PineForgeOffline PineScript Backtesting on Your Own PC

What Is PineForge and Who Is It For?

PineForge takes a strategy written in PineScript v6 and runs it against historical market data on your own computer. Instead of replaying charts by hand, you get a complete list of simulated trades, with entries, exits and results, that you can analyze or compare.

The people who benefit most are Pine script authors who want to test ideas over longer periods than a chart replay allows, and developers who need a dependable execution core for their own tools. Quant hobbyists who like scripted experiments will also find it useful.

A typical scenario: you have a moving-average crossover strategy that looks promising on a few months of charts. You want to see how it behaved across several years and across different parameter settings. Running those tests locally lets you change one variable at a time without rebuilding the analysis by hand.

PineForge is not a charting platform or a live dashboard. It concentrates on execution logic, so you will typically pair it with your own data preparation and reporting.

Deterministic Results and Reproducible Research

Deterministic means the same inputs always produce the same output. With PineForge, two runs with identical scripts, data and settings generate identical trade lists. That sounds obvious, but many backtesting setups drift between runs because of timing, randomness or changing configuration, which makes it hard to tell whether a result changed because of your idea or because of the tool.

Reproducibility matters most when you compare variants. Suppose you test three stop-loss values. If each run is repeatable, any difference in profit or drawdown comes from the stop-loss value itself. Keep a record of the script version, the date range and the parameters for each run, so you can recreate any result later.

A few habits help. Store each strategy version in its own file, avoid editing the data between runs, and write down the settings you used. These steps cost little and protect you from fooling yourself with a lucky run.

Determinism does not make a strategy profitable. It makes your measurements trustworthy, which is a prerequisite for sound decisions.

Can AI Agents Automate Strategy Research?

PineForge includes an MCP server, a standard way for AI assistants to call external tools. Through it, an agent can start the engine, pass in a script and read back results. That opens the door to workflows where an assistant runs a series of tests, summarizes what changed and suggests the next experiment.

Automation is useful for repetitive work: testing many parameter combinations, checking how a script behaves on different date ranges, or organizing output tables. An agent handles these tasks faster than a person clicking through options, and it does not tire of the routine.

Automation has limits. An assistant can misread results or propose changes that look good only because of noise in the sample. Treat each suggestion as a hypothesis, check the trade list yourself and rerun the test on fresh periods before you trust it.

If you are new to agents, start with a small task such as a single parameter sweep, and expand only after you are comfortable with how the tool reports results.

Supported coins

The project documentation mentions Bitcoin (BTC), Ethereum (ETH) and USDT as example assets. Because PineForge works on historical data you supply, you can test strategies on these markets as long as your data files cover them.