50 patterns in this category
When a Pre-Trade Risk-Check Tool Call Times Out or Returns a Transport-Level Error Instead of a Structured Response, the Agent's Next Reasoning Step Generates a Schema-Conforming Risk-Check Result From Its Own Expectations of What the Tool Usually Returns, and Proceeds as if the Call Had Succeeded
A Hard Constraint the Client States Early in a Long, Multi-Turn Advisory Session (e.g., "No Fossil-Fuel Holdings," "No Leveraged Products") Falls Outside the Model's Effective Context Window by the Time a Later-Turn Recommendation Is Generated, and the Agent Recommends a Security That Violates It Without Re-Checking Against the Original Constraint List
An Advisor-Facing Agent That Processes Multiple Clients Within a Single Continuous Session Carries a Computed Suitability Parameter (Risk Tolerance, Tax Bracket, Liquidity Need) From One Client Forward Into Its Reasoning for the Next Client, Without Any Literal Data From the First Client Appearing in the Second Client's Output
After a Trading Venue Renames a Field in Its Order-Routing API Response (e.g., `available_qty` to `remaining_quantity`), an Agent Consuming the Response Maps the New Field to the Old Expected Key by Semantic/Fuzzy Similarity Rather Than Failing Loudly, Silently Misreading a Different Field as the Quantity Value
When a Security's ISIN/CUSIP Is Absent or Malformed in the Reference-Data Feed, an Agent Tasked With Enriching Trade or Position Records Generates a Plausible-Looking Identifier by Pattern-Completing From the Security's Name and Exchange, Instead of Flagging the Record as Unresolved
An Agent Given an Explicit, Durable Client Constraint Infers a Broader or Conflicting Preference From the Client's Transaction History During Multi-Step Recommendation Generation and Silently Supersedes the Explicit Constraint Instead of Flagging the Conflict
When an Agent Is Used to Canonicalize Free-Text Counterparty or Issuer Names Into a Standard Form, Its Generative Rewriting Silently Drops or Alters a Legal-Entity Suffix (Ltd, LLC, GmbH, Inc, N.V.) That Materially Changes Which Legal Entity — and Therefore Which Counterparty-Risk Profile — the Record Refers To
An Agent Computing Net Asset Value Through a Multi-Hop Sequence of Tool Calls Re-Types the Intermediate Numeric Result as Text Between Steps Instead of Passing the Exact Value Forward, Introducing Rounding and Truncation Drift That Compounds Across the Chain
An Agent Resolving a Margin Call Through a Multi-Step Tool-Calling Sequence Continues Reasoning About Remaining Liquidation Needs Using the Account Equity and Margin-Usage Figures It Read Before Its Own Preceding Sell Order Executed, Rather Than Re-Fetching Current State After Its Own Action Changed It
After Autonomously Executing or Recommending a Trade, an Agent Asked to Produce the Compliance/Suitability Rationale Selectively Retrieves and Surfaces Only the Evidence That Corroborates the Trade It Already Made, Omitting Contradicting Signals It Had Equal Access To
An Agent Given Autonomy to Draft Routine, Recurring Regulatory Filings (Quarterly Amendments, Periodic Disclosures) Uses Its Own Prior-Quarter Filing Output as a Template and Implicit Ground Truth for Consistency Checking, So a Factual Error Introduced in One Quarter Is Not Caught but Actively Re-Validated and Carried Forward in Every Subsequent Quarter
A Sanctions-Screening Tool Returns a Genuine Positive Match, but a Downstream Abstractive-Summarization Step — Written to Produce a Concise Case Summary for the Compliance Analyst — Paraphrases the Hit Into Vague, Non-Actionable Language That Reads as "No Significant Findings"
Model Optimizes Against a Generic Benchmark That Does Not Match the Client's Actual Risk Profile, Goals, or Constraints
Agent Fails to Adjust Prices, Positions, or Historical Series for Splits, Dividends, Mergers, or Spin-Offs
Agent Attributes Risk, Exposure, or Performance Data to the Wrong Entity Within a Corporate Family Due to Incomplete Parent-Subsidiary Mapping
Assumed Diversification Fails When Correlations Spike (Tail Dependence Not Captured)
Model Recommends International Assets Without Accounting for Currency Risk; FX Movements Dwarf Asset Returns
A Compliance Agent Classifying a Newly Launched Financial Product Against the Applicable Regulatory Rule Set Selects the Rule Whose Description Is Most Lexically or Embedding-Similar to the Product's Marketing Description, Rather Than Matching on the Product's Structured Regulatory Classification Code, Applying the Wrong Rule Set to a Structurally Different Product
