Engineering Transparency

SimTrade is Not a GPT Wrapper

We built proprietary algorithms, integrated real market data, and created genuine trading infrastructure. Here's the proof.

Try the Platform View API Docs

What is a "GPT Wrapper"?

A GPT wrapper is an app that simply sends your input to an AI API, then displays the response. No proprietary logic, no real data, no persistent state. Just a middleman.

GPT Wrapper Pattern

// This is ALL a GPT wrapper does
async function analyze(userInput) {
  const response = await openai.chat({
    messages: [
      { role: "user", content: userInput }
    ]
  })
  return response.text // That's it
}

SimTrade Approach

// SimTrade: Real data + algorithms + AI
async function analyze(symbol) {
  const price = await alphaVantage.get(symbol)
  const rsi = calculateRSI(price.history)
  const macd = calculateMACD(price.history)
  const agents = await runMultiAgentDebate(data)
  await db.save({ symbol, rsi, macd, ... })
  return { technicals, consensus, risk }
}

Proprietary Technology

Built From Scratch, Not Prompted

Every core feature runs on proprietary code. AI enhances the experience, it doesn't replace the product.

10+ Indicators

Proprietary Technical Indicators

RSI, MACD, Bollinger Bands, Stochastic, ATR, OBV, VWAP, MFI - all calculated in-house with pure mathematics, not AI

// Pure math, no AI
function calculateRSI(prices: number[], period = 14) {
  const gains = [], losses = []
  for (let i = 1; i < prices.length; i++) {
    const change = prices[i] - prices[i-1]
    gains.push(change > 0 ? change : 0)
    losses.push(change < 0 ? -change : 0)
  }
  const avgGain = gains.slice(-period).reduce((a,b) => a+b) / period
  const avgLoss = losses.slice(-period).reduce((a,b) => a+b) / period
  return 100 - (100 / (1 + avgGain / avgLoss))
}

6 Strategies

Full Backtesting Engine

Test strategies against historical data with Sharpe ratio, max drawdown, win rate, and Monte Carlo simulation

// Real backtesting logic
function runBacktest(strategy, historicalData) {
  const trades = [], equity = [startingCapital]
  for (const candle of historicalData) {
    const signal = strategy.evaluate(candle, indicators)
    if (signal.action === 'BUY') {
      trades.push({ entry: candle.close, ... })
    }
  }
  return {
    sharpeRatio: calculateSharpe(returns),
    maxDrawdown: calculateDrawdown(equity),
    winRate: wins / trades.length
  }
}

5 Agents

Multi-Agent Debate Architecture

5 specialist AI agents (Fundamental, Technical, Sentiment, News, Volume) debate and reach consensus - not a single prompt

// Multi-agent consensus, not single prompt
const agents = [
  FundamentalAgent.analyze(data),   // P/E, Revenue
  TechnicalAgent.analyze(data),     // RSI, MACD
  SentimentAgent.analyze(data),     // Social signals
  NewsAgent.analyze(data),          // Breaking news
  VolumeAgent.analyze(data),        // Flow analysis
]
const consensus = await debateProtocol(agents)
return consensus.recommendation

5+ Data Sources

Real Market Data Sources

Alpha Vantage, CoinGecko, Yahoo Finance, Polymarket, Kalshi - actual market data, not generated text

// Real API integrations
const stockData = await alphaVantage.getQuote(symbol)
const cryptoData = await coinGecko.getPrice(coinId)
const predictionData = await polymarket.getMarket(slug)

// Data stored in PostgreSQL, not context
await supabase.from('trades').insert({
  symbol, price: stockData.price,
  user_id, executed_at: new Date()
})

Feature Comparison

SimTrade vs. GPT Wrappers

A side-by-side look at what we actually built versus what a typical AI wrapper provides.

