Research guide

Top DST Sponsors: How to Account for Missing Data in a Ranking

By Sequoia ResearchPublished Read as Markdown

What Is This Article About?

A list of top DST sponsors can change when the reviewer changes how missing facts are treated. This article shows how to keep those gaps visible without turning them into scores. It compares several research methods using one invented set of sponsor files. It also explains when a ranking should be withheld and what a useful unfinished comparison can still tell you.

Table of Contents

Why Missing Data Changes a DST Sponsor Ranking

A missing fact is part of the research result. It tells the reader where the evidence stops. If a comparison hides that boundary, a neat table may suggest a level of knowledge that the reviewer does not have. The trouble starts before anyone argues about which sponsor belongs first.

DST means Delaware statutory trust in this article. A sponsor comparison may cover a firm's people, past programs, fee records, reporting, or other defined features. Those are different research questions. A missing report about one program does not make every fact about its sponsor unknown. Likewise, a full biography does not fill a gap in the program's results.

Consider two files. One has a detailed fee schedule that reveals a charge the reviewer dislikes. The other has no usable schedule. If the second file receives a favorable score because no charge was found, limited access becomes an advantage. If it receives zero, absence becomes a claim of poor performance. Neither treatment describes what the reviewer learned.

The U.S. Securities and Exchange Commission, or SEC, explains that private placements may provide less information than registered offerings. Its bulletin was updated on September 21, 2026. A public search may therefore leave questions that require records beyond the pages inspected. This does not establish that a particular sponsor withheld a required document. SEC private-placement bulletin

A sound comparison names its evidence boundary. State which entities, programs, dates, and records were reviewed. Then explain what could not be established within that work. A reader can assess a narrow, honest result. They cannot recover omitted limits from a confident headline.

This article uses an original scoring exercise to expose those choices. The numbers are research-method examples, not actual sponsor scores, credit ratings, or investment recommendations. No real firm is ranked. The goal is to make the effect of missing evidence easy to trace.

Separate Unknown Facts From Zero and Not Applicable

Start with labels that describe the state of the fact. A verified zero belongs to a different category from a blank. If a defined charge is confirmed as zero in the relevant schedule, record zero with its source and scope. If the schedule is unavailable, record unknown. A search that found nothing does not turn an unknown fee into a free service.

“Not applicable” needs its own basis. Suppose a comparison asks about a fixed loan's next interest-rate reset. A reviewed fixed-rate loan may have no such reset within its stated term. A program with no loan may also make that question inapplicable. Those conclusions require the right debt records. They cannot be inferred from an empty debt field.

Separate a fact's value from the reviewer's progress. “Not yet checked” means the work is open. “Checked; source does not answer the question” means a review occurred but did not resolve it. “Conflicting sources” means two or more records require reconciliation. Each state calls for different next steps.

A document may exist without being available to the reviewer. It might sit in a private portal, fall outside the research scope, or require a request that has not been answered. Record only the explanation supported by the file. Do not invent a privacy restriction or refusal to make the gap seem less awkward.

Use dates with these labels. A 2024 document may establish a 2024 fact while leaving the same question open for a 2026 comparison. It remains useful historical evidence. Moving it into a current column without checking whether it still applies silently changes what the source supports.

For an internal worksheet, five basic states often suffice: supported value, supported zero, supported inapplicability, unresolved, and conflicting. Add a separate work-status field for requested, received, or reviewed. This avoids a common mistake: marking a task complete and then assuming its factual question must also be settled.

The labels are a proposed research practice, not a regulator's official rating system. Write their meanings beside the table. Even careful terms can confuse readers if one analyst uses “unavailable” for a missing source and another uses it for an unverified claim.

Build a Missing-Data Register Before Scoring

A missing-data register is a short record of each unanswered question. It should say what is missing, why the question matters, where the reviewer looked, and what would resolve it. This gives the gap a defined shape. “Need more research” is too broad to guide the next person.

Use one row per question. For example: “Need the investor-level cash proceeds for Cedar Program Three's final sale, including the reporting date and treatment of withheld funds.” That is more useful than “Track record incomplete.” The narrower wording makes it possible to judge whether a later reply actually answers the request.

Record the exact subject. A sponsor group, a management company, an issuer, and a property owner may have related names. A response about one does not automatically apply to another. A missing fact can appear resolved if the wrong entity's document is attached to the right row.

