Agentic Trading Challenge

Downloads

Download the starter code and challenge specification.

⇩ Starter Code (.tar.gz) ⇩ Challenge PDF

Overview

Build an agentic stock analysis workflow in Go. Your program accepts a stock ticker symbol, uses the DeepSeek Chat LLM to autonomously call tools (get price, fetch news, analyze sentiment, retrieve historical prices, financial metrics, insider trades, and analyst ratings), and produces a structured Buy / Sell / Hold recommendation as JSON.

Key constraint: No AI frameworks (LangChain, LlamaIndex, etc.). You must implement function calling manually — parse tool calls from raw LLM output, execute them, and inject results back into the conversation.

High-level flow:

CLI args (symbol) → LLM ⇔ Tools → Final AnalysisReport (JSON)

What’s Provided vs. What You Implement

Component	Status	File
Data types	Provided	`internal/models/types.go`
Mock tools (7 functions)	You Implement	`internal/tools/tools.go`
LLM client (HTTP call)	Provided	`internal/llm/client.go`
CLI entry point	Provided	`cmd/analyzer/main.go`
Test suite (40 tests)	Provided	`*_test.go`
Tool dispatch (`executeTool`)	You Implement	`internal/agent/agent.go`
Tool-call parsing (`ParseToolCall`)	You Implement	`internal/llm/client.go`
Agent loop verification	You Implement	`internal/agent/agent.go`
System prompt design	You Implement	`internal/agent/agent.go`

Project Structure

stock-analyzer/
├── Makefile
├── go.mod
├── cmd/analyzer/main.go
├── internal/
│   ├── agent/agent.go, agent_test.go
│   ├── llm/client.go, client_test.go
│   ├── models/types.go
│   └── tools/tools.go, tools_test.go
├── configs/config.example.yaml
├── testdata/sample_response.json
└── docs/challenge.tex, challenge.pdf

Getting Started

Download and extract the starter code:

tar xzf agentic-trading-challenge.tar.gz
cd stock-analyzer

Verify it compiles:
```
make build
```
Run the tests to see your starting point (7 pass, 33 fail):
```
make test
```
Set your API key:
```
export DEEPSEEK_API_KEY="your-key-here"
```
Get a key at platform.deepseek.com.
Implement the 7 mock tool functions in tools.go, executeTool in agent.go, and ParseToolCall in client.go.
Run:
```
./bin/analyzer AAPL
```

Available Tools

The mock tools return deterministic data so you can focus on the agent logic.

GetStockPrice(symbol)

{
  "name": "GetStockPrice",
  "arguments": { "symbol": "AAPL" }
}

Returns: symbol, price, change, change_pct, volume, high, low.

GetNews(symbol)

{
  "name": "GetNews",
  "arguments": { "symbol": "AAPL" }
}

Returns: array of news items with title, source, summary.

AnalyzeSentiment(text)

{
  "name": "AnalyzeSentiment",
  "arguments": { "text": "Apple Reports Record Q4 Revenue..." }
}

Returns: sentiment (positive/negative/neutral), confidence.

GetHistoricalPrices(symbol)

{
  "name": "GetHistoricalPrices",
  "arguments": { "symbol": "AAPL" }
}

Returns: array of 5 entries with date, open, high, low, close, volume (oldest first).

GetFinancials(symbol)

{
  "name": "GetFinancials",
  "arguments": { "symbol": "AAPL" }
}

Returns: symbol, revenue, net_income, eps, pe_ratio, market_cap, dividend_yield.

GetInsiderTrading(symbol)

{
  "name": "GetInsiderTrading",
  "arguments": { "symbol": "AAPL" }
}

Returns: array of insider trades with name, title, trade_type, shares, price, date.

GetAnalystRatings(symbol)

{
  "name": "GetAnalystRatings",
  "arguments": { "symbol": "AAPL" }
}

Returns: array of analyst ratings with firm, rating, price_target, date.

