Mosaic Theory: Research and Ethics
Table of Contents
An analyst can read the same annual report as everyone else and reach a different conclusion about a company’s future. The useful question is how the analyst got there: which facts were available, which assumptions were added, and whether any information came with restrictions. Mosaic theory explains the role of that analytical work while keeping the source of an insight central to the discussion.
This guide shows how to assemble a traceable investment inference, work through an entirely hypothetical public-information example, and recognize when new information calls for a pause and appropriate compliance or legal guidance. It uses CFA Institute ethics guidance and U.S. securities-law sources, which serve different purposes.
What Is Mosaic Theory?
In investment research, mosaic theory describes reaching a meaningful conclusion by analyzing a collection of information. CFA Institute’s Standard II(A) guidance recognizes analysis using public information and nonmaterial nonpublic information. An analyst’s own conclusion may be significant even though the analyst was never given the equivalent material inside fact directly. [1]
The inference is the analyst’s work. Its source trail should show what was observed, what was assumed, and how the conclusion followed. Calling the work a mosaic does not resolve concerns about material nonpublic information (MNPI).
For example, an analyst might combine published store counts, an announced expansion plan and historical sales economics to estimate future operating profit. That estimate is different from a finance employee privately supplying the company’s unreleased earnings result. Both may concern earnings, but the route to the information matters.
Mosaic analysis fits within fundamental analysis: it helps turn evidence about a business into an investment thesis. It does not establish that the thesis is correct, that the market has overlooked it, or that trading on every input is permitted.
Mosaic Theory vs. Insider Trading
Three separate sets of requirements can apply to research. CFA Institute standards govern its members and candidates. Securities law determines legal obligations under the applicable jurisdiction and facts. Firm policies control what employees may do within their organization. A conclusion about one does not settle the others.
| Framework | Question it addresses | Practical implication |
|---|---|---|
| CFA Institute Standard II(A) | Does the information or proposed action breach the professional standard? | Members and candidates must not act or cause others to act on material nonpublic information that could affect investment value. |
| U.S. securities law | Do the facts meet the requirements of an applicable prohibition? | Consider the information, duties, relationships and conduct; a mosaic label supplies no automatic safe harbor. |
| Firm policy | What contacts, information access and trades does the employer permit? | Follow required approvals, information barriers and trading restrictions even when the analyst believes a source is acceptable. |
The SEC’s general description of illegal insider trading includes securities trading based on material nonpublic information in breach of a fiduciary duty or another relationship of trust and confidence. Tipping and misappropriation can also be involved. The exact legal analysis depends on the relevant theory and facts; it is broader than asking whether the trader works for the issuer. [2]
Materiality and public availability are separate questions
Materiality concerns importance to a reasonable investor and the total mix of available information. Specificity, reliability and context matter. There is no universal dollar amount or percentage below which a private disclosure is automatically immaterial. An indirect confirmation of a major earnings change can matter even without an exact earnings number. [3]
Public availability concerns dissemination to investors generally. A private conversation, an invitation-only briefing or a message to a selected group does not become public simply because several people received it. Check when and how the underlying information became generally available, rather than assuming that a widely repeated rumor or forwarded document establishes public status. [1]
These questions explain why asymmetric information is a broader concept than illegal insider trading. Market participants can have different expertise and research resources. The existence of an information advantage alone does not establish either a violation or permission to trade.
Regulation FD has a separate issuer-disclosure role
Regulation FD addresses specified disclosures of material nonpublic information by covered issuers or people acting on their behalf to covered recipients. Subject to its scope and exceptions, it requires simultaneous public disclosure for an intentional disclosure and prompt public disclosure for a nonintentional one. It is not a general clearance process for an analyst’s trades. [4]
The SEC’s adopting release recognizes that a genuinely immaterial disclosure may help an analyst reach a significant conclusion through expertise. The same discussion cautions that an issuer cannot make material information immaterial by dividing it into ostensibly nonmaterial pieces. Assess the substance and context of the communications together; several innocuous-looking fragments do not prove that the information is acceptable. [3]
How to Build a Public-Source Mosaic
Start with a question narrow enough to test. “Will expansion increase operating profit this year?” is more workable than “Is this a good stock?” The first question identifies the inputs, time period and potential gaps before a valuation or trading decision enters the discussion.
