High-Throughput FX Rate Time Series Anomaly Detector with LlamaIndex
Global treasury departments rely on accurate foreign exchange rate forecasting and feeds. Malicious price manipulation attacks or feed corruption can trigger catastrophic automated currency conversions and massive liquidations if uncorrected. In this challenge, you will build a time-series FX rate anomaly detection agent using LlamaIndex. The agent will index streaming currency pair feeds, calculate dynamic Bollinger Bands and volatility spreads, and detect malicious rate spikes or synthetic price freezes in real time.
What you are building
The core problem, expected build, and operating context for this challenge.
Build an FX time-series anomaly detection engine using LlamaIndex to protect treasury workflows from price feed corruption.
How work is evaluated
Evaluates rate anomaly detection accuracy and false alarm rate on synthetic FX streams.
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
circuit_breaker_pass
Verifies circuit breaker triggers on >5% instant jump
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
detection_recall
Recall percentage for synthetic price corruption events • target: 0.98 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Construct a dynamic LlamaIndex vector and tabular index over streaming FX rate ticks
Implement time-window query tools for detecting rate deviations exceeding 3 standard deviations
Trigger automated oracle feed suspensions when anomaly confidence exceeds thresholds
Generate detailed post-incident forensic summaries for quantitative risk officers
Reference links and supporting material
Synthetic tick-by-tick multi-currency FX spot rates with injected oracle manipulation spikes.
Challenge setup
Evaluates rate anomaly detection accuracy and false alarm rate on synthetic FX streams.
Challenge input
JSON array of FX rate ticks with timestamps
LlamaIndex
Data framework for structured and unstructured vector indexing.
Evaluated output
JSON specifying anomaly_detected, tick_timestamp, and circuit_breaker_triggered
- Verifies circuit breaker triggers on >5% instant jump
- Recall percentage for synthetic price corruption events • target: 0.98 • range: 0-1
- Benchmark: FinMarket-Anomaly-Bench
- Detection Recall target: 0.98
- 1 public reference case
- Python execution harness
- Python sandbox (unavailable on Versalist)
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
- Llama Index · Required
- LlamaIndex · Optional
- Llama 3.2 · Optional
Evaluation contract
- circuit_breaker_pass · Weight 1
- detection_recall · Weight 1
Metrics
- detection_recall · Target 0.98
Benchmarks
- FinMarket-Anomaly-Bench
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON array of FX rate ticks with timestamps
Output: JSON specifying anomaly_detected, tick_timestamp, and circuit_breaker_triggered
Execution harness: Python
Evaluation tools
- LlamaIndex · Agent Framework
Recipe freeze marker
No recipe freeze time is recorded.