Biotech research works best when each investment claim is tied to a document that can support it. A scientific paper can help assess a drug’s biological rationale. A trial registry can show the planned study population and outcome measures. Financial filings can reveal whether the company has the funding and contractual rights needed to turn development progress into shareholder value. None of these sources answers every question. Read them together, keeping their dates, limitations and connections visible.
The practical task is to build an evidence trail from the drug candidate to the company’s economics. That means matching development codes across documents, distinguishing planned analyses from later interpretations, and checking whether the cash position still supports the published timetable. Within a broader biotech investing framework, this process helps separate an attractive scientific story from an investment case that survives scrutiny. The method below focuses on healthcare biotech research, with particular attention to publicly traded drug developers and U.S. disclosure sources.
Build a source map before reading the company story
Start with a narrow research question. “Is this a good biotech company?” is too broad to investigate efficiently. “Does the lead trial test the patient population used in the sales forecast?” gives the work a useful boundary. Write down the claim, identify the document needed to test it, then record what remains unresolved. A company presentation can help locate assets and milestones, but treat its statements as research leads. Your notes should make it possible to move from a claim to the underlying evidence without repeating the entire search.
| Research question | Starting document | Follow-up check |
|---|---|---|
| What supports the drug’s biological rationale? | Original research paper and supplementary material | Does the evidence concern this candidate, a related molecule, or only the target? |
| What was the trial designed to establish? | Registry entry, protocol and statistical analysis plan | Which version applied before the relevant analysis? |
| What did the study actually find? | Full results paper and available results tables | Do populations, denominators and data cutoffs match? |
| Can the company fund the next milestone? | Financial statements, notes and subsequent disclosures | What spending or financing occurred after the reporting date? |
| Who receives the commercial proceeds? | Licensing agreement and related disclosures | Which territories, costs and payment obligations belong to each party? |
| What has the regulator decided? | Approval letter, product label and available review documents | Does the decision cover the indication assumed in the investment case? |
Keep a separate identity record for each asset. Include its current name, earlier development codes, sponsor, licensed partners and trial identifiers. Add the disease and treatment setting under investigation. This small administrative step prevents a large analytical mistake: combining evidence from different compounds or different studies because their names look related. Record the document’s publication date and the period it describes as separate fields. A document published recently may describe experiments, patients or finances from much earlier.
Find scientific papers tied to the actual drug candidate
Search aliases, targets and trial identifiers
Search the asset’s current name and former development codes, then widen the search to its target and disease. In PubMed, use OR to combine aliases and AND to connect separate concepts; parentheses keep groups of terms together. A query might combine two candidate names with a disease term, followed by a separate search using the trial identifier. PubMed provides citations and abstracts, with links to full text where available, rather than hosting every article itself. The PubMed search and retrieval instructions cover search fields, filters and saved searches. Keep a record of the queries you used so that another search can reproduce the same starting point.
Screen papers for relevance before reading them closely. Label each as candidate evidence, target evidence, technology evidence or background disease research. A review article can provide vocabulary and point to original experiments, but it should not replace those experiments in the evidence record. When a company cites laboratory findings, check the compound, experimental model and dose rather than assuming the paper tested the commercial candidate. The distinction between human evidence and what preclinical research can establish deserves its own line in the notes. Biological plausibility and demonstrated patient benefit answer different questions.
Read methods and supplementary material before accepting the headline
For a clinical results paper, begin with the study population, treatment groups and analysis methods. Find the participant flow diagram, the numbers assigned to each group and the numbers included in each analysis. Then inspect the outcomes, effect estimates and harms. The CONSORT 2025 reporting statement provides a structured framework for reporting randomized trials, including access to the protocol and statistical analysis plan, funding disclosures and participant flow. Use that framework to identify information worth locating, not as a score that proves a study is reliable. Complete reporting makes assessment possible; it does not settle the assessment.