A Data-Quality Agent Deduplicating Issuer Records Across Multiple Source Feeds Using Embedding Similarity Over Issuer Names, Rather Than Matching on a Unique Identifier Such as LEI or CUSIP Issuer Code, Merges Two Distinct Issuer Entities With Coincidentally Similar Names Into a Single Record, Corrupting Downstream Holdings and Exposure Calculations
A Transaction-Cost-Analysis Agent Benchmarking a Trade's Execution Quality Against Historical Comparable Orders Selects the "Most Similar" Past Order Using Embedding Similarity Over Free-Text Order Notes, Rather Than Matching on Instrument, Order Size, and Time-of-Day Liquidity Regime, Producing a Benchmark That Was Never Executed Under Comparable Conditions
A Market-Data Freshness-Monitoring Agent Checking Whether an Illiquid Instrument's Price Is Plausibly Current Selects a "Comparable" Reference Instrument Using Embedding Similarity Over Free-Text Descriptions Rather Than Matching on Sector, Duration, and Credit-Quality Attributes, Producing a Freshness Benchmark That Moves Differently From the Instrument Being Checked
Agent Incorporates Self-Reported Corporate ESG Disclosures Into Portfolio Scoring at Face Value, Without Adjusting for Known Greenwashing or Disclosure-Quality Variance
An Agent Generating a Client-Facing Fund Comparison or Recommendation Document Retrieves Most Required Disclosure Fields (Expense Ratio, Standardized Performance, Minimum Investment) From the Firm's Actual Fund Documents, but When Retrieval Misses One Specific Field for One Fund, Fills the Gap With a Plausible-Sounding Fabricated Number Rather Than Marking the Field as Unavailable
A portfolio-recommendation model recommends factor exposures with no awareness of market-wide crowding, causing correlated unwind risk when crowded positions all retreat at once.
An Execution Agent Receives a "Filled" or "Success" Status From an Order-Management/Broker Tool and Narrates the Trade as Correctly Executed Based on That Status Alone, Without Comparing the Confirmation Payload's Own Symbol, Quantity, Side, and Price Fields Against the Order It Actually Submitted
Agent Relies on a Customer's Initial KYC Risk Rating Without Triggering a Refresh When Materially Risk-Relevant Account Activity Changes
Model Underestimates Risk from Leverage; Recommends Leveraged Strategies Without Adequate Risk Buffer
Model Recommends Illiquid Assets; Prices Assumed Frictionless but Actual Bid-Ask Spreads Destroy Returns
Model Uses Future Information in Backtest That Wouldn't Be Available in Real-Time Trading
Agent Sizes and Times Orders Without Accounting for Its Own Price Impact, Especially Across Correlated Positions or Repeated Trades
Model Handles Missing Data Incorrectly (Mean Imputation, Deletion); Introduces Bias or Information Loss
A Data-Cleansing Agent Notes in Free Text That a Field It Cleaned Was Ambiguous or Low-Confidence, but the Structured Cleansed Record Handed Off to a Downstream Risk- or Pricing-Consuming Agent Has No Field for Cleansing Confidence, So the Downstream Agent Treats the Value as Fully Reliable
An Account-Opening Agent Notes in Free Text That a New Client's Stated Residency and the Jurisdiction Implied by Their Funding Source Do Not Match, but the Structured Client Profile Handed Off to the Compliance-Screening Agent Has No Field for a Jurisdiction Conflict, So the Client Is Screened Only Under Their Stated Residency's Rules
A Pre-Trade Risk Agent Notes in Free Text That an Order, Combined With Existing Positions, Would Push a Concentration or Leverage Limit Into a Marginal Breach Under a Plausible Adverse Price Move, but the Structured Pre-Trade Check Result Handed Off to the Execution Agent Has Only a Pass/Fail Field on Static Current-State Limits, So the Execution Agent Routes the Order as Clear
A Market-Data Ingestion Agent Notes in Free Text That a Price Feed Showed No Update Across an Unusually Long Window and May Be Stale, but the Structured Price Record Handed Off to the Downstream Valuation Agent Has No Field for Staleness Suspicion, So the Valuation Agent Treats the Last Received Price as Current
Portfolio or Strategy Compliant in One Jurisdiction but Violates Rules in Another; Model Doesn't Handle Jurisdiction-Specific Rules
Model Uses Outdated Regulatory Rules; Recommendations Violate New Regulations Effective Retroactively or Overlooked
Model Optimized for Historical Regime (Bull Market) Fails in Regime Change (Bear Market, Stagflation)
