About BacktestMarket
Founded in 2015 by traders who needed data they could trust. A decade later, the mission hasn't changed — only the scale and the stakes have.
“When doing something you will have to face those who wanted to do the same, those who wanted to do the opposite, and the majority of those who did not want to do anything.”
Our History
The Beginning
Cesare and Andrea met in spring 2015 and immediately recognised a shared frustration: quality historical market data was either unreliable, poorly formatted, or priced out of reach for independent researchers. They decided to build what they could not find.
First Products
The first Expert Advisors and historical datasets were published on BacktestMarket. The catalogue was small, but each dataset was verified bar by bar — a standard of rigour that became the company's identity.
Growing the Catalogue
The data library expanded steadily: forex spot rates, metals, equity indices, bonds, commodities, and futures — each with documented timezone, rollover logic, and split/dividend adjustment methodology. A community of quantitative traders and researchers grew around it.
The Age of AI Data
The explosion of large language models and AI trading systems changed what "good data" means. Clean, precisely timestamped time-series became raw material for machine learning pipelines alongside human backtests. BacktestMarket's uncompromising quality standard — built for humans — turned out to be exactly what models need too.
Trusted by Quants & Machines
Researchers, quantitative funds, and AI developers across the world rely on BacktestMarket data. The mission remains the same as 2015: provide the most reliable, consistently maintained historical market data available anywhere.
Data in the Age of AI
For most of BacktestMarket's history, "data quality" meant one thing: can a trader trust these bars to backtest a strategy? Correct timestamps, no phantom spikes, accurate rollover logic, proper dividend adjustment. The bar was already high — and we met it.
Since early 2023, a second demand has arrived alongside the first. Large language models, reinforcement learning agents, and quantitative AI systems all need the same thing as human researchers: clean, consistently formatted, machine-readable time-series. The difference is that models consume millions of bars at once, and a single dirty row can silently poison an entire training run.
This is not a coincidence that BacktestMarket data works well for AI pipelines. The same properties that make data useful for backtesting — deterministic formatting, no lookahead, documented gaps, explicit timezone — are exactly the properties that make it safe to feed into a model. We built for rigour, and rigour transfers.
The broader shift matters too. As AI-driven strategies become more prevalent across institutional and retail trading, the quality of the underlying historical data becomes a competitive variable, not just a hygiene factor. Bad data produces bad models. Researchers who cut corners on data quality will eventually be outcompeted by those who don't.
BacktestMarket has been the data layer for serious quantitative work since 2015. That role is more important now than it has ever been.
The Team
The people who build and maintain BacktestMarket.
“Price is what you pay. Value is what you get.”
— Warren Buffett