How we test what we build — and why the result is honest.
Tick-level XAU/USD data from 2008 to today. 18 years of market data including the 2008 financial crisis, Corona crash 2020, 2022 inflation, 2023 banking crisis, 2024 rate shocks.
For each user profile we compute up to 100,000 parameter combinations. Every variant is simulated on historical data with modeled spreads and slippage.
Strategies are optimized on a training period, then tested on a later out-of-sample period. Whatever fails the test is discarded.
Trade order is randomly permuted, spreads artificially widened. Strategies that only work under ideal conditions are filtered out.
Out of 100,000 variants, typically less than 1 % survive all tests. You only receive variants that passed all 4 stages.