For laboratory research, ask what the experiment directly establishes. Does it show that the drug binds its intended target, changes a biological process, or improves an outcome in the chosen model? Those findings should remain separate in your notes. Trace the company’s strongest scientific claim to the figure or experiment supporting it, then read the controls and qualifications around that result. Our guide to evaluating drug targets and mechanisms of action covers the scientific judgment involved. Here, the research discipline is to preserve the distance between the observed finding and the commercial claim built on it.
Read the supplementary files as part of the paper, not as optional storage. Look there for eligibility criteria, detailed analysis methods and results omitted from the main tables. When the protocol or analysis plan is available, save its version number and date alongside the article. Check whether the paper’s comparator and patient population answer the question your investment case requires. A beautifully presented result can still address the wrong commercial question. Detailed assessment of randomisation, blinding and control groups belongs in clinical trial design analysis; the source review should establish which design was actually used.
Track publication status and document versions
Label preprints clearly. They can provide useful early access to methods and results, but medRxiv manuscripts have not been certified by peer review, as stated in its description of preprints and publication screening. Save the version you reviewed and check whether a journal article subsequently appeared. If it did, compare the conclusions and analyses rather than assuming the documents are interchangeable. Apply the same discipline to conference material: identify whether you have an abstract, a poster or a complete presentation. Record exactly what is available, and avoid filling missing methods with assumptions.
Before relying on a paper, look for corrections, retractions and later follow-up reports. Read the funding statement and author disclosures, then check the work on its merits. Sponsor involvement is a reason to inspect the methods and reporting carefully, not a substitute for doing so. Pay particular attention when a company’s presentation emphasizes a subgroup while the paper gives greater weight to a broader analysis. Keep that difference visible and investigate it through the framework for selective reporting, subgroup claims and missing data. Do not turn an unresolved discrepancy into either an accusation or a clean bill of health.
Read trial registries as dated development records
Find the study using its identifier whenever possible, then confirm the sponsor, intervention, disease and protocol number. In the registry record, locate the enrollment figure, eligibility criteria, treatment groups, outcome measures and study dates. ClinicalTrials.gov separates study details from posted results and other supporting information; its instructions for reading a study record identify those sections. Build your extraction sheet around the fields that matter to the investment question. If the thesis depends on a particular dose or patient subgroup, do not settle for a registry title that broadly resembles the company’s description.
Read status labels literally. “Active, not recruiting” means the study continues without recruiting new participants. “Completed” describes normal study completion, not a favorable outcome. The primary completion date concerns final data collection for the primary outcome measures; the study completion date also covers the remaining outcome measures and adverse-event collection. Estimated dates remain estimates. Registry information also passes through quality-control checks that do not verify its scientific validity. These distinctions appear in the ClinicalTrials.gov study terms and record guidance. They prevent common shortcuts, including treating a completion date as a promised results announcement or a government-hosted listing as scientific endorsement.
Compare historical versions before interpreting changes
Review earlier versions, rather than reading only the current record. ClinicalTrials.gov provides historical study records through its Record History function, documented in the National Library of Medicine description of study-record history. Compare the original entry, a version before the relevant analysis and the current version. Note changes to enrollment, endpoints, eligibility and dates. A change deserves a question, not an automatic verdict. Ask what changed, when it changed, whether an explanation is available and whether the change affects the claim being evaluated. Keep the registry submission date separate from the date on which a protocol amendment actually took effect.
Check registration timing as well. The ICMJE clinical trial registration policy requires prospective registration for consideration by journals following that policy, subject to its stated historical provisions. This is a publication policy, not a universal description of every legal registration obligation. For investment research, the practical question is whether the public record helps establish what was planned before outcomes were known. A late registration or a thin original entry reduces what you can verify from that record alone. Look for dated protocols and analysis plans before reaching a judgment.
