CapEx & OpEx — System Financials
Variable and fixed costs of running the Shanfeld Equities AI intelligence engine
Monthly OpEx
$256.50
All-in operating cost
Annual Run Rate
$3,078.00
Projected yearly
API Costs
$187.00
73% of total
Infrastructure
$69.50
27% of total
Cost / AUM
0.0 bps
Annual cost basis
API Costs by Agent — Variable OpEx
Infrastructure — Fixed OpEx
Dashboard hosting + CI/CD
Hosting
Free tier — tunnel to Mac Studio
Networking
Decision logging + analytics (estimated)
Data
Fidelity portfolio sync
Data
shanfeldequities.com
Infrastructure
24/7 operation — M2 Ultra estimated
Hardware
NAS storage — 2x 4TB Seagate Barracuda
Hardware
Cost Optimization Opportunities
Route simple queries through Claude Sonnet 4.5 ($3/M) instead of Opus ($15/M). Saves ~$25/mo for non-strategic queries.
Gemini 2.0 Flash at $0.10/M input is 150x cheaper than Opus. Use for high-volume quantitative scans.
Implement semantic caching for similar questions within a session. Reduces API calls by ~30%.
Reduce system prompt sizes across agents. Each 1K token saved across 300 daily calls = significant savings.
Use Anthropic Batch API (50% discount) for non-real-time analysis: end-of-day reviews, portfolio rebalancing.
Run a small local model (Llama 3) for query classification before routing to paid APIs. Reduces unnecessary API calls.
ROI Analysis
Breakeven Alpha
0 bps
Annual alpha needed to cover system costs
Cost per Decision
$0.12
Avg cost per consensus analysis (~50/day)
Cost per $1K Managed
$3078000.00
Annual cost per $1K of portfolio
vs Hedge Fund Fees
99% cheaper
2/20 hedge fund = ~$0/mo vs $256.50
Note: API costs are estimates based on projected usage. Actual costs vary based on query complexity, response length, and daily volume. All costs shown assume the current paper trading configuration. Production trading may increase API usage 2-3x.