Backtests and Live Recommendations Use Data That Was Not Actually Available at the Decision Timestamp (Look-Ahead via Restated/Revised Data)
Agent Recommends Target Allocations but Fails to Account for Execution Lag, Letting Portfolios Drift Far From Target Before Rebalancing Triggers
Model Overweights Recent Performance; Recommends Assets That Performed Well Recently but Are Mean-Reverting
Agent Screens Counterparties Against a Cached or Infrequently Updated Sanctions/Watchlist Snapshot, Missing Recently Added Entities
Execution Agent's Pre-Trade Cost Model Systematically Understates Realized Slippage, Especially for Larger or Illiquid Orders
A Post-Trade Analytics Agent Generating a Free-Text Explanation for Elevated Execution Slippage Constructs a Plausible Causal Story Linking a Temporally Co-Occurring but Unrelated News Event to the Slippage, When the Actual Driver Was the Execution Algorithm's Own Order-Slicing or Venue-Selection Behavior
Agent Generates Recommendations or Risk Calculations Using Cached or Delayed Price Data Without Detecting Staleness
Model Trained Only on Survived Assets; Ignores Assets That Failed or Were Delisted
Model Optimizes for Mean-Variance Efficiency While Ignoring Fat-Tail and Black-Swan Risk
Model Recommends Portfolio Without Accounting for Tax Impact; After-Tax Returns Much Lower Than Pre-Tax Predictions
Agent Routes Orders to the Venue With the Best Quoted Price Without Accounting for Realistic Fill Probability, Rebate Structure, or Information Leakage Risk at That Venue
Agent Executing Orders Across Multiple Related Accounts or Strategies Inadvertently Creates Wash-Trade-Like Patterns by Crossing Its Own Buy and Sell Orders Without a Cross-Order Check
Financial-services agents face compound failures across data quality, market data timeliness, portfolio construction, regulatory compliance, and execution—each independent in mechanism but tightly coupled in impact, where data quality degradation propagates into pricing, risk calculations compound the data errors through poor correlation modeling, and execution errors realized the corrupted recommendations. None of the five goals reliably catches errors made at earlier stages: a data-quality team fixing entity resolution errors has no visibility into whether downstream risk or compliance calculations would have caught the error, and a portfolio-construction agent optimizing for historical Sharpe ratios has no awareness that its backtest data contains look-ahead bias or survivorship bias that will make live performance substantially worse. The gap is structural: financial-services systems are built as sequential pipelines (data → pricing → risk → recommendations → execution) where each stage assumes upstream data and models are correct, yet each stage independently fails for different reasons.
| Goal | Covers | Patterns |
|---|---|---|
| Data Quality (GitHub) | Entity resolution, hierarchy mapping, missing-data imputation, point-in-time temporal accuracy, multi-agent handoff confidence drops | 5 |
| Market Data Freshness (GitHub) | Stale price feeds, corporate-action adjustments, freshness benchmarking, ingestion-to-valuation handoff staleness signals | 4 |
| Portfolio Recommendation Accuracy (GitHub) | Backtesting biases (survivorship, look-ahead, overfitting, recency), risk modeling (correlation, tail risk, leverage, factor crowding), implementation friction (rebalancing lag, taxes), data quality (ESG greenwashing, currency blindness, liquidity mispricing) | 15 |
| Regulatory Compliance (GitHub) | Outdated rules, multi-jurisdiction conflicts, KYC refresh staleness, sanctions-list freshness, product classification mismatches, jurisdiction-flag handoff drops | 6 |
| Trading Execution (GitHub) | Slippage underestimation, market-impact blindness, venue-selection mismatches, fill-confirmation validation gaps, wash-trade detection, TCA narrative spuriousness, risk-limit handoff drops | 8 |
Total: 38 patterns
Five financial-services goals are tightly coupled pipelines, not independent concerns, because errors cascade downstream:
Data Quality → All Other Goals. Clean entity identity, accurate hierarchies, and point-in-time-correct historical data are prerequisites for every downstream stage. A mismerged issuer entity cascades into incorrect risk aggregation (regulatory compliance exposure), wrong pricing (portfolio recommendations miss the true concentration risk), and wrong position records (execution agents inherit corrupted data).
