Product / UX
Product scope, workflows, responsive UI, dark/light experience and multilingual foundations.
CASE STUDY / 01 · SOLO-BUILT PRODUCT
AurumTrade
AT/01PRODUCT SYSTEM
Solo-built end-to-end by Denys Novikov: a full-stack, AI-assisted trading performance product that unifies trade capture, exchange data, behavioral journaling, playbook discipline, analytics and subscription infrastructure.
Founder · Product EngineerFull-Stack · AI · DataSolo-built end-to-end
21product views
5AI workflows
17CSV variants
27RLS policies
01 / CAPTUREManual + CSVTrades, screenshots, context
02 / INTERPRETAnalyticsP&L, R:R, drawdown, behavior
03 / REFLECTJournal + PlaybookEmotion, rules, discipline
04 / ASSISTPersonalized AIReview, coach, psychology, chat
01 / PRODUCT PROBLEM
Trading data is fragmented.
A useful trading journal cannot stop at a table of wins and losses. The product connects structured trades with exchange exports, screenshots, execution quality, emotions, journal notes, playbook rules and retrospective analysis — then makes that context available to analytics and AI-assisted review.
02 / SOLO OWNERSHIP
One product.
End-to-end ownership.
Full-Stack
21 application views, 15 API route modules, authentication, CRUD, imports, uploads and account lifecycle.
Data / PostgreSQL
Supabase schema, RLS ownership, indexes, triggers, constraints and database-level free-tier enforcement.
AI Systems
Five authenticated OpenAI workflows grounded in trades, playbooks, journal entries and emotion data.
Analytics
Performance metrics, equity curve, behavioral feedback and a 200-path Monte Carlo-style projection.
SaaS / Billing
Stripe Checkout, Customer Portal, signed webhooks and server-side AI entitlement foundations.
03 / SYSTEM ARCHITECTURE
Data in. Context out.
INPUT / 01Manual TradeTrade context + execution notes
INPUT / 02Exchange CSVDetection + mapping + normalization
→PRODUCT CORENext.jsUI + authenticated APIs
→DATASupabaseAuth · PostgreSQL · RLS · Storage
→OUTPUTAnalytics + AIPerformance · journal · coaching
Stripe Checkout→Signed Webhook→Plan State→AI Entitlement
04 / DATA INGESTION
17 export variants.
One ingestion path.
01Detect
→02Preview
→03Map
→04Normalize
→05Insert
Selected exchange export formats plus manual mapping. This is not presented as universal or lossless exchange reconciliation.
05 / AI CONTEXT ARCHITECTURE
Personal history,
not generic commentary.
TRADESJOURNALPLAYBOOKEMOTIONS
CONTEXT ASSEMBLYOpenAI / GPT-4oAuthenticated · server-side · plan-gated
01Trade Review
02Trade Score
03Coach
04Psychology
05Chat
AurumTrade does not claim to predict market outcomes. The AI layer is personalized review and coaching grounded in the user's own records.
06 / DATA + ACCESS
Access rules live
below the interface.
8application tables
27RLS policies
18indexes
20free-plan trade cap
07 / ANALYTICS
Performance + behavior,
one feedback loop.
Win rateTotal / avg P&LR:RDrawdownLong / shortBest setupStreaksEquity curvePair breakdownWeekday / monthGrade / emotionRecent trades
200 paths
Interactive Monte Carlo-style performance projection using historical win rate and average win/loss values. A simplified projection, not a market simulator or predictive model.
08 / PRODUCT MATURITY
Complete beta product.
Known hardening work.
The forensic audit confirmed broad end-to-end implementation while identifying the next engineering phase: schema reconciliation, stronger trade-domain validation, AI contracts, billing lifecycle hardening, privacy alignment, automated tests, CI/CD and operational observability.
09 / OUTCOME
From scattered trading artifacts to one connected product system.
AurumTrade demonstrates solo end-to-end product engineering across UX, full-stack architecture, data security, analytics, AI-assisted workflows and SaaS foundations — without reducing the work to a dashboard.