Include a source location, such as a page, table, section, or dated message. A long report may mention sale proceeds at several levels. The property sale price, program cash balance, and cash sent to investors are not interchangeable. The register should show which one was needed and which one was found.

Assign an owner and a next action, but keep the action separate from the conclusion. “Ask the research contact for the final statement” is a task. “Final cash remains unverified” is the current finding. A request being sent changes the task's state; it does not change the finding.

Also record whether the missing fact blocks the intended comparison. Some gaps prevent any reasonable score on a required feature. Others limit a footnote or an optional detail. Make this choice using the question the comparison is meant to answer. Avoid declaring every gap minor simply because a publication deadline is near.

The Financial Industry Regulatory Authority, or FINRA, tells covered firms to investigate private-placement claims with care. Its May 2023 notice discusses primary documents, independent analysis, misleading selection of past results, and follow-up on concerns. A spreadsheet does not establish that a firm met those duties. The limited point here is that a source gap deserves investigation rather than a guessed value. FINRA Regulatory Notice 23-08

A Fictional Sponsor Comparison With Uneven Evidence

Assume an analyst defines four research features. They concern the completeness of a program inventory, fee records, reporting examples, and team records. Each uses a written zero-to-100 rubric. Higher values mean a closer fit with that rubric. They do not mean higher expected returns or less investment risk.

The analyst gives the four features weights of 40%, 30%, 20%, and 10%. These invented weights reflect the exercise's chosen priorities. No regulator or rating agency supplied them. Different weights could answer a different research question, even if all source records stayed the same.

Three fictional profiles have the following entered values. A dash means unresolved evidence and has no numerical value. The names Cedar, Maple, and Spruce refer only to this example; they are not disguised reviews of actual firms.

Research feature Weight Cedar Maple Spruce
Program inventory records 40% 80 70 90
Fee records 30% 60 Unresolved 50
Reporting examples 20% 90 80 Unresolved
Team records 10% 70 80 60

All four cells are supported for Cedar within this invented file. Maple lacks support for the fee-record feature. Spruce lacks support for the reporting feature. Ten of twelve cells have support, or about 83.3% by simple count. That count says nothing about whether the missing cells are important.

Weighted coverage tells another part of the story. Cedar has support for 100% of the original feature weight, Maple for 70%, and Spruce for 80%. These are evidence-coverage figures under the selected method. They are not probabilities that the profiles are accurate, safe, or suitable.

The case is deliberately small so the reader can check every calculation. Real research may involve many more records and layered entities. The same need remains: show the gap and the denominator before showing a rank. A larger table does not fix an unclear method.

Method One: Keep Only Complete Sponsor Profiles

The complete-profile method preserves all four features and their original weights. A profile gets an overall score only when every required feature has usable support. The method keeps the question stable, but it can leave few profiles available for comparison.

Cedar's score is 75. Its four contributions are 32 from program inventory, 18 from fee records, 18 from reporting, and seven from team records. They sum to 75. Maple and Spruce receive no aggregate score under this method because each has an unresolved active feature.

Cedar should not be called the top sponsor just because it is the only scored row. The file has one complete profile, not evidence that this profile outranks the other two. A suitable display reads “75 under the stated rubric; comparative rank withheld.” Keep both excluded profiles visible below it.

This method works when each feature is central to the research question. For example, if the comparison promises to examine fee documentation, the analyst cannot drop that feature and still claim to have answered the whole question. Waiting for support may be more useful than forcing a complete-looking ranking.

The cost is reduced coverage of the chosen group. Complete records may be easier to obtain for familiar firms, older programs, or participants that answered quickly. The method does not correct this access bias. Report the number included and the number withheld so readers can see how narrow the result is.

A complete file also is not necessarily a strong file. All four scores could be low, supported by current records. Conversely, a profile withheld for one gap might later score well. Completeness measures whether the required cells can be used; it does not decide the result those cells will produce.

If a second complete profile later becomes available, the comparison can expand. Preserve the first version and explain the new evidence. Do not rewrite the original page to suggest that two profiles had always been scored. The history matters when someone wants to understand why a rank first appeared.

Method Two: Compare Only Shared Evidence

The shared-evidence method keeps all three profiles but uses only features supported for every one of them. In the example, program inventory and team records qualify. Fee records and reporting are removed from every profile's scored calculation. They remain visible as excluded features.

The retained features account for 50% of the original weight: 40% plus 10%. Within the reduced model, their weights become 80% and 20%. The new weights sum to 100%, but the model still retains only half of the original planned weight. Both numbers belong beside the result.

