Status: research-only. This document does NOT constitute legal or tax advice. Before filing or acting on anything below, consult a securities attorney licensed in the relevant federal + state jurisdictions, and a privacy attorney for the GLBA/CCPA section. Last updated: 2026-04-29 UTC. Sources as of that date — verify freshness before any filing.
An LLM that translates user-authored strategies into executable DSL is the strongest argument for staying outside the Advisers Act "investment adviser" definition; the moment the AI proposes or scores strategies, that argument weakens materially. The hybrid broker model (Alpaca default + BYOB via aggregator) does not by itself trigger investment-adviser registration, but it does not eliminate it either — the advice question is analytically separate from the custody question. The single gating question before any live-trading feature ships is: "Does Raxx's AI propose, score, or classify strategies on behalf of the user, or does it only translate the user's own language into a structured format?" The answer to that question — and only that question — determines which regulatory lane Raxx is in.
Statutory text — 15 U.S.C. § 80b-2(a)(11):
"any person who, for compensation, engages in the business of advising others, either directly or through publications or writings, as to the value of securities or as to the advisability of investing in, purchasing, or selling securities, or who, for compensation and as part of a regular business, issues or promulgates analyses or reports concerning securities"
Source: https://www.law.cornell.edu/uscode/text/15/80b-2
The statutory definition is satisfied when ALL THREE prongs are met. SEC Release IA-1092 (1987) codified the three-prong test:
A — Compensation. Any economic benefit received, including subscription fees, SaaS fees, or platform fees bundled with advisory activity. The compensation does not need to be a separate charge for advice; it need only be part of aggregate compensation received for services that include advisory activity. (Source: https://www.kitces.com/blog/abcs-financial-coach-register-investment-adviser-status-sec-series-65-66-nasaa/ citing SEC Release IA-1092)
B — Business. Advice must occur with regularity — "anything other than rare, isolated and non-periodic instances" satisfies this element. A recurring automated system running on user accounts satisfies "business" with essentially no argument to the contrary.
C — Advising others. Providing advice about specific securities to third-party users. This element is the easiest to satisfy and the hardest to avoid in a user-facing platform.
Critical corollary: Raxx charges subscription fees (or will). The moment the platform does anything advisory alongside those fees, Prong A is satisfied. B and C are trivially satisfied for a recurring automated platform. The only real argument for exclusion is that what Raxx does is not "advising" — it is execution of the user's own advice.
The statute (§ 80b-2(a)(11)(A)–(H)) lists eight categorical exclusions:
| Exclusion | Applies to Raxx? |
|---|---|
| (A) Banks | No |
| (B) Lawyers/accountants/engineers/teachers whose advice is incidental | No |
| (C) Broker-dealers whose advice is solely incidental + no special compensation | Potentially, if Raxx registers as a B-D — but Raxx is not currently a B-D |
| (D) Publishers of bona fide newspapers, news magazines, financial publications of general circulation | Potentially (the "publisher exclusion") — analyzed below |
| (E) Advice only on U.S. government obligations | No |
| (F) NRSROs | No |
| (G) Family offices | No |
| (H) SEC-designated persons | No |
Source: https://www.law.cornell.edu/uscode/text/15/80b-2
Lowe v. SEC, 472 U.S. 181 (1985) is the leading Supreme Court case on the publisher exclusion under § 80b-2(a)(11)(D). The Court held:
Sources: https://supreme.justia.com/cases/federal/us/472/181/ https://www.law.cornell.edu/supremecourt/text/472/181
Application to an LLM-based strategy platform:
The publisher exclusion is narrow and unlikely to protect Raxx's AI in most scenarios. The exclusion was designed for mass-distributed newsletters with no individualization — the antithesis of a platform that reads a user's account, parses their natural-language input, and executes orders against their specific positions. Courts and the SEC have consistently read the exclusion to require (1) general circulation, (2) impersonal content, and (3) no ongoing client relationship. A per-user AI that operates inside a logged-in account fails at least elements (2) and (3).
The Seeking Alpha precedent (2024): A district court dismissed an SEC complaint against Seeking Alpha, holding that a financial publication aggregating user-submitted articles was within the publisher exclusion. (Source: https://katten.com/judge-dismisses-case-against-seeking-alpha-implications-for-publishers-of-financial-information) However, Seeking Alpha is a general-circulation publication; the ruling does not extend to per-account personalized execution tools.
Research finding subject to attorney review: The publisher exclusion is almost certainly unavailable for Scenarios B and C (scored/proposed strategies). It is theoretically arguable for Scenario A (pure parser) but even there the per-user, per-account, execution-linked nature of the service distinguishes it from Lowe.
Wealthfront Advisers LLC and Betterment LLC both registered as SEC investment advisers upon launch and remain registered today. Both have Form ADV filings publicly available through the SEC's Investment Adviser Public Disclosure system: - Wealthfront: https://adviserinfo.sec.gov/firm/summary/148456 - Betterment: https://adviserinfo.sec.gov/firm/summary/149117
The SEC's 2017 robo-adviser guidance (IM Guidance Update 2017-02) defined the term "robo-adviser" to include "registered investment advisers that use computer algorithms to provide investment advisory services online with often limited human interaction." The guidance treated robo-advisers as full investment advisers subject to the same fiduciary duty, disclosure, suitability, and compliance obligations as human advisers, with specific guidance on: - Disclosure requirements (Form ADV Part 2A brochure must describe the algorithm) - Suitability obligations (questionnaire design, investor profile sufficiency) - Compliance program obligations
Source: https://www.sec.gov/newsroom/press-releases/2017-52
Operational distinction between Raxx and registered robo-advisers:
| Attribute | Wealthfront / Betterment | Raxx (proposed) |
|---|---|---|
| Portfolio construction | Platform proposes + executes | User authors strategy; platform executes |
| Discretion | Platform has trading discretion | User retains discretion (in Scenario A/B) |
| Goal intake | Questionnaire drives allocation | User-written natural-language strategy |
| Asset custody | User account custodied at platform or DTCC clearing | User account at Alpaca or user's own broker |
| Registered as IA | Yes — SEC-registered, Form ADV filed | Not currently registered |
Research finding subject to attorney review: The Wealthfront/Betterment model involves the platform holding discretionary trading authority. Raxx's Scenario A does not give the platform discretionary authority — the user authors the strategy, the platform executes against the user's authorization. That distinction is substantively meaningful and is the foundation of the "pure parser" legal argument. However, the SEC has not issued a no-action letter specifically clearing this exact model, so the analysis is untested at the agency level.
Under the post-Dodd-Frank framework, investment advisers with under $100 million in regulatory AUM register at the state level, not with the SEC. A startup Raxx will be below this threshold for years.