When a question calls for observations from stores, suppliers or other operating channels, see how to design and evaluate channel checks. That workflow adds sampling and triangulation discipline while retaining the information-use boundaries explained here.
Consider Harbor Stores, a fictional retailer. Every document, date and amount in this exercise is invented. Assume the described annual report and company updates were broadly released to the public before the analyst’s July 15, Year 1 research cutoff. There are no private conversations or restricted datasets in this example.
The fictional annual report covers Year 0. A later public plan describes intended Year 1 openings, and a July 1 update confirms actual openings. The ledger distinguishes what those records say from what the analyst still has to estimate.
| Public record | What it establishes | Analyst assumption | What remains unknown |
|---|---|---|---|
| Year 0 annual report | 100 established stores at year-end; annual sales of $2 million per store; 12% operating margin for that business scope. | The existing stores remain open all Year 1 and sustain the same sales and margin. | Future same-store sales, closures and cost changes. |
| Year 1 expansion announcement | Management plans to add 20 stores by midyear. | The plan will be executed on schedule. | Whether each opening actually happens. |
| July 1, Year 1 public update | All 20 new stores opened on July 1. | New stores operate for half the year at a selected sales pace and margin. | Actual sales ramp, startup costs and new-store profitability. |
The opening plan and the opening confirmation answer different questions. Once the July update appears, timing is better supported. It still does not confirm future revenue or margin. Keep the historical metric, the management plan and the analyst forecast labeled separately.
Check that each metric refers to the same business scope. A company-wide margin may include activities beyond its stores, and a new-store contribution margin may omit central costs included in consolidated operating profit. Segment reporting can help identify those differences. A calculation is only meaningful when its inputs describe compatible operations.
Mosaic Theory Example: From Stores to Profit
Use the fictional public inputs to build a conditional estimate. Assume the existing 100 stores operate for a full year at $2 million annual sales each and a 12% operating margin. For this simplified model, the margins include the relevant operating costs and allocations; there are no additional unmodeled startup or central costs, closures or changes in existing-store economics.
The $24 million figure is the analyst’s conditional estimate for the existing fleet. It is not an unreleased company result. Carrying historical inputs forward is itself an assumption, even when every source document is public.
Separate a full-year run rate from the first-year contribution
The July 1 openings leave the new stores only half of Year 1 to generate sales. Also define sales realization as their sales pace while open relative to the mature-store benchmark of $2 million annually. A realization assumption of 80% means an annualized pace of $1.6 million while open, before applying the half-year timing factor.
Under those assumptions, new-store revenue is $16 million and operating profit is $1.92 million. Adding that contribution to the $24 million existing-store estimate gives $25.92 million for Year 1. This uses a simplified half-year convention rather than a daily opening schedule.
A different question is what 120 stores could earn over a complete year at the mature sales pace and 12% margin. That calculation is 120 × $2 million × 12% = $28.8 million. Treating this full-year run rate as the Year 1 estimate would skip both the opening timing and the assumed ramp.
Test which assumptions drive the conclusion
Keep existing-store operating profit fixed at $24 million and compare four scenarios for the 20 new stores. Change only their time open, sales realization and operating margin as labeled. “Half year” means the July 1 openings; the 9% margin case applies only to new stores. All scenarios retain the model’s cost-coverage assumptions.
| Scenario | Operating profit (USD millions) |
|---|---|
| Full year, 100%, 12% | 28.80 |
| Half year, 100%, 12% | 26.40 |
| Half year, 80%, 12% | 25.92 |
| Half year, 80%, 9% | 25.44 |
Hypothetical totals. Labels give time open, sales realization and new-store margin; existing-store profit stays $24 million.
The half-year, 80%-realization estimate is $2.88 million below the $28.8 million mature full-year case. The lower-margin case reduces the estimate further to $25.44 million. The evidence supports a range of conditional outcomes, and the assumptions explain the differences.