Cedar scores 78, calculated as 80 times 0.8 plus 70 times 0.2. Maple scores 72, from 70 times 0.8 plus 80 times 0.2. Spruce scores 84, from 90 times 0.8 plus 60 times 0.2. Spruce leads only this two-feature exercise.

That result cannot support the headline “Spruce Is the Best Overall Sponsor.” The calculation excludes both fee records and reporting. It answers a narrower question about two defined kinds of evidence. A title and summary should make that scope clear before a reader sees the ordered table.

Cedar's score rises from 75 in the full model to 78 in the reduced one. No underlying Cedar fact improved. The model changed. Comparing those two numbers as if they showed progress would confuse a change in method with a change in evidence or behavior.

This approach is useful when the retained features form a meaningful question on their own. It is weak when the excluded features contain the reason for doing the research. A two-feature model may be easy to calculate yet poor at meeting the reader's need. The analyst must explain the choice, not merely display its arithmetic.

Avoid letting one new profile silently redefine an established model. Adding a fourth profile with missing team records could remove that feature from the common set. Every old score would then change. Record the participant list and feature list together so the reader can trace why a result moved.

Method Three: Show Separate Findings Without a Ranking

A comparison does not need an aggregate score to be useful. The separate-findings method shows each feature, source, value, and gap without combining them into a single rank. It preserves distinctions that a weighted total would compress.

In the fictional case, the reader can see that Cedar has support for all four features. Maple's fee-record evidence remains open. Spruce's reporting evidence remains open. The reader can also see the supported values in each row. No overall winner is declared from that pattern.

This is often the clearest option when the features measure different things. Team history, cost records, and report access need not be converted into one scale for an investor to ask better questions. Some tradeoffs require judgment about the reader's needs, not more decimal places.

Separate findings can still include a focused comparison. For example, the analyst might describe which files contain dated reporting examples under a fixed rule. That is a comparison of that feature. It should not expand into a broad conclusion about service quality unless the evidence actually supports that claim.

Show the status of each question in plain language. “Fee schedule requested; no usable copy reviewed by the cutoff” is clearer than a gray box with a question mark. A reader who prints the table without its color key should still understand the gap.

Keep the order of profiles neutral where practical, such as alphabetical order. If the first position is just a display choice, say so. A numbered list can look like a ranking even when the surrounding text says there is none. Design choices should reinforce the meaning of the evidence.

The research output can end with an action list: obtain the missing fee record, check the reporting sample's scope, and confirm the inventory period. That is a useful result. An unfinished score does not make the work worthless when the remaining questions are precise and traceable.

Use Ranges to Show What Remains Unresolved

A range can illustrate how much an unknown score could affect the model without filling the gap. It is a sensitivity exercise: calculate the extremes allowed by the rubric, keeping all supported values fixed. The endpoints are possible mathematical limits, not predictions about the missing evidence.

Maple's supported contributions total 52: 28 from inventory, 16 from reporting, and eight from team records. Its missing fee-record score carries 30% weight. If that unknown score could run from zero to 100, Maple's full-model total could run from 52 to 82. The actual score remains unknown.

Spruce's supported contributions total 57: 36 from inventory, 15 from fees, and six from team records. Its missing reporting score carries 20% weight. The full-model total could run from 57 to 77 under the same zero-to-100 limits.

Cedar's known 75 lies within both ranges. Therefore, the entered evidence does not establish that Cedar must lead either incomplete profile in the full model. It also does not establish that Maple or Spruce will lead. The ranges show unresolved order, not a forecast of who will win.

A midpoint would be another assumption. Maple's range midpoint is 67, but there is no evidence here that 67 is more likely than any other allowed result. Presenting it as an expected score would create knowledge the file does not contain. Keep the bounds separate from any supported estimate.

Sometimes further research narrows the allowed range. If reliable evidence limits Maple's fee-record score to 40 through 60 under the rubric, its total would fall between 64 and 70. Cedar's 75 would exceed that entire interval within this model. Explain the new evidence that justifies the tighter bounds; do not narrow them for convenience.

Even a clear ordering of modeled scores remains limited by the rubric, weights, dates, and selected profiles. A robust numerical order does not prove better investment outcomes. The range analysis addresses uncertainty in a score, not uncertainty in every economic event that could affect an investor.

