MVP to production: scaling a trading platform.
Operational changes required as a trading-platform MVP transitions to production.
The MVP
A seed-stage trading-platform MVP typically demonstrates clean UI, working trade flow, paper trading, and broker integration. It generally lacks replayable event streams, observability, reconciliation, and incident procedures.
Failure modes
When real users operate the platform with real capital, several failure categories surface.
Data divergence
Positions diverge from the broker. P&L diverges from the bank statement. Trade history disagrees with exchange reports. Reconciliation pipelines from production launch are required.
Order types
Production exposes stop orders, OCO orders, GTC orders, after-hours orders, and orders for delisted instruments. An explicit order state machine is required.
Corporate actions
Splits, spin-offs, symbol changes, and listings are first-class subsystem requirements rather than ad hoc handlers.
Tenant isolation
Bad client integrations consume rate limits. Bot retries amplify failures. Per-tenant quotas, per-user observability, and rate limits are required from launch.
On-call
Markets operate around the clock during weekdays. An on-call rotation is required even at small team sizes.
Foundational planning
Production foundations are most cost-effective to build before scale. Retrofit after scale is materially more expensive.