Registration triggers in priority states:
| State | Trigger | De minimis exemption | Initial fee | Annual renewal |
|---|---|---|---|---|
| California (DFPI) | IA status + 6+ CA clients | Fewer than 6 CA clients + no CA place of business | $125 (IARD) | $125 |
| New York (AG Office, GBL § 359-eee) | IA status + 6+ NY clients | Fewer than 6 NY clients (no NY place of business required per AG FAQ) | $200 (IARD) | $200 |
| Texas (SSB) | IA status + 6+ TX clients | Fewer than 6 TX clients + no TX place of business | ~$150 (IARD) | ~$150 |
Sources: - CA: https://dfpi.ca.gov/regulated-industries/broker-dealers-and-investment-advisers/state-licensed-investment-adviser/ - NY: https://ag.ny.gov/investment-advisers-faq - TX: https://www.ssb.texas.gov/securities-professionals/dealer-adviser-registration/getting-started-registered-investment-adviser-3 - NASAA state-by-state: https://www.nasaa.org/industry-resources/investment-advisers/state-investment-adviser-registration-information/
Critical point for a multi-state SaaS: If Raxx determines it is an investment adviser and registers in any state, it must register in every state where it has 6 or more clients. At Raxx's target scale (hundreds or thousands of users), that means 50-state registration. Ongoing compliance cost for a 50-state registered IA is material — IARD fees alone are roughly $5,000–$10,000/year, plus per-state filing, exam, and annual report requirements. This is before attorney fees for maintaining compliance program, Form ADV, and annual update filings. (Source: https://www.nasaa.org/73254/nasaa-announces-2025-fee-schedule-for-investment-adviser-registration-depository-system/)
As of April 2025, the SEC is evaluating raising the AUM threshold for mandatory federal registration, which could shift the state-vs-federal registration split. The threshold has been unchanged since 2012 ($100M). Watch for SEC rule proposal. Source: https://www.wealthmanagement.com/regulation-compliance/sec-to-review-aum-threshold-for-state-advisors
The analysis below is research-only. A securities attorney must confirm applicability to Raxx's specific implementation. The posture calls are structured as probability assessments for planning purposes, not legal opinions.
Description: Raxx's AI only translates the user's natural-language strategy into a structured DSL. No proposals, no scoring, no feasibility ratings. The user authors the entire strategy concept. The system executes what the user authorized. User's broker holds the account.
Three-prong analysis: - Compensation: Satisfied (Raxx charges subscription fees for the platform). - Business: Satisfied (recurring, regular service). - Advising: Arguable no. The platform is executing the user's own advice, not giving advice. The analogy is a word processor that formats a legal document — it does not practice law. The LLM is performing natural-language parsing, not reasoning about what the user should do.
Key risks: 1. The "compensation" prong is met even though advice itself is not the product, under SEC Release IA-1092's bundled-fee reading. However, the bundled-fee reading only matters if the advisory element also exists. If no advisory element, no adviser status. 2. A court or regulator could characterize even neutral parsing as "advising" if the parse materially shapes the execution — e.g., if the LLM fills in gaps in an ambiguous instruction in a way that reflects a judgment about what the user "probably meant." Gap-filling is the key risk. 3. Without clear UX documentation that the user is the strategy author and the system is only a translator, the factual record could support a different characterization.
Posture call: Likely exempt — but depends entirely on implementation. This is the strongest argument for non-adviser status. The argument holds as long as the AI's role is genuinely ministerial (translation only, no gap-filling, no scoring, no implicit recommendations). Confirm implementation details with a securities attorney before launch. This is ambiguous enough that a no-action letter request to the SEC staff could be valuable — see Questions for Attorney below.
Description: Same as A, but Raxx surfaces feasibility/risk readouts: PDT $25K rule flag, "your seed × 50%/yr return is aggressive," risk-band classification (conservative / moderate / aggressive), or similar.
Three-prong analysis: - Compensation: Satisfied. - Business: Satisfied. - Advising: Likely satisfied. Once Raxx's system issues "analyses or reports concerning securities" (language from the statute), the Advisers Act's second definitional clause applies even without the first. Feasibility scoring is an analysis of a securities strategy. The PDT flag is a regulatory note, not advice — but "your 50%/yr target is aggressive" is closer to a classification of the advisability of a strategy.
The SEC's 2025 Exam Priorities specifically flagged that examiners will scrutinize "compliance policies and procedures as well as disclosures to investors" when AI is "integrated into advisory operations" and that "an adviser cannot defer its fiduciary responsibility to an algorithm." This language suggests the SEC views strategy-scoring AI as part of advisory operations. Source: https://www.kitces.com/blog/artificial-intelligence-compliance-considerations-investment-advisers-sec-securities-exchange-commission-legal-regulation-framework/
Posture call: Ambiguous — leans toward "likely registered required." The line between a purely mechanical rule-check (PDT flag = informational) and a risk classification (aggressive/conservative = advisory) is legally significant but practically blurry. The safest path if Raxx adds scoring: assume adviser status and plan registration accordingly, or structure the scoring so narrowly mechanical (pure rule-of-law flags, not strategy quality judgments) that the advisory character is absent. This requires specific attorney guidance on each proposed output type.
Description: Raxx's AI proactively proposes strategies based on user goals (e.g., "you want 10% annual return with low volatility — here are three strategy templates").
Three-prong analysis: - Compensation: Satisfied. - Business: Satisfied. - Advising: Clearly satisfied. This is the core of investment adviser activity — recommending specific securities strategies to specific users for compensation. Wealthfront does this and is a registered IA. There is no credible argument that strategy proposal falls outside the definition.
Posture call: Likely required to register as investment adviser. Research finding subject to attorney review. This scenario almost certainly triggers registration unless an applicable exemption exists (none of the § 80b-2(a)(11) exclusions are available here). Operating a Strategy Proposer without registration would constitute an unregistered investment adviser under § 203 of the Advisers Act, which is a civil and potentially criminal violation.
| Scenario | Posture | Confidence | Attorney priority |
|---|---|---|---|
| A: Pure parser | Likely exempt | Moderate | High — confirm before launch; consider no-action letter |
| B: Parser + scorer | Ambiguous | Low | High — each scorer output type needs separate analysis |
| C: Strategy proposer | Likely required to register | High | Blocking — do not ship without attorney signoff |
When an LLM misinterprets a user's instruction and Raxx executes a resulting order the user did not intend, Raxx's liability posture implicates at least three theories:
Negligence: A platform that holds itself out as capable of executing user strategies via AI owes a duty of reasonable care in the design and operation of that parsing system. A misparse that results in an unintended trade could constitute breach of that duty if the system did not implement reasonable safeguards (confirmation steps, audit logging, notional caps). The damages (the financial loss from the unintended trade) are foreseeable. Standard for "reasonable care" in this context is still developing — no controlling case law specifically addresses LLM-mediated order misexecution in a retail fintech context as of the research date.