Why Filling Gaps With Averages Can Mislead

Replacing unknown values with an average is tempting because it makes a table easy to finish. In this example, the two known fee-record scores are 60 and 50. Their average is 55. Assigning 55 to Maple would produce a full-model score of 68.5.

The two known reporting scores are 90 and 80. Their average is 85. Assigning 85 to Spruce would produce 74. The resulting order would be Cedar at 75, Spruce at 74, and Maple at 68.5. That order relies on two invented substitutions, even though the arithmetic is correct.

There is no reason in the entered evidence to assume that the missing files resemble the available ones. The gap may reflect access, timing, a different record format, or something else not established. The average of two other profiles does not answer a question about Maple or Spruce.

A clearly labeled scenario may use assumed values to test the model. That is different from placing those values into the evidence table as facts. Keep the original blanks intact, name the assumption, and show how the scenario changes the outcome. Do not let an exported chart lose those labels.

Another shortcut is to score each profile only on its own known features and stretch that subtotal to a full scale. Maple would become about 74.3, using 52 divided by 0.7. Spruce would become 71.25, using 57 divided by 0.8. Cedar stays at 75.

Those numbers use different feature sets. Maple's score ignores fee records, while Spruce's ignores reporting. Placing them in one ranked column makes unlike calculations look alike. The shared-evidence method avoids this particular problem by excluding the same features for everyone, though it has its own limits.

The simplest question for a reviewer is: “Which numbers in this table came from evidence, and which came from a rule for filling gaps?” If the answer requires reverse-engineering a formula, the display needs work. A reader should be able to identify an assumption before relying on the result.

Check Missing Track Records and Uneven Time Periods

Missing results deserve special care because completed programs can be easier to describe than ongoing ones. Start with a list of the programs that belong in the chosen scope. Then record which are completed, ongoing, transferred, or unresolved under defined terms. Do not begin with only the outcomes that were easy to find.

Imagine an original research set of 20 programs. Eight have completed cash exits, ten remain open, and two have unresolved status. The analyst has full outcome records for six of the eight completed exits. This supports 75% coverage of completed-exit records, calculated as six divided by eight. It does not support a 75% success rate.

The same six records cover 30% of the 20-program set by count. That is another denominator, answering another question. Neither figure is weighted by invested capital. Keep counts, capital amounts, and investor accounts in separate columns unless the method clearly connects them.

Ongoing programs should not be assigned zero realized returns just because they have not exited. They may have distributions and current estimates, but those data answer different questions from a completed investor cash outcome. Preserve the categories instead of forcing all rows into a single exit-return average.

The SEC's performance-claims bulletin asks readers to examine how results are calculated, what fees are included, and which periods and conditions the figures cover. Its September 2022 guidance remains relevant to the method here; it is not a newly issued 2026 rule. SEC performance-claims bulletin

A gap may also concern time rather than a missing document. A program purchased in one year and sold in another cannot be treated as the same holding period as a younger open program. Record start date, end date or cutoff, and status. A shared sponsor name does not make periods comparable.

If a record contains a forecast, label it as a forecast even when actual results are missing. Replacing an unknown realized result with the original target quietly changes the measure. A complete-looking column assembled from targets, appraisals, and cash outcomes is less useful than separate, correctly labeled columns with visible blanks.

Decide Which Missing Facts Block the Comparison

Not all missing facts have the same effect. Decide in advance which facts are required to answer the main question. A comparison of current fee structures may be blocked by a missing fee schedule. A comparison of whether selected public pages identify their authors may proceed without that schedule, because it answers a narrower question.

Write the rule before inspecting the resulting order where possible. Otherwise, there is a risk of choosing a missing-data policy because it produces a preferred leader. A change in policy can be justified, but the explanation should describe the research need rather than celebrate the new result.

Consider a proposed rule that requires at least 80% weighted evidence coverage. Cedar qualifies at 100%, Spruce at 80%, and Maple does not at 70%. Yet Spruce still lacks reporting evidence. If reporting is essential to the stated purpose, reaching 80% does not solve that problem. A threshold cannot replace the reason for the comparison.

Use both a coverage rule and explicit required features when appropriate. Explain what each does. The coverage rule controls the quantity of supported weight; the required-feature rule protects a specific part of the question. Neither rule proves the underlying sources are reliable without source review.

Avoid treating a missed deadline as a negative answer. If a research request is unanswered by the cutoff, record that status. A late reply may affect when the article can be updated. It does not by itself show poor investment performance, misconduct, or refusal to cooperate.