These are operating-profit estimates, not net income or free cash flow. Financing, taxes, capital expenditure and working-capital needs require additional analysis. Use consistent profit margin definitions before carrying a research inference into a valuation.
When New Information Requires a Pause
Return to the Harbor Stores exercise. After completing the public-source model, the analyst unexpectedly receives a message from someone claiming to work in finance: “I have the unreleased internal earnings number.” That message is outside the public ledger. Its arrival raises a new information-control question, even if the original forecast was built appropriately.
Pause trading and recommendation action while the information is assessed. Do not solicit the number or circulate the message among investors. Preserve the communication and consult designated compliance or qualified securities-law counsel about its status, restrictions and next steps. Refer any existing orders or instructions to them as well.
The response does not depend on first proving a legal violation yourself. The source’s access, the information’s reliability and significance, public dissemination, relevant duties and applicable rules need assessment. An earlier public-source thesis does not by itself establish that a later trade is permissible after receiving potential MNPI. U.S. Rule 10b5-1 uses an awareness standard subject to specific affirmative defenses; “I had already decided” does not establish that those conditions are met. [5]
Follow the firm’s process for limiting access and resolving the issue. Do not delete records, independently distribute the information to make it “public,” or assume removing one line from a spreadsheet solves the problem. The checklist is a way to recognize and escalate uncertainty, not a legal determination about the fictional message.
Expert networks and confidentiality agreements
An industry expert can explain technology, terminology and publicly observable business conditions. That is different from an employee supplying confidential quarterly results or unreleased test data. In its February 2011 expert-network charges, the SEC expressly distinguished legitimate expert advice from trading on MNPI obtained in violation of a confidentiality duty. The release describes allegations, not a blanket prohibition on expert consultation. [6]
A non-disclosure agreement does not establish permission to trade. Regulation FD includes an exception for disclosure to a person who expressly agrees to maintain confidentiality, while Rule 10b5-2 identifies confidentiality agreements among the circumstances creating a duty of trust or confidence for misappropriation analysis. The agreement can enable a legitimate confidential business discussion while restricting subsequent use of the information. [4] [7]
For the firm-level process governing controlled access, see information barriers and wall-crossing. That guide distinguishes deliberate information sharing, watch and restricted lists, and research controls from the analyst’s task of assembling a mosaic.
How to Document the Research
A useful record should let another person reconstruct what the analyst knew when the inference was made. It should also expose weak assumptions before they become an investment decision. CFA Institute’s mosaic guidance specifically recommends retaining supporting research. [1]
- Record provenance and timing. Save the source title, publisher, release date, access date and relevant page or passage. For a filing, record its identity and reporting period; for a call, note whether the public had access.
- Separate statements from assumptions. “Management announced 20 planned openings” is a sourced statement. “All stores will reach mature sales immediately” is a modeling assumption.
- Show the transformation. Retain formulas, units, scope adjustments and scenario definitions. Someone checking Harbor Stores should see why the half-year factor is separate from sales realization.
- Look for disconfirming evidence. Ask what would weaken the thesis: delayed openings, lower sales, omitted operating costs or closures. Keep conflicting public evidence alongside the supportive material.
- Record unresolved information questions. Note restrictions, uncertain public status and the appropriate escalation. Follow firm retention and access rules; avoid placing restricted material in a broadly shared research folder.
- Version the conclusion. Date the inference and explain later changes. Preserve the earlier analysis so a subsequently released result is not mistaken for information available at the original cutoff.
A concise conclusion for the example would be: “At the July 15 cutoff, the public opening update supports half-year timing for 20 new stores. Assuming 80% mature sales realization and a 12% operating margin, total Year 1 operating profit is estimated at $25.92 million. Actual ramp and costs remain unverified.” This states what is supported and what could change.
Documentation improves traceability. It does not cure an inappropriate disclosure or authorize a trade. Keep investment analysis and information-permission decisions explicit throughout the process.
Common Mistakes
1. Counting sources instead of evaluating them. Ten articles repeating one anonymous claim provide less independent corroboration than their number suggests. Trace the original source and evaluate the evidence rather than treating repetition as confirmation.