Breach of contract: If Raxx's Terms of Service represent the platform as accurately executing user-authored strategies, a systematic failure to do so — including via LLM misparse — could be characterized as a breach of the platform's core service promise. TOS language disclaiming AI accuracy is standard but not a complete shield (see 2.2 below).
Suitability/fiduciary: This theory applies only if Raxx is a registered investment adviser or broker-dealer. For Scenario A (pure parser, unregistered), suitability claims are not available to a plaintiff. However, registration in Scenario B or C immediately opens the door to suitability claims under the Advisers Act fiduciary duty or Reg BI (for broker-dealers).
Research finding subject to attorney review. No binding case law or regulatory guidance specifically addresses LLM-mediated retail trading order misexecution as of 2026-04-29. The negligence and contract theories above are applications of general principles to the AI context; the specific standard of care has not been adjudicated.
Well-drafted TOS reduces but does not eliminate liability. Key provisions that courts in CA and NY have analyzed:
Liability caps: Courts in California and New York generally enforce contractual liability caps between commercial parties and, with more scrutiny, in consumer contracts. For consumer-facing platforms, a cap at the subscription fee paid is typically the minimum defensible floor. Caps at zero (total disclaimers) are more aggressively challenged. California's Unfair Competition Law (Cal. Bus. & Prof. Code § 17200) and Consumer Legal Remedies Act have been used to challenge zero-liability disclaimers in consumer fintech contexts. (Unsourced — confirm with CA consumer protection attorney.)
Arbitration clauses: As of 2024, the FTC's Non-Compete Clause Rule did not directly affect fintech arbitration clauses, and the Supreme Court's pro-arbitration case law (Viking River Cruises, 2022; Coinbase v. Bielski, 2023) generally enforces individual arbitration clauses in consumer agreements. Class-action waivers combined with individual arbitration are enforceable in federal courts (AT&T Mobility v. Concepcion, 2011) and in California for individual arbitration claims, though PAGA claims remain a CA-specific risk. Source: https://www.law.cornell.edu/wex/regulation_best_interest_(reg_bi)
AI output disclaimers: As of 2026, no controlling case law specifically tests "AI-generated output is not guaranteed accurate" disclaimer language in a trade-execution context. The SEC's position (2025) that "an adviser cannot defer its fiduciary responsibility to an algorithm" suggests that if Raxx is registered as an IA, a TOS disclaimer of AI accuracy cannot fully disclaim fiduciary obligations. For unregistered platforms (Scenario A), general disclaimer language has more force but is still challenged under unconscionability doctrine when the damage is consumer financial loss.
FINRA arbitration: FINRA arbitration applies only to FINRA member firms. Raxx is not currently a FINRA member and is not a broker-dealer. If Raxx ever registers as a B-D, FINRA arbitration provisions would apply. Not relevant to current architecture.
The following patterns appear in SEC guidance, FTC guidance on deceptive practices, and general negligence mitigation best practices. None are specifically mandated by statute for an unregistered platform operating in Scenario A, but they create a documented record of reasonable care.
| Safety rail | Regulatory / case law basis | Risk-reduction value |
|---|---|---|
| Plain-English confirmation of parsed DSL before save | No direct mandate for unregistered platforms; SEC 2017 robo-adviser guidance requires IA suitability confirmation | High — creates evidence that user reviewed and accepted the interpretation |
| Per-execution notional cap | No direct mandate; analogous to broker-dealer suitability requirements under Reg BI | Medium — limits downside of any single misparse |
| Daily max-fire per strategy | No direct mandate; standard risk control in algo-trading compliance | Medium — limits accumulation of misparsed orders |
| Kill-switch in daily digest notification | No direct mandate; analogous to "right to withdraw" disclosures | Medium — demonstrates user control |
| Audit log of every parse + every fire | No direct mandate for unregistered platform; required for registered IAs under Advisers Act Rule 204-2 | High — essential in any litigation or regulatory inquiry |
| Human-readable strategy description shown at activation | SEC 2017 robo-adviser guidance requires disclosure of "how the algorithm works" for registered IAs | High — demonstrates informed consent |
Research finding subject to attorney review. None of these UX patterns are legally mandated for an unregistered Scenario A platform. However, implementing them creates the factual record needed to defend against negligence claims (reasonable safeguards were in place) and may be required if Raxx registers as an IA.
Both Embroker and Vouch offer Tech E&O insurance for fintech startups. As of 2025, Tech E&O premiums softened following the prior hard market cycle, and the CFPB's reduced enforcement posture in 2025 is lowering perceived fintech risk for underwriters.
Standard Tech E&O covers software errors, API failures, and service outages. It does not clearly cover LLM hallucination-driven order misexecution — underwriters treat this as an AI-specific exposure.
AI-specific E&O / AI Insurance (Vouch explicitly offers this product as of 2025) covers: - Model hallucinations generating incorrect guidance that causes client financial loss - Algorithmic bias or discriminatory model outputs - Defense costs and settlement for AI-errors-and-omissions claims
Sources: - https://www.vouch.us/blog/errors-omissions-vs-ai - https://www.embroker.com/coverage/tech-errors-omissions/
Cost ranges for pre-launch fintech with AI trading execution risk: Neither Embroker nor Vouch publishes specific rates for this risk profile. Vouch's published framework cites revenue, industry, criticality of AI system, and governance maturity as the primary rating factors. For context from industry sources (unsourced primary — confirm with broker): - Standard Tech E&O for a pre-revenue fintech: estimated $3,000–$8,000/year for $1M per-occurrence / $2M aggregate limits. - AI-specific E&O with trading execution exposure: estimated $8,000–$20,000+/year at similar limits, with possible exclusions for intentional misuse or user-provided bad input. Confirm pricing with a fintech-specialist insurance broker.
Recommendation to Kristerpher: Before launch, obtain quotes from both Vouch and Embroker specifically noting: (a) LLM-mediated order execution; (b) user-authored strategies; (c) Alpaca and BYOB broker connectivity. Ask each carrier whether their policy covers AI hallucination-driven order misexecution, what the exclusions are, and what governance/controls they require as underwriting conditions. This is not legal advice — consult an insurance broker + your attorney on policy adequacy.