Mock Data Reference

Stock Prices

Symbol	Price	Change	Change%	Volume
AAPL	178.72	+2.35	+1.33%	52.3M
GOOGL	141.80	−0.95	−0.67%	23.1M
TSLA	248.50	+8.20	+3.41%	98.8M
NVDA	875.30	−12.40	−1.40%	45.2M
META	505.75	+15.30	+3.12%	32.8M
AMZN	178.25	+0.75	+0.42%	28.5M
Other	100.00	+0.50	+0.50%	10.0M

News Headlines

AAPL (3 articles)

Apple Reports Record Q4 Revenue (Reuters)
Apple Announces New AI Features (Bloomberg)
Analysts Raise Apple Price Target (CNBC)

GOOGL (2 articles)

Google Cloud Revenue Surges 28% (Reuters)
DOJ Antitrust Case Update (WSJ)

TSLA (2 articles)

Tesla Deliveries Beat Expectations (Bloomberg)
Tesla Expands Charging Network (Electrek)

NVDA (3 articles)

NVIDIA AI Chip Demand Plateaus (Reuters)
NVIDIA CEO Sells $50M in Stock (Bloomberg)
NVIDIA Valuation Concerns Rise (WSJ)

META (2 articles)

Meta Reality Labs Shows Turnaround (Bloomberg)
Meta Ad Revenue Grows 25% (Reuters)

AMZN (2 articles)

AWS Growth Slows to 12% (CNBC)
Amazon Expands Same-Day Delivery (Reuters)

Sentiment Analysis

Keyword-based: positive words (record, beat, surge, growth, strong, ...) vs negative words (decline, drop, fall, antitrust, lawsuit, ...). Tie → neutral.

Historical Prices

Returns 5 most recent trading days (oldest first). Last entry's close matches current price. AAPL shows upward trend; NVDA shows downward trend.

Financials

Symbol	Revenue (B)	EPS	P/E	Market Cap (B)
AAPL	383.28	6.42	27.8	2,800
GOOGL	307.39	5.80	24.4	1,780
TSLA	96.77	4.73	52.5	790
NVDA	60.92	12.06	72.6	2,150
META	134.90	15.02	33.7	1,300
AMZN	574.78	2.90	61.5	1,870

Insider Trading

AAPL: 2 Buy trades (bullish). NVDA: 2 Sell trades (bearish). Unknown symbols: empty.

Analyst Ratings

AAPL: 3 ratings (2 Buy, 1 Hold). NVDA: 3 ratings (1 Sell, 1 Hold, 1 Buy). META: 2 Buy. Unknown: empty.

Expected Output

{
  "symbol": "AAPL",
  "recommendation": "Buy",
  "confidence": 0.85,
  "reasoning": "Strong Q4 earnings, positive analyst sentiment, and upward price momentum.",
  "price": { "symbol": "AAPL", "price": 178.72, ... },
  "news": [ ... ],
  "sentiment": { "sentiment": "positive", "confidence": 0.85 }
}

Acceptance Criteria

Automated (must all pass)

make build — compiles without errors
make test — all 40 tests pass
./bin/analyzer AAPL — produces valid JSON to stdout

Manual Review

Agent makes ≥ 2 tool calls before producing a final answer (not hardcoded)
recommendation is one of Buy, Sell, or Hold
confidence is in [0.0, 1.0]
reasoning is a non-empty string

Scoring Rubric

Criterion	Weight	Description
Agent Loop	30%	LLM drives tool selection, multi-turn conversation, clean termination
Function Calling	25%	Robust parsing (raw JSON, code fences, surrounding text), correct dispatch
Code Quality	20%	Idiomatic Go, proper error handling, clear naming
Output	15%	Valid JSON, reasonable recommendation with supporting data
Edge Cases	10%	Unknown symbols, malformed responses, API errors

DeepSeek API Reference

Full docs: api-docs.deepseek.com · Chat completions: create-chat-completion

POST https://api.deepseek.com/chat/completions
Content-Type: application/json
Authorization: Bearer <DEEPSEEK_API_KEY>

Model ID	Description
`deepseek-chat`	DeepSeek-V3 — use this for the challenge
`deepseek-reasoner`	DeepSeek-V3 with chain-of-thought reasoning

Bonus (Optional)

Streaming — SSE streaming with real-time partial output
Retries with backoff — exponential backoff for 429/500/503
Multi-turn memory — remember previous analyses across runs
Concurrent tool calls — use goroutines to execute independent tools in parallel
Additional tests — extend the suite with your own edge-case tests