason to seek qualified statistical work when an interval matters. It is not a reason to add one automatically to every review. NIST, Confidence Intervals for Proportions.

“No observed failures” should therefore remain a description of the reviewed record. It does not prove that failure is impossible. The definition of failure, sample selection, observation time, and outcome evidence still matter. A clean history may be useful while providing limited evidence about a condition that never occurred during it.

Use plain descriptions when formal inference is not supported. “All four reviewed cases met this test” is precise. “The next case is almost certain to meet it” is a new claim that needs separate support. The first sentence does not establish the second, however confidently it is presented.

Test Exclusions Before Trusting a Ranking

An exclusion can be valid. A study of completed cash outcomes may leave active programs out of its final-outcome statistic. A study of one property type may exclude other sectors. What matters is whether the rule fits the question, is applied consistently, and leaves an understandable record of the omissions.

Prepare an exclusion table with the program ID, reason, source, and effect on the count. “Outside launch-date range” is a clearer reason than “not relevant.” “Final payment record unavailable” is clearer than “incomplete.” Precise labels help a second reviewer tell a scope choice from an evidence gap.

Then test plausible alternative rules. Does the broad conclusion change if an eligible but disputed record is included? Does it depend on removing the largest program? Does it disappear when both sponsors use the same observation window? These checks reveal sensitivity; they do not authorize selecting whichever rule gives the preferred answer.

Keep an invalid record out even if including it would make the result look more balanced. A duplicated program is still a duplicate. A forecast is still a forecast. Fairness means applying sound definitions consistently, not adding unsupported rows until a table reaches an appealing number.

The SEC's investment-adviser marketing guide addresses selected performance in a specific regulatory setting. It describes restrictions on presenting fewer than all substantially similar portfolios, subject to exceptions, and conditions for extracted performance. The guide covers ads from advisers registered, or required to register, with the SEC. These provisions do not establish that every incomplete public DST review violates a rule. SEC Investment Adviser Marketing guide.

For an ordinary reader, the practical question remains straightforward: can someone explain why each included record belongs and each excluded record does not? A ranking built on an unexplained selection should be treated as a claim with an unresolved method, even when the calculation itself is correct.

Write a Sample Note That a Reader Can Audit

A useful sample note sits beside the statistic. It gives the reader enough detail to understand the group without opening every source first. Put the full register behind it for those who need to verify identities, cash records, dates, or exceptions.

For the Aster exercise, the note could say: “We identified twenty offerings. Eight were marked completed in cash by the cutoff. Six had sufficient final records for the defined nominal cash threshold. Four of those six met it. Two completed outcomes remained unresolved; ten programs were active and two had unclear completion status.”

Add the coverage basis if it matters: “The six reviewed completed programs represent 75% of known completed programs and about 66.7% of their initial equity.” Then state the threshold's exclusions, including time value and personal taxes. This keeps the reader from treating a cash-recovery count as a total-return result.

The register should preserve the source for each outcome and each classification. Include a date checked, the version used, and any question still open. If the same source supports several programs, say so. The goal is a trail that another reviewer can follow, not an impressive number of citations.

Assign unresolved items to the next research step. One may need a final distribution statement; another may need the complete offering list; a third may need clarification of a sale versus a transfer. Different questions require different records. Asking for “more performance information” may produce another headline instead of the missing detail.

Update the Comparison Without Changing Its Meaning

Keep each published or shared snapshot with its cutoff and rules. When new records arrive, create an update that explains the change. This lets a reader separate newly completed programs, corrected figures, newly obtained evidence, and a revised selection method.

Suppose Aster's two incomplete completed files later become available. One meets the threshold and one does not. The completed-group count becomes five of eight, or 62.5%. Evidence coverage reaches eight of eight, or 100%. The lower observed proportion does not mean an earlier program performed worse; the record became more complete.

If an active program later completes, that is a different update. The completed-group denominator grows. Keep its outcome date separate from the date you received the documents. A later review can include a newly documented old event without implying that the event itself happened during the review month.

A method change deserves its own explanation. Switching from program counts to initial-equity weights can change the headline sharply. So can widening the launch years or adding another property type. Show the old and new definitions rather than presenting the result as a simple continuation of the same series.

End the review with the finding the evidence can support. A sample may document a narrow historical pattern, reveal a reporting gap, or show that a ranking is sensitive to selection. Each is useful. None requires turning limited past observations into a promise about the next DST offering.

Frequently Asked Questions

How Many Completed DST Programs Make a Review Reliable?

There is no universal count. Reliability depends on the question, selection rule, source quality, observation time, and how different the results are. A larger count can add useful history, but it cannot correct a list that omits relevant outcomes or mixes incompatible measures. Ask to see the defined group and its coverage.

Should Active Programs Count as Failed Investments?

No. An active program has no final outcome yet under a completed-outcome test. Track it separately, with its age and current status. Excluding it from that narrow statistic does not mean it should disappear from the sponsor's wider history or from a review of ongoing operating problems.

What Is the Difference Between Sample Coverage and Performance?

Coverage measures how much of a defined group has sufficient evidence for the analysis. Performance measures the stated outcome within the analyzed records. Reviewing six of eight programs gives 75% count coverage. If four of the six meet a threshold, the observed proportion is about 66.7%. The two percentages answer different questions.

Is an Equity-Weighted Result Better Than a Program Count?

It serves a different purpose. Initial-equity weights show how much starting capital is represented by each program. Equal program weights describe the program count. Neither automatically gives the average investor's experience or the return of the combined capital. State the chosen weights and retain the unweighted count for context.

Does Zero Observed Loss Mean the Next Offering Is Safe?

No. It describes the reviewed cases under a particular loss definition and cutoff. The sample may be small, selected, unfinished, or exposed to similar conditions. It may not include a future stress that matters. Check the next offering's own terms and risks instead of converting the historical count into a guarantee.

Why Can an Overall Ranking Reverse Within Separate Periods?

The two samples may contain different proportions of programs from each period. One sponsor can have more programs in the period where both show stronger outcomes. Its combined percentage can then be higher even while its percentage is lower within each period. Show group counts before drawing a comparative conclusion.

What Should a Review Disclose When Some Records Are Missing?

State which programs are missing, what evidence is absent, and how those gaps affect the calculation. Keep unknown outcomes separate from zero or failure. If useful, show the possible count range under explicit assumptions, while labeling it as a boundary exercise. Do not present that range as a confidence interval or forecast.

About Sequoia Research

Sequoia Research, LLC provides independent research on DSTs and other real estate investment offerings. This educational library helps readers examine the sources, methods, and limits behind investment claims.

Sources