SnapTrade is a B2B API provider. Its developer documentation and site positioning indicate that Raxx (the platform company) is SnapTrade's customer, and the end user (Raxx's subscriber) is SnapTrade's downstream principal for data access purposes. Source: https://snaptrade.com
This structure means Raxx is the contracting party with SnapTrade, and the pass-through liability and indemnification terms in SnapTrade's Developer Terms of Use govern. As of the research date, SnapTrade's full Developer Terms were not publicly available for review; Kristerpher or his attorney should obtain and review the current Developer Terms before any live integration. Key clauses to examine:
Research finding subject to attorney review. The SnapTrade Developer Terms must be reviewed by a contracts/securities attorney before integration. Raxx assumes aggregator-level risk if SnapTrade's terms impose significant indemnification obligations on Raxx for user-initiated data access.
Note on aggregator dependency: Kristerpher indicated Raxx does not want to depend on a single aggregator. Other providers with similar functionality include Plaid (primarily banking, but expanding to investment accounts), MX Technologies, and Finicity (now a Mastercard company). Each has different terms, brokerage coverage, and pricing. An aggregator-agnostic integration layer with an abstraction over multiple providers reduces dependency risk but adds integration complexity. This is a product architecture question, not solely a legal one — flag for software-architect.
GLBA Privacy Rule (15 U.S.C. §§ 6801–6809; FTC implementing rule at 16 C.F.R. Part 313):
The FTC's 2019 consent decree position is that companies whose services facilitate financial operations on behalf of financial institutions may themselves be "financial institutions" subject to GLBA's privacy and data security requirements. A 2018 U.S. Treasury report similarly took the view that data aggregators and consumer fintech application providers accessing consumer financial account and transaction data are financial institutions subject to GLBA. Sources: - https://www.ftc.gov/business-guidance/privacy-security/gramm-leach-bliley-act - https://cdp.cooley.com/fintech-faces-expanded-applicability-of-glbas-privacy-and-security-requirements/
What this means for Raxx: If Raxx is a "financial institution" under GLBA: - It must provide a privacy notice to users at account opening and annually. - It must give users an opt-out right before sharing nonpublic personal information (NPI) with non-affiliated third parties. - It must implement an information security program (Safeguards Rule, 16 C.F.R. Part 314). - Account position data and transaction history pulled from a user's broker account are clearly NPI under GLBA — they relate to the user's financial account transactions and balances.
Consent required for position import + backtesting use: GLBA's opt-out right is not a consent requirement for internal use — Raxx can use NPI for the direct financial services purpose for which it was collected (executing the user's strategy, backtesting) without separate consent. However, sharing NPI with third parties (including SnapTrade, in the data processor chain) requires either a GLBA-compliant privacy notice + opt-out or a service-provider exception (the third party processes data solely on Raxx's behalf under contractual restrictions). The SnapTrade integration should be structured as a service-provider/data-processor relationship with appropriate contractual protections to use the service-provider exception.
CCPA / CPRA — "Sensitive Personal Information" classification:
Under CPRA (Cal. Civ. Code § 1798.140(ae), effective January 1, 2023): "Financial account, debit card, or credit card number in combination with any required security or access code, password, or credentials allowing access to an account" is sensitive personal information (SPI) requiring heightened protections.
Brokerage account positions and transaction history are tied to the user's financial account and access credentials — this data is SPI under CPRA once accessed via the account link. As SPI, CPRA requires: - Right to limit use and disclosure (separate from the general right to opt out of sale) - Cannot be used for purposes beyond the originally disclosed purpose without consent - Heightened disclosure in the privacy notice
Sources: - https://oag.ca.gov/privacy/ccpa - https://www.akingump.com/en/insights/blogs/ag-data-dive/california-expands-definition-of-sensitive-personal-information-covered-under-ccpa
CCPA threshold: CPRA applies to businesses that: (a) have $25M annual gross revenues, OR (b) buy/sell/share personal information of 100,000+ California consumers or households, OR (c) derive 50%+ of revenues from selling/sharing personal information. At early Raxx scale, threshold (a) is unlikely to be met; threshold (b) could be triggered if 100,000 CA users are onboarded. Pre-launch, Raxx should implement CCPA-compliant privacy practices regardless, because retrofitting after crossing a threshold is costly and risky.
Research finding subject to attorney review. Whether Raxx is a GLBA "financial institution" depends on the nature of its services and how it accesses financial data — this is unsettled for fintech aggregator-adjacent platforms and requires a privacy attorney to confirm based on Raxx's specific data flows.
Alpaca's regulatory status (as of 2025): Alpaca Securities LLC is a FINRA-registered broker-dealer and self-clearing firm (as of 2024 it became self-clearing; in 2025 it became a member of OCC, FICC, and Nasdaq). Alpaca holds SIPC membership. Alpaca is the qualified custodian for accounts opened through its Broker API. Sources: - https://alpaca.markets/support/alpaca-registered-broker - https://alpaca.markets/broker
KYC obligations: Alpaca's Broker API offers "KYC as a Service" — Raxx can either white-label Alpaca's KYC flow or run its own CIP (Customer Identification Program) procedures. Per Alpaca's documentation: "Non-registered technology partners can leverage Alpaca's KYC as a service to get to market faster. Partners are responsible for maintaining a robust Customer Identification Program (CIP) and Know Your Customer (KYC) procedures, in accordance with Anti-Money Laundering (AML) regulations." Source: https://alpaca.markets/broker-resources/guide/alpaca-broker-api-guide-kyc-process
This language indicates that when Raxx uses Alpaca's KYC-as-a-Service, Alpaca performs the CIP verification but Raxx retains some KYC/AML responsibility — likely as an "introducing" or "referring" party in the clearing relationship. The exact allocation of KYC obligation depends on the Alpaca Broker API agreement, which Kristerpher's attorney should review.
Reg BI applicability: Regulation Best Interest (17 C.F.R. § 240.15l-1) applies to broker-dealers making recommendations to retail customers. Raxx is not a registered broker-dealer and thus is not directly subject to Reg BI. However, two indirect exposures exist:
Introducing/referring party exposure (unsourced — confirm with securities attorney): Some securities attorneys have argued that a platform that routes users to a specific broker and effectively "recommends" that broker's account-opening flow may have de facto introducer obligations, even without formal B-D registration. This theory has not been broadly adjudicated in a retail fintech context. It is more relevant if Raxx receives revenue-sharing or referral fees from Alpaca for each account opened.
If Raxx ever recommends Alpaca over the user's own broker (BYOB option): Encouraging a user to use Alpaca's default rather than their existing Fidelity/Schwab account could be characterized as a securities-adjacent recommendation. The safer posture is to present Alpaca-default and BYOB as equal options with no platform-driven endorsement of Alpaca beyond "default for new users."