Feature

SimTrade

GPT Wrapper

Real market data from APIs

Proprietary indicator calculations

Persistent user portfolios in database

Full backtesting with statistics

Multi-agent debate architecture

Paper trading simulation engine

Payment & subscription infrastructure

Social features (leaderboards, copy trading)

Just sends prompt to GPT API

Displays AI response as product

System Architecture

Real Infrastructure, Real Data Flow

SimTrade integrates multiple data sources, runs proprietary calculations, and persists everything in a real database.

Data Sources

Alpha Vantage

CoinGecko

Yahoo Finance

Polymarket

Kalshi

Processing Layer

Technical Indicators

Backtesting Engine

Risk Calculator

Multi-Agent System

Portfolio Manager

Persistent Storage

User Portfolios

Trade History

Achievements

Subscriptions

Referrals

AI is used forNatural language insightsAgent debatesTrade explanations- not as the core product

Technology Stack

Built on Proven Infrastructure

Next.js 15Framework

SupabaseDatabase

PostgreSQLStorage

Alpha VantageMarket Data

CoinGeckoCrypto Data

StripePayments

PolymarketPredictions

KalshiEvent Markets

See It In Action

The best way to understand SimTrade is to use it. Start paper trading with real market data today.

Create Free Account Explore Features

// This is ALL a GPT wrapper does async function analyze(userInput) { const response = await openai.chat({ messages: [ { role: "user", content: userInput } ] }) return response.text // That's it }

// SimTrade: Real data + algorithms + AI async function analyze(symbol) { const price = await alphaVantage.get(symbol) const rsi = calculateRSI(price.history) const macd = calculateMACD(price.history) const agents = await runMultiAgentDebate(data) await db.save({ symbol, rsi, macd, ... }) return { technicals, consensus, risk } }

// Pure math, no AI function calculateRSI(prices: number[], period = 14) { const gains = [], losses = [] for (let i = 1; i < prices.length; i++) { const change = prices[i] - prices[i-1] gains.push(change > 0 ? change : 0) losses.push(change < 0 ? -change : 0) } const avgGain = gains.slice(-period).reduce((a,b) => a+b) / period const avgLoss = losses.slice(-period).reduce((a,b) => a+b) / period return 100 - (100 / (1 + avgGain / avgLoss)) }

// Real backtesting logic function runBacktest(strategy, historicalData) { const trades = [], equity = [startingCapital] for (const candle of historicalData) { const signal = strategy.evaluate(candle, indicators) if (signal.action === 'BUY') { trades.push({ entry: candle.close, ... }) } } return { sharpeRatio: calculateSharpe(returns), maxDrawdown: calculateDrawdown(equity), winRate: wins / trades.length } }

// Multi-agent consensus, not single prompt const agents = [ FundamentalAgent.analyze(data), // P/E, Revenue TechnicalAgent.analyze(data), // RSI, MACD SentimentAgent.analyze(data), // Social signals NewsAgent.analyze(data), // Breaking news VolumeAgent.analyze(data), // Flow analysis ] const consensus = await debateProtocol(agents) return consensus.recommendation

// Real API integrations const stockData = await alphaVantage.getQuote(symbol) const cryptoData = await coinGecko.getPrice(coinId) const predictionData = await polymarket.getMarket(slug) // Data stored in PostgreSQL, not context await supabase.from('trades').insert({ symbol, price: stockData.price, user_id, executed_at: new Date() })

Command Palette

SimTrade is Not a GPT Wrapper

What is a "GPT Wrapper"?

Built From Scratch, Not Prompted

SimTrade vs. GPT Wrappers

Real Infrastructure, Real Data Flow

Data Sources

Processing Layer

Persistent Storage

Built on Proven Infrastructure

See It In Action

SimTrade is Not a GPT Wrapper

What is a "GPT Wrapper"?

Built From Scratch, Not Prompted

SimTrade vs. GPT Wrappers

Real Infrastructure, Real Data Flow

Data Sources

Processing Layer

Persistent Storage

Built on Proven Infrastructure

See It In Action