Build a timetable that separates enrollment completion, primary data collection and management’s expected announcement window. Add the reporting date of each statement. When those dates diverge, first test whether they refer to different milestones. Do not manufacture a delay by comparing unlike dates, or dismiss a genuine delay because one field still looks unchanged. The investment relevance of these events belongs in an assessment of biotech catalysts and development milestones. Your research file should supply the underlying chronology, including which dates are confirmed and which remain management estimates.
Use financial filings to test funding and ownership
For U.S. issuers using these forms, read the annual 10-K alongside the latest 10-Q. The annual report contains audited financial statements; quarterly reports provide unaudited financial updates. Business descriptions, risk factors and management’s discussion add context to the numbers. Foreign issuers may report under different forms, so establish the company’s reporting regime before searching. The SEC investor bulletin on reading 10-K and 10-Q reports sets out these documents and their principal sections. It also makes a boundary clear: the company prepares the reports, and the SEC does not vouch for their accuracy. Filing status does not turn a scientific assertion into independent validation.
Read the cash flow statement and notes together
Extract cash, investments and operating cash use, keeping their reporting dates attached. Then inspect restrictions, debt obligations, contractual commitments and events disclosed after the balance-sheet date. Do not substitute accounting loss for cash consumption. Build a bridge from reported liquidity to the amount potentially available for the development plan, explaining each adjustment. Compare management’s funding statement with your own assumptions about trial spending. The detailed calculations belong in cash burn and cash runway analysis. The source review should establish which figures are reported, which are estimates and which obligations could change the available funding.
Read share information with equal care. Record the shares outstanding and the date to which that figure applies, then inspect financing disclosures, options, warrants and convertible instruments. Avoid using a stale share count with a newer cash balance without explaining the mismatch. Search for financing activity between the reporting date and your research date. An authorization to issue securities is different from a completed financing, so identify what actually occurred. Our guide to share dilution and equity financing risk covers the ownership consequences. Here, the goal is to collect a dated record that supports those calculations without mixing completed transactions with possible future issuance.
Read agreements beyond the announcement
Search subsequent current reports and open the relevant exhibits. Form 8-K covers prescribed events and other disclosures, while its exhibits can include agreements and associated announcements. The SEC guidance on Form 8-K items and exhibits distinguishes material agreements, financing events and other disclosures. Use the item number to orient the search, then inspect the attachment that contains the terms. A press release attached to a filing remains a company announcement; its location does not replace the missing protocol, detailed results or contract. If a document incorporates an earlier agreement by reference, follow that trail too.
For licensing arrangements, extract the asset, territories, development responsibilities and payment structure. Separate cash already received from contingent milestones. Identify who pays for future trials and whether the company retains commercial rights in the market assumed by your model. If parts of the agreement are redacted or unavailable, leave those terms unresolved rather than reverse-engineering a favorable answer. The valuation work belongs in assessing upfront payments, milestones and royalties. A research note should make the ownership chain readable enough that someone can distinguish the drug’s potential sales from the proceeds that might reach the listed company.
Add regulatory evidence when the investment claim depends on approval
When a thesis depends on an approval, search for the decision documents rather than relying on a pipeline graphic. Drugs@FDA contains product information and, where available, approval letters, labels and review materials. Coverage varies by product and period; the FDA describes the database’s scope in its Drugs@FDA coverage and document guide. Check the application, indication and document date before extracting a finding. A review of one indication should not be presented as a decision on another. Save the document that supports the claim, not a search-results page that might look different when reopened.
Read the approved indication against the patient population and treatment setting used in the commercial forecast. Inspect relevant warnings, dosing requirements and any obligations stated in the decision materials. Keep management’s description of regulatory discussions separate from a regulator’s published decision. If public documents do not resolve an issue, say so in the research record. Do not translate “discussion,” “submission” or “review” into approval. The broader process is covered in FDA and EMA drug approval procedures; this part of the work is document matching, establishing exactly what was decided and which investment assumptions that decision can support.