Custody Rule (17 C.F.R. § 275.206(4)-2): As confirmed by LII's text of the rule, custody concerns arise only when an investment adviser holds or has authority over client assets. In the BYOB model, user assets remain at Fidelity/Schwab/Alpaca (the user's broker). Raxx does not hold client funds or securities and does not have withdrawal authority over user accounts. The custody rule does not apply to Raxx under this architecture — but this analysis changes the moment Raxx has any ability to withdraw funds from user accounts for any purpose (fee deduction, etc.). Source: https://www.law.cornell.edu/cfr/text/17/275.206(4)-2
| Scenario | Investment Adviser Registration Required? | Custody Rule Triggered? | Reg BI Exposure? | GLBA/Privacy Required? | CCPA SPI Triggered? |
|---|---|---|---|---|---|
| A: Pure parser, Alpaca default | Likely no | No | No (indirect introducer question) | Yes (if financial institution) | Yes (financial account data) |
| A: Pure parser, BYOB | Likely no | No | No | Yes | Yes |
| B: Parser + scorer, Alpaca default | Ambiguous — likely yes | No | No | Yes | Yes |
| B: Parser + scorer, BYOB | Ambiguous — likely yes | No | No | Yes | Yes |
| C: Strategy proposer, any broker | Likely yes (register required) | No | No (unless Raxx becomes B-D) | Yes | Yes |
| Any scenario + Raxx takes fee from Alpaca per account | Ambiguous — increases registration risk | No | Yes — introducing broker question | Yes | Yes |
No statutory deadlines are triggered today because Raxx is pre-launch and not yet registered as anything. However, the following timing considerations apply:
Before any live-trading feature ships to any user: Securities attorney review of Scenario A posture is required. This is the single blocking dependency for live-trading launch. Estimate: 2–4 weeks for a focused securities attorney engagement.
Before any strategy-scoring feature ships (Scenario B): Attorney review of each proposed scoring output type. This can run in parallel with launch prep but must complete before the feature is enabled.
If Raxx determines registration is required (Scenario B or C): Form ADV preparation and state registration takes 30–90 days depending on state. Do not delay starting this process once the decision is made — operating as an unregistered IA is a violation from day one of operation.
GLBA Safeguards Rule: If Raxx is a GLBA financial institution, the FTC's Safeguards Rule (16 C.F.R. Part 314, amended 2021 effective January 10, 2022) requires an information security program with specific technical, administrative, and physical safeguards. This must be in place before handling any user financial data — including position imports for backtesting. Source: https://www.federalregister.gov/documents/2021/12/09/2021-25735/privacy-of-consumer-financial-information-rule-under-the-gramm-leach-bliley-act
Recommended specialty: Securities law attorney with investment adviser practice and fintech experience. Matthew Crosby (engaged for IP) likely does not cover this — his practice is IP. This requires a separate engagement with a fintech-specific securities attorney. See attorney referral note at bottom of this document.
Top 3 ship-blocking questions:
Scenario A posture confirmation: Given the specific implementation — Claude parses user-authored natural-language strategy into a structured DSL, user reviews and confirms the parsed DSL before activation, Raxx executes against the user's Alpaca or BYOB broker account — does this constitute "advising others as to the advisability of investing in, purchasing, or selling securities" under 15 U.S.C. § 80b-2(a)(11)? Should Raxx seek a no-action letter from the SEC staff?
Scoring outputs — adviser trigger: If Raxx adds any of the following readouts, which individually trigger investment adviser status: (a) PDT $25K rule flag; (b) "this strategy returned X% in backtest"; (c) risk-band label (conservative/moderate/aggressive); (d) "your seed × projected return rate is above typical for this strategy type"?
Alpaca-as-default and referral fees: If Raxx receives any revenue from Alpaca tied to account openings or assets under management, does that create introducing-broker obligations, broker-dealer registration requirements, or investment adviser status? What is the cleanest structure for the Alpaca default-broker relationship that does not trigger B-D registration?
Additional questions for a full engagement:
Is a no-action letter request to the SEC staff cost-effective for a pre-launch startup in Scenario A, or is the posture clear enough that attorney-confirmed analysis is sufficient?
Should Raxx's TOS include an explicit "you are the investment adviser, not Raxx" user acknowledgment? What language survives challenge?
If Raxx eventually adds strategy recommendations (Scenario C), what is the minimum viable compliance program for a state-registered IA: Form ADV, investment policy statement, annual review obligations?
What is the standard for "reasonable compensation" in the context of a registered IA whose fees are subscription-based rather than AUM-based?
Does SnapTrade's role as order-routing aggregator create any B-D obligation for Raxx as the platform directing orders to user's broker via the aggregator?
Recommended specialty: Privacy attorney with GLBA fintech experience. Separate engagement from securities attorney; this is a distinct practice area.
GLBA "financial institution" determination: Based on Raxx's data flows — connecting to user broker accounts via SnapTrade aggregator, importing positions and transaction history, using that data for backtesting — is Raxx a "financial institution" under GLBA? Does the FTC 2019 consent decree position apply?
GLBA Safeguards Rule gap analysis: What technical and administrative controls does Raxx need to satisfy 16 C.F.R. Part 314 before handling user financial account data?
CCPA / CPRA SPI obligations: User brokerage account positions and transaction history imported via aggregator — does this constitute "sensitive personal information" under CPRA § 1798.140(ae)? What is the required handling protocol for SPI, including the "right to limit" disclosure?
Consent architecture for position import: What consent language is required in Raxx's privacy policy and account-linking flow to satisfy both GLBA opt-out requirements and CPRA SPI limitations? Can a single disclosure + consent UI cover both?
SnapTrade as data processor: What contractual protections must the SnapTrade Developer Terms contain to qualify SnapTrade as a "service provider" under GLBA (avoiding third-party sharing restrictions) and a "service provider" under CCPA (avoiding a "sale" of personal information)?
Per project memory, Matthew Crosby is engaged as Raxx's IP counsel. Based on the issues in this document, his practice (IP — trademarks, patents) does not cover:
Recommended: ask Matthew Crosby for a referral to a fintech-specific securities attorney. Attorneys at firms with dedicated fintech securities practices (e.g., Wilson Sonsini, Cooley, Gunderson Dettmer, Morrison Foerster, or a boutique like Hutter Farrelly) are appropriate. For a focused engagement on investment adviser status, a single attorney experienced in SEC IA regulatory work can likely give a preliminary posture opinion in a 2-hour consultation.
Estimated engagement cost (rough, for preparation purposes only — not a quote): - Initial securities attorney consultation (2 hours): $700–$2,000 depending on firm/market - Written posture opinion on Scenario A + key trigger questions: $3,000–$8,000 - If no-action letter is warranted: $15,000–$40,000+ (multi-month SEC staff process) - Privacy attorney GLBA/CCPA gap analysis: $2,000–$6,000 initial engagement - Ongoing compliance counsel if registration is required: $15,000–$50,000+/year
These ranges are based on general market knowledge and are not a substitute for obtaining actual quotes. Confirm with the specific attorney.
