Skip to content
Sableframe
Sableframe
PRIVATE RESEARCH BUILD

Stratifi

A safety-first workspace for systematic trading research.

A personal trading platform that brings strategy research, paper execution, risk controls, audit trails, and validation evidence into one disciplined workspace.

Status
PRIVATE RESEARCH BUILD
Source
Private
Capabilities
  • Financial tooling
  • Backtesting
  • Risk systems
Stratifi dashboard under a paper mode and mock broker banner, with tiles for a $100,000 paper equity balance, cash, P&L, drawdown, risk profile, and bot status, an equity curve, and the watch list for one paper strategy.
Paper tradingDashboardThe research workspace overview on a paper-trading account.

The problem

Systematic trading research is only useful if its evidence can be trusted. A backtest that draws a clean-looking equity curve from flawed assumptions is more dangerous than an obvious failure.

Stratifi brings strategy research, paper execution, and risk controls into one workspace built around that concern.

The system

  • Strategy library

    Strategies are pure functions with no I/O, so the same logic drives backtests and paper trading.

  • Backtesting

    Fills at the next bar’s open with modelled slippage and commission. Daily-bar backtests are treated as reliable; intraday ones are not.

  • Validation

    Cost-stress reruns and walk-forward tests decide whether a result is fit to act on.

  • Risk engine

    Every order passes checks for loss, drawdown, exposure, sizing, data freshness, and trading hours, and every result is recorded.

  • Paper execution and kill switches

    Orders go to a paper-trading broker adapter or a deterministic mock, with system lock, pause, and manual close controls.

How it was built

A React and TypeScript interface built with Vite talks to a Python FastAPI service. PostgreSQL stores strategies, backtests, orders, risk checks, and audit events, and scheduled jobs drive the strategy loop.

Orders flow from a strategy through the risk engine to an execution service and a broker adapter. The only broker integration is Alpaca paper trading, and live trading is disabled by default.

  1. Research

    • Strategy library
    • Backtest engine
    • Validation

  2. Decide

    • Strategy service
    • Risk engine

  3. Execute

    • Execution service
    • Alpaca paper adapter
    • Mock broker

  4. Record

    • PostgreSQL
    • Risk checks and audit events
A proposed order is approved or blocked by the risk engine before it reaches the paper broker, and every step is recorded.

Technology

  • React
  • TypeScript
  • Python
  • FastAPI
  • PostgreSQL
  • Alpaca paper-trading integration

Design decisions

  1. The risk engine never short-circuits

    Every check runs on every order and each result is stored, so a blocked trade always shows all of its reasons.

  2. One sizing formula

    Backtests and paper trading size positions the same way, so research and execution stay comparable.

  3. Validation is part of the engine

    Robustness checks run as part of analysis rather than as an optional habit. An edge that disappears under higher costs is not treated as an edge.

  4. Decimal money math

    All monetary calculations use decimal arithmetic rather than floating point.

Selected views

  • Stratifi backtest detail for a demo run on synthetic bars, with a run explanation, return, drawdown, Sharpe, win-rate, and capital tiles, and an equity curve, under a paper mode and mock broker banner.
    Synthetic dataBacktest detailAnalysis and validation evidence for a demo backtest on synthetic price history. The results are not real performance.
  • Stratifi Risk Controls page showing the running bot state with pause and full system lock buttons, an emergency exit to close all positions, and editable position, loss, circuit-breaker, and trade-quality limits.
    Demo dataRisk and systemRisk controls, bot state, and the emergency exit, with demo values and no account details.
  • Stratifi Risk Controls on a phone under the paper mode and mock broker banner, with the running bot state, pause and full system lock buttons, the emergency exit, and the first risk setting.
    Paper tradingOn a phoneRisk controls at a 390px viewport, with the paper mode banner in view.

Current state

Private research build under active development. It runs against paper trading only, intraday backtests are not treated as reliable, and market-regime rotation is still untested. The source code is private.