Reconcile documents before assigning investment meaning
Create an evidence ledger with separate fields for reported fact, interpretation and modelling assumption. “The registry lists 240 participants” is a reported fact. “The broader population may reduce the apparent treatment effect” is an interpretation requiring analysis. “Approval occurs in two years” is a model assumption. These categories should not merge because a presentation uses confident wording. Link every reported fact to a document location, such as a table, figure or filing note. Once that record is assembled, assessing the investment meaning of trial results becomes a more disciplined exercise, with the evidential gaps visible before a valuation changes.
Match denominators and data cutoffs whenever two reports appear inconsistent. Check whether the figures describe all enrolled participants, treated participants or a narrower analysis group. Confirm the dose, follow-up period and definition of the outcome. For safety, compare the analysis populations and observation periods before comparing percentages. A longer follow-up report may answer a different question from an earlier announcement. The detailed interpretation of adverse events and treatment discontinuations requires more than copying a headline rate. Your first job is to establish whether the two figures are genuinely comparable, and to document any mismatch that remains.
A hypothetical example of cross-checking a biotech announcement
Consider a fictional developer announcing a 60% response rate and stating that it is funded through its next results announcement. Neither statement should enter a model unchanged. The following hypothetical research record illustrates how a favorable headline can produce several different follow-up questions without proving that the announcement is wrong.
| Document finding | What to investigate | Treatment in the research note |
|---|---|---|
| The registry lists a planned enrollment of 240 participants. | Does the announcement cover the full study or an earlier cohort? | Keep planned enrollment separate from the reported analysis population. |
| The response figure covers 30 evaluable patients. | How many were treated, and why were others excluded? | Record the denominator and missing information beside the rate. |
| The reported endpoint is secondary. | What happened on the primary endpoint? | Do not describe the study as successful based on this result alone. |
| The cash balance predates a new trial commitment. | Does the funding estimate include the additional spending? | Rebuild the funding bridge using consistent dates and assumptions. |
Suppose the same fictional company reports $120 million of unrestricted cash and investments, and the analyst assumes future cash use of $15 million per quarter. Simple division gives eight quarters. That is arithmetic under an assumption, not a verified funding forecast. The researcher still needs to test whether the spending estimate includes the announced trial, whether contractual payments fall due, and whether cash use changes before the next milestone. Keep the company’s guidance alongside the independent estimate, with the reason for any difference. Do not silently replace one with the other.
The investment response should reflect what the evidence changes. A denominator question might affect the strength assigned to an efficacy signal; a funding gap, the financing assumptions. Neither automatically determines the correct share price. Carry unresolved questions into valuation sensitivity analysis rather than hiding them inside a single forecast. Where the available documents cannot support a firm judgment, use alternative cases or postpone that judgment. An explicit unknown is more useful than an estimate whose apparent precision comes from missing evidence.
Maintain a research log that can be updated
Finish each research session by making the record reusable. Save the document title, URL, publication date and relevant page or table. For trial records, preserve the version date; for results, the data cutoff; for financial statements, the reporting period. Keep downloaded documents where access terms permit, using filenames that distinguish versions. Avoid overwriting an earlier thesis when new evidence arrives. Write a dated update explaining what changed and which assumption it affects, so the history shows the reasoning rather than only the latest opinion.
- Claim: State the proposition being tested in one sentence.
- Evidence: Record the supporting document and exact location.
- Comparison: Identify the other document used to check consistency.
- Open question: Explain what public evidence does not resolve.
- Update trigger: Name the filing, dataset or decision that should prompt another review.
Set monitoring priorities around those open questions. Save literature searches, review new filings and revisit trial records at the milestones already recorded in the timetable. Check for amended documents and corrections before adding a new version to the file. Give each unresolved issue a next action: locate the protocol, reconcile the patient count, read the financing exhibit or wait for the promised dataset. If the answer would not change the investment judgment, reduce its priority. If the entire thesis depends on it, leave the decision open until the evidence is strong enough to carry that weight.