Status: research-only. This addendum does NOT constitute legal or tax advice. Before acting on anything below, consult a securities attorney licensed in the relevant federal + state jurisdictions. Research finding subject to attorney review throughout. Added: 2026-04-29 UTC. Sources verified as of that date.
Scenario D is the product framing confirmed by Kristerpher on 2026-04-29 (issue #479, comment ID 4340196567). It is distinct from Scenarios A, B, and C analyzed above.
What Scenario D does:
What Scenario D does not do:
15 U.S.C. § 80b-2(a)(11) defines an investment adviser as a person who, for compensation, engages in the business of:
"advising others, either directly or through publications or writings, as to the value of securities or as to the advisability of investing in, purchasing, or selling securities..."
Source: https://www.law.cornell.edu/uscode/text/15/80b-2
The statute contains two alternative triggers: (i) advice as to value, or (ii) advice as to advisability of investing in/purchasing/selling. Scenario D must be analyzed against both.
A Scenario D notification does not address the value of any security. It does not produce a price target, a fair-value estimate, or a valuation of the underlying instrument. A statement that "your 30-delta iron condors with 45-DTE entry have returned 88.4% on 32 set trades" is a historical fact about the user's own realized returns — not a statement about what any security is worth today or in the future. This prong is not triggered by Scenario D.
Research finding subject to attorney review.
This is where the analysis is most delicate. The question is whether a pattern-match notification, anchored entirely in the user's own past trades, constitutes advice "as to the advisability of" purchasing or selling a security.
The SEC Release IA-1092 (1987) three-prong test requires that a person:
Prongs 2 and 3 are trivially satisfied for Raxx (subscription fees; recurring automated system). The entire analysis turns on Prong 1 — whether Scenario D constitutes "advice."
Source: SEC Release IA-1092 (1987): https://www.sec.gov/files/rules/interp/1987/ia-1092.pdf
The statutory word "advising" implies communication directed toward a conclusion — a recommendation to act or not act. The critical distinction for Scenario D is between:
(i) "Here is your own historical pattern and current conditions match it."
This formulation presents historical facts about the user's own trading behavior. It is analogous to a brokerage account statement that shows realized returns on past trades — which no one argues is investment advice. The platform is functioning as a mirror: it is returning to the user information that was generated by the user's own prior decisions. The platform is not telling the user what to do; it is telling the user what they already did, in the context of conditions that are now statistically similar.
The SEC's information-versus-advice framework, developed in the context of its 2022 Request for Comment on information providers (SEC Release IA-6050, S-7-18-22), identified "significant discretion" — i.e., the provider making active judgments that drive the user toward a specific outcome — as the threshold for crossing from information into advice.
Source: https://www.sec.gov/rules-regulations/2022/06/s7-18-22
A Scenario D notification as formulated (i) exercises no such discretion: the user authored the strategy; the user generated the return; the notification merely identifies that today's conditions resemble the conditions on the prior 32 occasions. The platform is performing pattern recognition, not reasoning.
(ii) "Your historical pattern suggests this trade has 88% odds."
This formulation makes an inferential leap from historical correlation to a probabilistic prediction about a current trade. It attributes the historical win-rate to the current setup as if it were predictive — a forward-looking probability statement. The moment the notification shifts from "here is what happened to you before" to "here is what will probably happen if you trade now," it has crossed from information about the user's past into advice about future actions. This is the form the statutory "advisability" prong is designed to reach.
Research finding subject to attorney review: The line almost certainly runs through the language of the notification copy. Formulation (i) has the stronger case for non-advisory status. Formulation (ii) is likely advisory. The actual Raxx notification mockup language ("for the last 30 days, your 30-delta iron condors...have returned 88.4% on 32 set trades") reads as formulation (i): it states historical fact, in the past tense, without a forecast verb. However, the phrase "setup matches your 90-day winning profile" introduces an interpretive conclusion — "winning" is an evaluative label applied by the platform, not a neutral statistical report. A conservative attorney would flag "winning profile" language as potentially advisory and recommend replacing it with neutral descriptive language (e.g., "conditions match the market parameters on your last 32 iron condor entries").
Lowe v. SEC, 472 U.S. 181 (1985): Source: https://supreme.justia.com/cases/federal/us/472/181/
Lowe's holding: the publisher exclusion under § 80b-2(a)(11)(D) protects publishers of bona fide, general-circulation financial publications that deliver identical, impersonal content to all subscribers, because such publications do not create the "fiduciary, person-to-person relationship" characteristic of investment advisory relationships.
Why Lowe's holding does not directly apply to Scenario D:
Scenario D is personalized by definition. The notification content is unique to each user — it reflects that specific user's 32 trades, not a market-wide commentary. General-circulation exclusion requires content identical to all subscribers; Scenario D inverts that: the content is generated from that user's data and is meaningful only to that user. Lowe's holding protects the newsletter; Scenario D is not a newsletter.
Why Lowe's reasoning may partially support Scenario D's case:
Lowe's conceptual framework identified the "fiduciary, trust-and-reliance relationship" as the thing the IA Act was designed to regulate. The Court wrote that the Act centered on "the business of rendering personalized investment advice" — but the operative word is "advice." The Court distinguished advisers (who direct client action) from publishers (who provide information that readers may or may not act upon).
Scenario D's user retains complete discretion. There is no client-adviser trust-and-reliance dynamic: Raxx cannot see what the user ultimately decides; it cannot be relied upon to monitor the user's portfolio; it does not have a fiduciary relationship with the user in any contractual sense. The notification is delivered; the user decides. This is closer to the "reader/publisher" dynamic than the "client/adviser" dynamic.
The critical Lowe distinction that does NOT help Raxx:
Lowe's personalization test was about whether different subscribers received different content tailored to their individual circumstances. Under that framing, Scenario D fails the Lowe test: each user receives different content because each user's trading history is different. This is the precise personalization that Lowe identified as characteristic of the adviser relationship.
Research finding subject to attorney review: Lowe's holding is unavailable to Scenario D because the publication is personalized per-user. Lowe's reasoning is partially supportive because the functional relationship lacks the trust-and-reliance quality the IA Act targets and the user retains complete decisional autonomy. A securities attorney should assess whether the "no fiduciary relationship" argument — drawn from Lowe's reasoning but not its holding — has traction before a court or regulator given the per-user personalization.
