Bare minimum subscription to backtest intraday strategy with lean cli and use optimization?

Hi Sergio—welcome! The short answer: you can already do both minute-data backtests from the CLI and run cloud optimizations on your current Quant Researcher plan. You don’t need a more expensive tier; you just pay a little per optimization run with QuantConnect Credits (QCC).

What your current plan supports

• Lean CLI cloud backtests: fully supported on paid tiers like Quant Researcher. (quantconnect.com)
• Cloud optimization from the CLI: supported via lean cloud optimize; you rent “optimization nodes” per hour using QCC (pay-as-you-go). Example node pricing shown in the docs (O2-8 ~$0.15/hr, O4-12 ~$0.30/hr, O8-16 ~$0.60/hr). (quantconnect.com)
• Optimization tools are included at the Quant Researcher tier (you also get up to two backtesting nodes for concurrent backtests). (quantconnect.com)

How to run everything from the CLI (minute data included)

• Cloud backtest with parameters:
  • lean cloud backtest "My Project" --push --open --parameter ema_fast 12 --parameter ema_slow 26
  • Cloud backtests use QC’s Dataset Market data, so you don’t have to buy/download local data for these runs. (quantconnect.com)
• Cloud optimization (interactive wizard):
  • lean cloud optimize "My Project" --push
  • The wizard will ask for objective, parameter ranges, constraints, node type, and number of nodes; it then estimates time and cost and charges QCC only for what you actually use. (quantconnect.com)
• Tip: expose tunable parameters in your algorithm and read them via algorithm parameters so you can vary them with --parameter from the CLI. If you’d like, I can generate a tiny template that does this.

Keeping costs low while you learn

• Do intraday work in QuantConnect Cloud. Most price data is free to use in the cloud, so your backtests don’t incur data fees—only compute for optimization when you choose to run it. (quantconnect.com)
• Avoid downloading minute data locally unless you really need it. Local downloads are priced per symbol-day (for US Equities minute data, 5 QCC ≈ $0.05 per symbol-day, and Security Master is a separate annual license). Those costs add up quickly compared to running in the cloud. (quantconnect.com)
• Start with short date ranges and small universes when optimizing; scale up only for promising settings.
• Cloud optimization caps at three parameters per job; if you need more, you can run local optimizations (but that requires local data). (quantconnect.com)

Suggested “affordable” setup

• Stay on Quant Researcher (your current plan).
• Top up a small QCC balance (e.g., $10–$20) and run cloud optimizations as needed. The docs show small optimization batches often cost well under a dollar. (quantconnect.com)
• Keep all minute-data backtests and optimizations in the cloud to avoid any local data purchases.

If you want, I can:

• Generate a minimal algorithm template that reads parameters via GetParameter so you can sweep settings from the CLI.
• Provide a ready-to-run CLI script to batch multiple lean cloud backtest runs and summarize the results.

Tell me your target asset class and a couple of parameters you want to tune, and I’ll set that up for you.

Sources: Costs - QuantConnect.com, QuantConnect Paper Trading - QuantConnect.com

Ahhh, good bot, thanks!!!

I think my issue was that I was running “lean backtest” instead of “lean cloud backtest”. After running “lean cloud backtest” I was able to execute the backtest from the command line. thanks! 

It seems that when you run “lean backtest” it tries to download the data and that was causing issues such as:

DATA USAGE:: Total data requests 2892
DATA USAGE:: Succeeded data requests 2
DATA USAGE:: Failed data requests 2890
DATA USAGE:: Failed data requests percentage 100%