At the same time, a reader may decide that an unresolved material question prevents a decision now. That is a limit on the decision supported by the file. It need not become an accusation about the firm. Careful language can state both facts: the answer is not established, and the reader's next step remains blocked.

For hands-on comparison of complete profiles, shared features, and disclosure-only output, use the Missing-Data Treatment Explorer. It makes the policy choice visible. User-entered scores and source confirmations still require judgment; a functioning calculator cannot supply missing evidence.

Publish the Limits and Update the Right Version

Put essential limits near the result, not only in a distant methodology page. The reader should see the sample, cutoff, retained features, withheld profiles, and meaning of the score in the same area as the ranking. Someone who reads only that section should not mistake a narrow exercise for a market-wide conclusion.

A useful result statement for the shared-evidence example is: “Spruce leads this fictional three-profile comparison on inventory and team records. Those features retain half the original planned weight. Fee and reporting evidence are excluded from all scores.” This is less sweeping than a best-sponsor claim and much closer to the work performed.

When missing data arrive, save the original source and record its date. Check the exact entity, period, and definition before changing the cell. A new document can answer part of the question while leaving another part open. Receiving a file is not the same as resolving every gap linked to it.

Suppose Maple's fee-record score is later supported at 40 and Spruce's reporting score at 100. Maple's full score becomes 64, and Spruce's becomes 77. Cedar remains 75. The full-model order is now Spruce, Cedar, Maple. These are new supported inputs within the invented exercise, not proof that the earlier unknowns should have been guessed.

The shared-evidence version and the later full version use different features. Preserve both with clear dates and labels. Explain which gaps closed and which model now applies. Avoid describing the change as a simple rise or fall in sponsor quality when it largely reflects better coverage.

Finish with a record that another person could reproduce: selected profiles, source locations, scoring rules, weights, gap policy, calculations, and unresolved questions. If the file cannot support a rank, say so plainly. The value of sponsor research lies in the claims it can defend, including an honest statement that a comparison is still incomplete.

FAQ

Should a missing sponsor score be set to zero?

No. Zero is a numerical result under a defined rule. A missing score means the needed value is not established. Keep it blank or explicitly unresolved, with the reason and next step. A verified zero can be used when the evidence actually supports it. The distinction prevents limited access from becoming an unsupported negative finding.

Is the sponsor with the most complete file automatically best?

No. Completeness concerns how much of the chosen evidence is available and usable. It does not establish investment quality or expected performance. In the fictional case, Cedar is initially the only complete profile. That supports a score under the stated rubric, but no comparative first-place claim over the two incomplete profiles.

Can I compare sponsors using only the factors they share?

Yes, if the shared factors answer a meaningful and clearly stated question. Apply the same retained factors to every profile and show what was excluded. Disclose both the original weight retained and the newly scaled weights. If an excluded factor is central to the promised comparison, withhold that broader conclusion rather than imply the reduced model covers it.

Does an unanswered request prove that a sponsor is hiding information?

No. The research record may establish only that a specified request remained unanswered by a stated date. Do not infer a reason without evidence. A material unanswered question can still prevent a decision. Record the request, scope, timing, and missing answer, then explain the practical limit without adding an unsupported claim about motive.

Is a score range the same as a forecast?

No. A range in this article shows the mathematical effect of allowing an unknown input to vary within stated limits. It assigns no probabilities. Maple's 52-to-82 range does not predict a result near its midpoint. A tighter range needs a defensible basis, and any ordering remains limited to the chosen rubric and evidence.

How should ongoing DST programs appear in a track-record review?

Keep ongoing programs separate from completed cash exits and other outcomes. Record their status, period, and available information. Do not turn an absent exit result into zero or replace it with a target. In the example, six usable completed-exit files cover 75% of eight completed exits and 30% of the 20-program set; neither percentage is a success rate.

What should change when missing evidence becomes available?

Verify the new source against the exact question, then update the relevant cell and calculations. Preserve the earlier version and explain whether the model, sample, or evidence changed. A revised rank may reflect newly supported inputs rather than a change in the firm's behavior. Keep any remaining gaps visible after the update.

About Sequoia Research

Sequoia Research, LLC provides independent research on DSTs and other real estate investment offerings. This library explains how to inspect investment claims, compare evidence, and recognize limits in a research file. Its educational examples do not replace review of a specific offering or professional advice for an investor's circumstances.

Sources