The prior Scenario B analysis called "feasibility readouts" ambiguous because scoring a strategy implies a judgment about its quality. Scenario D's 88.4% return figure functions contextually as a score of the current setup — it signals "this looks good" — even though it is derived exclusively from the user's own historical data rather than universe-wide data or platform-generated analysis.
The data-source question under the IA Act:
The Investment Advisers Act does not, on its face, distinguish between advice computed from universe-wide data and advice computed from user-specific data. The statute asks whether the output constitutes "advice as to...the advisability of...investing in, purchasing, or selling securities." The source of the data informing that advice is not an enumerated element.
However, source-of-data matters for two analytical reasons:
1. Who authored the underlying judgment.
When the score is derived from external market data, the platform has exercised its own analytical judgment to produce a conclusion about what the user should do. When the score is derived exclusively from the user's own prior trades, the analytical judgment (i.e., "I am going to keep trading iron condors under these conditions") was made by the user on 32 prior occasions. The platform is pattern-matching against a conclusion the user already reached — repeatedly — and is reporting back: "you have historically been consistent here." This is qualitatively different from the platform independently concluding "iron condors are advisable here."
The SEC's 2022 IA-6050 framework identified "significant discretion" exercised by the provider as the indicator of advisory activity. The discretion exercised in Scenario D is minimal: the platform selects a threshold for statistical match (a product design decision) but does not evaluate whether the strategy is good, appropriate, or advisable for the user. That evaluation was already made by the user.
2. The no-new-information argument.
The pattern-match notification conveys information the user implicitly already has access to: the user placed those 32 trades; they know their own results. Raxx is aggregating and surfacing that information in a timely way. Surfacing information the user already owns — their own trade history — is analytically weaker as "advice" than introducing new information (external market analysis, strategy recommendations) that the user did not already have.
What the IA Act likely does care about:
Even where data is user-sourced, if the notification is functionally a prompt to act — if the product design is to fire the notification specifically when conditions make the trade attractive and to suppress it otherwise — then the notification is functionally making a recommendation regardless of whose data underlies it. The IA Act looks at substance, not form. An attorney will ask: does Raxx send the notification when the pattern match is strong, and not when it is weak? If yes, the platform is implicitly ranking the attractiveness of potential trades and surfacing the ones that look good — which resembles scoring regardless of data source.
Research finding subject to attorney review: The user-own-data origin meaningfully strengthens the case for Scenario D versus Scenario B's universe-wide feasibility scoring, because the underlying judgment was authored by the user. But the source-of-data distinction is not a statutory safe harbor. The critical question is whether the notification is functionally making a recommendation, regardless of which dataset underlies it. The notification copy and the triggering logic together determine the answer.
Short answer: the IA Act does not care about statistical sufficiency. The line stays at "is it advice" regardless of sample size.
The Investment Advisers Act has no floor on the reliability of advice that triggers registration. Advisers who give terrible, statistically meaningless advice are still investment advisers. Statistical noise does not exclude a communication from being "advice."
Why sample size matters operationally:
However, sample size matters for a different reason: it affects whether the notification is materially misleading and whether the platform is making an implicit predictive claim it cannot substantiate.
N = 1: A single prior trade is not a "profile." Calling it a "winning profile" with one data point is misleading — it implies a pattern that does not exist. The SEC's Marketing Rule (Rule 206(4)-1 under the Advisers Act) prohibits registered advisers from making misleading performance claims. For an unregistered platform, FTC Section 5 unfair and deceptive practices prohibitions apply, as do California's UCL (Bus. & Prof. Code § 17200). Surfacing a "your winning profile" notification on 1 or 2 trades creates consumer protection risk independent of the IA Act.
N = 5: Still thin. Survivorship bias is substantial at this sample. Any quantitative claim (88.4% win rate) derived from 5 trades has such a wide confidence interval as to be potentially misleading if presented without material caveats.
N = 30+: Industry practice for options strategies generally holds that 25–50 occurrences begin to produce statistically observable patterns, though this varies by strategy type and market regime. At this range, the "your historical profile" framing is defensible as a statement about historical behavior, provided it is accompanied by clear disclosure that past performance does not predict future results.
N = 0: No notification fires. This is not a question of law; it is a product question. The pattern-match feature is undefined where the user has no history. The platform should enforce a minimum threshold for this reason.
Research finding subject to attorney review: The IA Act does not impose a statistical floor. But minimum-trade thresholds for firing the notification matter for consumer protection (FTC and state unfair practices law) and for the internal coherence of the "user-authored strategy" argument — a "profile" built on 1 trade is not meaningfully user-authored; it is a single data point elevated into a recommendation by the platform. A minimum threshold of 25–30 prior instances (with standard disclaimer language) is the most defensible floor from both a consumer-protection and IA Act characterization perspective. Below 10, the "profile" framing is difficult to sustain.
What the SEC has said about adviser-status disclaimers:
The SEC has consistently held that disclaimer language alone cannot disclaim investment adviser status if the substance of the activity meets the definition. An adviser who calls its advice "educational content" in its TOS is still an adviser if the substance of the output directs users toward specific securities transactions.
Source: SEC Release IA-1092 implicitly; see also SEC 2017 Robo-Adviser Guidance (IM Guidance Update 2017-02): https://www.sec.gov/newsroom/press-releases/2017-52
What a Scenario D disclaimer must accomplish:
A disclaimer for Scenario D should not attempt to disclaim IA status by calling the output "not advice." That approach does not survive scrutiny if the output is, in fact, advice. Instead, the disclaimer's job is to:
Specific disclaimer language elements that have legal load-bearing weight:
| Element | Why it matters | Draft language |
|---|---|---|
| Authorship attribution | Establishes the strategy as user-authored, not platform-generated | "This notification reflects your own prior trading history, not a recommendation from Raxx." |
| Historical-only characterization | Blocks the forward-looking "advisability" framing | "Past performance reflects trades you have already placed. It does not predict future results or constitute a forecast." |
| Non-advisory framing (substantive, not label) | Correct if notification is in fact informational | "Raxx does not advise you on whether to trade. This notification tells you that current market conditions resemble the conditions present during your own prior trades." |
| No reliance instruction | Undercuts "trust and reliance" element of advisory relationship | "Do not rely on this notification as a basis for any trading decision. You are solely responsible for your trading decisions." |
| Statistical caveat | Required for consumer protection; also weakens predictive inference | "Historical win rates are based on a limited sample and vary with market conditions. No statistical guarantee of future results is made or implied." |
Disclaimer language that is labeling fiction and will not protect Raxx:
Research finding subject to attorney review. Every disclaimer element above should be reviewed by a securities attorney before launch. The substantive characterization of what Raxx does — not the label on the output — determines adviser status. The disclaimer's job is to accurately describe the substance, thereby reinforcing the characterization that Raxx is not advising.