Market Data Freshness → Pricing/Valuation/Risk/Execution. Stale prices corrupt everything downstream: risk calculations overestimate time available to liquidate (stale prices don’t reflect current market conditions), execution agents underprice the cost of moving positions, and portfolio managers make rebalancing decisions based on stale valuations.
Portfolio Recommendation Accuracy → Regulatory Compliance + Execution. Backtests inflated by survivorship or look-ahead bias recommend strategies that fail in live markets but appear compliant under the test conditions. Strategies optimized for one jurisdiction’s rules may violate another’s. Execution costs not modeled in the recommendation lead to realized slippage that erodes or reverses the intended alpha.
Regulatory Compliance → Execution/Risk. Compliance rules gate which strategies are permissible; compliance failures are high-severity but lower-frequency. When a strategy passes compliance screening because rules are outdated or misapplied (multi-jurisdiction gap), the strategy can expose the firm to sanctions or enforcement.
Trading Execution is the final realization point: execution failures are where data-quality corruption, pricing errors, and portfolio misspecification finally surface as unintended positions or regulatory violations, or where execution cost and market impact erase recommendation alpha. Execution feedback (realized slippage vs. estimated, order fills vs. intended) is the signal that should drive recalibration of earlier stages.
To localize a failure by symptom: Recommendation performance lags backtest significantly → check Portfolio Recommendation (backtesting bias or execution friction); Risk calculations don’t match actual exposure → check Data Quality (entity resolution) and Market Data Freshness (stale hierarchy or prices); Actual regulatory enforcement or exam finding → check Regulatory Compliance (outdated rules, multi-jurisdiction gaps); Trade executes at materially worse cost than estimated → check Market Data Freshness (stale prices), Trading Execution (slippage model or venue routing), and Portfolio Recommendation (liquidity mispricing).
No. Each goal has structural gaps that model improvements alone cannot fix: (1) Data Quality requires versioning and provenance tracking, not better embeddings for entity resolution; (2) Market Freshness requires event-coupled cache invalidation and mandatory SLA gating, not better prediction of staleness; (3) Portfolio recommendations require multi-scenario stress testing and full-cost accounting, not better Sharpe-ratio optimization; (4) Compliance requires versioned rule registries and multi-jurisdiction gating, not better rule interpretation; (5) Execution requires real-time order-book data and pre-trade cross-account checking, not better slippage estimation.
Implement multi-goal audit: for a held-out test quarter, (1) audit data quality (entity deduplication, hierarchy versioning) against source documents, (2) verify market data freshness against independent feeds, (3) backtest portfolio recommendations on both historical and crisis periods checking for biases, (4) check regulatory rule versions against effective dates and all applicable jurisdictions, (5) measure execution quality (realized slippage vs. estimate, fill-vs-intended reconciliation). Report failures per goal; focus remediation on the goal with highest impact.
Separate agents with explicit handoff schemas and gating. Integration increases model size for marginal gain; separation enables: (1) independent validation of each goal’s outputs, (2) faster retraining when rules or data change, (3) clearer audit trails for regulatory compliance. Require structured handoffs (not free text) and gating: portfolio recommendations must pass compliance checks before reaching traders; execution confirmation must validate against original intent.
Data Quality has highest leverage: almost every downstream failure in compliance, pricing, and execution chains back to data-quality issues. Solving data-quality prevents cascading errors more effectively than local fixes to downstream stages.