// THE PROBLEM

You shouldn't need a CS degree
to test a trading idea.

You've got a strategy: "Buy tech stocks when they dip 5%"

You tried:

  • ×Python tutorials (assume you're a CS grad)
  • ×QuantConnect (cool, but you need to code)
  • ×TradingView (great charts, limited execution)
  • ×Hiring a quant ($150/hr, questionable results)

What if you could just... describe it?

5 backtests free • No credit card required

livermore-ai
$ livermore chat
You:
Buy QQQ when it drops 3% in a day, sell when it recovers 5%
Livermore:
Got it. I'll backtest a mean reversion strategy on QQQ with:
• Entry: -3% daily move
• Exit: +5% recovery
• Period: 2020-2024

Running backtest...
✓ Backtest complete
Return:+24.3%
Sharpe:1.42
Max Drawdown:-8.1%
Win Rate:64%
Livermore:
Strategy looks solid. Deploy to paper trading?
// HOW IT WORKS

Three Steps. No Code.

From idea to deployed strategy in minutes

1

You describe your strategy

You:
"Buy SPY when it drops 3% in a day, sell when it recovers 5%"
2

AI interprets and backtests

STRATEGY IDENTIFIED
Type: Mean Reversion
Entry: -3% daily move
Exit: +5% recovery
Period: 2020-2024
BACKTEST RESULTS
Return: +24.3%
Sharpe: 1.42
Max Drawdown: -8.1%
Win Rate: 64%

Powered by institutional-grade backtesting technology - the same techniques used by professional quant funds

3

Deploy to paper or live trading

Connected to your Alpaca account - trades execute automatically

// TRANSPARENCY

What Actually Happens Under the Hood

No black boxes. Here's exactly how it works.

1. OpenAI Models

Parses your natural language strategy

"Buy when RSI < 30"
↓
{
  entry: "RSI < 30",
  exit: "RSI > 70",
  indicators: ["RSI"]
}

2. Backtesting Engine

Fine-tuned collection of institutional-grade trading models

25+ ML models and strategies
Same techniques used by quant funds
Historical data: 2015-2024

3. Alpaca Trading

Executes trades automatically

Paper trading (risk-free testing)
Live trading (Pro plan)
Your account, your control

Why this matters: We use the same quantitative techniques professional trading firms rely on. You get institutional-grade capabilities, without needing a PhD in statistics.

// HONESTY

Real Talk: What This Is & Isn't

No hype. Here's what you should know.

What Livermore AI Is Good For

  • →Testing trading ideas quickly without coding
  • →Paper trading strategies before risking real money
  • →Learning quantitative concepts hands-on (Sharpe ratio, IC, drawdown)
  • →Running 25+ ML models including ensembles
  • →Deploying to Alpaca paper/live trading

×What This Won't Do

  • ×Guarantee profits (nobody can, markets are uncertain)
  • ×Replace learning about markets and risk management
  • ×Work with every broker (Alpaca only for now)
  • ×Support high-frequency trading (daily strategies only)
  • ×Make trading risk-free (all trading involves risk)

Think of it as: Grammarly for trading strategies.
It helps you build better, faster. But you're still the trader.

// USERS

Who Actually Uses This?

Real people, real use cases

S

Sarah, Retail Trader

Uses: Free plan, paper trading only

"I trade on Robinhood. I had ideas but couldn't code them. Now I test 5 ideas before market open every month. Game changer for validating hunches."
M

Mike, Software Engineer

Uses: Pro plan, 3-model ensembles

"I can code, but setting up quant infrastructure is tedious. This lets me focus on strategy, not infrastructure. I run LightGBM + XGBoost + CatBoost ensembles on live trades."
A

Alex, Finance Student

Uses: Free plan, learning quantitative finance

"My professor mentioned Sharpe ratios and Information Coefficient. I learned by actually seeing them on my own strategies. Way better than textbook examples."
// LEARN WHILE YOU BUILD

Trading Concepts, Explained

We teach you the metrics that matter

Sharpe Ratio

Return ÷ Volatility
Good:> 1.0
Great:> 2.0
Your strategy:1.42 ✓

Why it matters: Would you rather make 20% with smooth returns, or 20% with wild swings? Sharpe tells you which is which.

Max Drawdown

-8.1%
Conservative:< 10%
Moderate:10-20%
Aggressive:> 20%

Why it matters: This is the worst peak-to-trough decline. If you can't stomach a 20% drop, don't run a strategy with 22% max drawdown.

What's an Ensemble?

Instead of 1 model:
LightGBM
Use 3 models:
LightGBM + XGBoost + CatBoost

Why? Same reason you don't trust one poll. Combining models reduces overfitting and improves performance by 20-40%.

Available on: Pro ($29/mo) - up to 3 models

Information Coefficient (IC)

0.08
Random:~0.00
Good:> 0.05
Excellent:> 0.10

Why it matters: Measures how well your predictions correlate with actual returns. Above 0.05 means your model has genuine predictive power.

All these metrics are calculated automatically when you backtest. We explain them in plain English so you can make informed decisions.

Ready to test your first strategy?

Start with 5 free backtests. No credit card required.