The following guardrails, which Kristerpher specifically identified, are evaluated below for how much legal weight each carries in the IA Act exemption argument.
Load-bearing weight: High.
This is the single most important structural guardrail. It operationalizes the core legal argument: that the "strategy" referenced in the notification was authored by the user, not the platform. Without a meaningful minimum, the platform is in the business of surfacing pattern matches on statistically thin data — which looks more like generating recommendations than recognizing the user's own established patterns.
The threshold should be set with two objectives: (a) statistical defensibility of the "your profile" framing (recommend no fewer than 25 prior instances, ideally 30+); and (b) documented in product design decisions so that, in any regulatory inquiry, Raxx can demonstrate the threshold was a principled product decision, not an arbitrary one.
The threshold also needs to be documented in Raxx's terms of service or product disclosures: "Pattern notifications fire only when you have 25 or more prior trades matching the identified setup."
condition match" — no recommendation verbs
Load-bearing weight: Very high — this is the pivotal structural guardrail.
The legal analysis in D.1 above concludes that the line between non-advisory and advisory runs through the language of the notification. Prohibiting recommendation verbs ("consider," "this is a good time to," "you should," "recommended setup") from the notification copy is essential.
This guardrail requires: - A copywriting standard embedded in product design: no notification copy may contain forward-looking recommendation language. - Review of all AI-generated notification text before any template ships, verifying no recommendation verbs are used. - A hard requirement that the notification reference past-tense factual statements (what the user has done) rather than present-tense evaluations (what the current setup is worth).
Note: This guardrail must survive any future changes to the notification copy. If Kristerpher later wants to A/B test more "helpful" notification language that is closer to formulation (ii) in D.1.4, the IA Act risk analysis must be re-run before that copy ships.
Load-bearing weight: High.
Any forward-looking return statement — "this setup has historically returned X, implying Y upside" — converts the notification from historical reporting to predictive advisory output. This guardrail must be absolute. The 88.4% figure in Kristerpher's mockup is a historical return (past-tense); it is defensible. The same figure framed as "expected return: 88.4%" is not.
This also implicates the SEC Marketing Rule's prohibition on misleading performance claims for registered advisers, and FTC Section 5 for unregistered platforms.
Load-bearing weight: Highest (dispositively important for the IA Act analysis).
This is the most structurally important guardrail for the Investment Advisers Act analysis. Registered investment advisers are characterized by the element of discretion — they direct client assets. Raxx explicitly does not have discretionary authority over user accounts. If the feature design were to include any "auto-act on this notification" path — even opt-in — the platform would be a step away from discretionary advisory status.
Kristerpher has confirmed AI is NOT in the order-firing path and that automation handles execution deterministically. This confirmation should be documented in the product specification, in TOS, and in any eventual disclosure document.
The specific requirement is: between the notification firing and any order executing, at minimum one explicit user-initiated action must occur (a click, a confirmation, a separate navigation step). The notification must not be co-located with an "execute" button that reduces the user's deliberate decision to a single click on the notification itself.
| Guardrail | Load-bearing weight | Minimum specification |
|---|---|---|
| N >= 25 prior executions threshold | High | No notification below 25 prior matching trades; document in TOS + product spec |
| No recommendation verbs in copy | Very high | Copywriting standard enforced in product design; attorney review of final templates |
| No forward-looking return forecast | High | All return figures must be historical (past-tense); no "expected" or "implied" framing |
| Explicit user click required; no auto-act | Highest | At minimum one deliberate action between notification and order; AI NOT in order path (already confirmed) |
Summary of the analysis:
Scenario D occupies the strongest defensible position in the A-B-C-D spectrum:
| Scenario | Posture | Confidence |
|---|---|---|
| A: Pure parser | Likely exempt | Moderate |
| B: Parser + scorer | Ambiguous (leans registration required) | Low |
| C: Strategy proposer | Likely required to register | High |
| D: Pattern recognizer | Ambiguous (leans exempt) | Moderate |
Why Scenario D leans exempt:
Why it is not clearly exempt:
The scope limit Kristerpher confirmed (2026-04-29): AI is NOT in the order-firing path.
This is the single most important fact for the IA Act posture. The clearest path to investment adviser status is discretionary control over execution. Because AI augments understanding only, and execution is handled deterministically by user-authorized rules, Raxx has no discretionary authority over user accounts in the advisory sense. This scope confirmation is load-bearing for the exemption argument and must be maintained as a hard product constraint, not merely a preference.
What Scenario D does for the "ship blocker" question:
The ship-blocker framing in the original brief was: "Does Raxx's AI propose, score, or classify strategies on behalf of the user?" For Scenario D, the answer to that question is substantially cleaner than for B or C:
Scenario D meaningfully reduces — though does not eliminate — the ship-blocker risk. If Raxx launches with (a) the operational guardrails in D.6, (b) the notification copy complying with D.1.4 formulation (i), and (c) the confirmed scope limit that AI does not touch the execution path, then the IA Act argument for non-adviser status is materially stronger for D than for any prior scenario. It is still untested; it still requires attorney signoff before launch; but it is no longer in the "high risk of registration required" category.
Research finding subject to attorney review.
Recommended pro: Securities attorney with investment adviser practice and fintech experience. (Separate engagement from Matthew Crosby, who is IP counsel.)
Core posture question: Under 15 U.S.C. § 80b-2(a)(11), does a notification that surfaces statistical information about a user's own prior trade history, when current market conditions match the historical pattern, constitute "advising others as to the advisability of investing in, purchasing, or selling securities"? Specifically: does the "own-data" origin of the statistical pattern affect the analysis, and does the user's complete discretion over execution affect the analysis?
Notification copy review: The specific notification copy must be reviewed by a securities attorney before launch. The legal conclusion depends materially on the language. Provide the attorney the exact notification templates, including all variant copy, and ask: which versions of this text constitute advice, and which do not? (This is the most critical, and most specific, attorney task for Scenario D.)
Triggering logic review: Does the selective nature of the notification — it fires when the match is strong, not when it is weak — constitute an implicit recommendation about when to trade? Is there a product design change (e.g., always showing the user their pattern statistics in a persistent dashboard, with a separate "current conditions" indicator, rather than a triggered notification) that would strengthen the non-advisory framing?
No-action letter cost-benefit for Scenario D: Is a no-action letter request to SEC staff warranted specifically for the Scenario D pattern-recognition model? Or is the non-adviser argument strong enough on the existing record that attorney-confirmed analysis is sufficient?
The "winning profile" language: Ask the attorney specifically whether the phrase "winning profile" — or any evaluative label applied to the historical pattern — is advisory, and what neutral language accomplishes the same product goal.