2. Calling every small private detail nonmaterial. Materiality depends on context and the total mix. A series of confirmations can convey something important, and an issuer cannot avoid that issue by fragmenting its disclosure.
3. Turning a forecast into a fact. Public historical sales and an opening announcement do not disclose future profit. Label estimated growth, sales realization, margins and costs so the conclusion does not acquire false certainty.
4. Mixing time periods and metric definitions. A year-end store count is not a full-year operating count. A mature annual run rate is not a first-year result. A store contribution margin is not necessarily the same as company operating margin.
5. Trusting a source’s reassurance. “Everyone knows,” “this is fine” or “our agreement covers it” does not establish public status, materiality or permitted use. Follow the actual compliance process when the information raises concerns.
6. Assuming the label proves legal or investment success. A documented mosaic is still subject to applicable law, standards and firm restrictions. A soundly sourced forecast can also be wrong or already reflected in price.
Limitations of Mosaic Theory
A plausible narrative can connect weak evidence just as easily as strong evidence. Check source quality, competing explanations and the assumptions that could overturn the conclusion.
Public inputs can be incomplete or misleading. Disclosures can contain errors, stale observations and optimistic plans. Several sources may depend on the same underlying statement. Combining them does not eliminate those weaknesses.
The information boundary can be difficult to assess. Materiality, dissemination, confidentiality duties and the surrounding circumstances require judgment. U.S. sources do not resolve other jurisdictions’ rules, and this educational example cannot provide clearance for a real research program.
A business inference is only one part of valuation. Harbor Stores’ estimated operating profit says nothing by itself about financing, future investment requirements or the price worth paying for its shares. Those questions belong in an intrinsic value analysis.
Research does not guarantee excess returns. Other investors may have reached the same view, and the current price may already reflect it. The efficient market hypothesis addresses that separate question about information and prices. Mosaic theory does not promise an exploitable edge.
Frequently Asked Questions
Sources
Primary sources checked November 13, 2026. CFA guidance below was updated in April 2024. The eCFR sections were displayed as current through November 10, 2026; the SEC adopting release and enforcement release provide dated background.
- CFA Institute: Standard II(A) Material Nonpublic Information — the standard, mosaic guidance, public status, industry experts and compliance practices.
- SEC Investor.gov: Insider Trading — general U.S. definition, tipping and misappropriation.
- SEC: Selective Disclosure and Insider Trading — 2000 adopting release, especially Section II.B.2 on materiality, mosaics and fragmented disclosure.
- 17 CFR 243.100: General Rule Regarding Selective Disclosure — covered disclosures and confidentiality exceptions.
- 17 CFR 240.10b5-1: Trading on the Basis of Material Nonpublic Information — awareness standard and conditional affirmative defenses.
- SEC: SEC Brings Expert Network Insider Trading Charges — February 3, 2011 allegations and the distinction between expert advice and misuse of confidential information.
- 17 CFR 240.10b5-2: Duties of Trust or Confidence — nonexclusive duty circumstances under the misappropriation theory.
Disclaimer
This article is educational and does not constitute investment, legal or compliance advice. CFA Institute standards, U.S. securities law and firm policies are distinct, and other jurisdictions may differ. Harbor Stores and all associated documents and figures are hypothetical. Consult appropriate compliance personnel or qualified securities-law counsel about actual information access, restrictions and proposed investment actions.
Keep reading
More in Equities
-
Fundamental Analysis: A Step-by-Step Guide to Stock Analysis
Fundamental analysis is the systematic process of evaluating a stock’s intrinsic value by examining financial statements, profitability metrics, valuation…
-
P/E Ratio: Price-to-Earnings Ratio Explained
The price-to-earnings ratio (P/E) measures how much investors pay for each dollar of a company’s earnings, serving as the most widely used equity valuation…
-
DCF: Discounted Cash Flow Valuation Explained
Discounted cash flow analysis is the gold standard of intrinsic valuation in finance. Rather than relying on what the market currently pays for comparable…
