Most forex indicators fail.
Stop guessing and see the historical data of what actually survived across 635,000+ setups.

What is in the database

Two datasets, one query-credit balance

Every result row is a bar-by-bar backtest on historical OHLC price data. The full grid is included, so the losing setups are visible alongside the top of the list.

Strategies include RSI, MACD, Bollinger Bands, Ichimoku, Stochastic, Parabolic SAR, Williams %R, CCI, Awesome Oscillator, Bull Bear Power, Momentum, Ultimate Oscillator, Keltner Channels, and Advance/Decline Ratio, alongside SMA and EMA crossover grids. Pairs cover EUR/USD, GBP/USD, USD/JPY, AUD/USD, USD/CHF, USD/CAD, NZD/USD, EUR/GBP, GBP/JPY, and EUR/JPY. Every result was simulated from 1 January 2019 through 31 December 2025.

Paid access also includes a quarterly dataset. Each completed calendar quarter is backtested as its own batch on the same strategies, pairs, and timeframes, and tagged with a quarter ID that can be filtered on in the QTR Query page. Quarters are appended after they finish. Both datasets share the same query-credit balance.

How it works

Three steps from question to ranked answer

1. 1

   Pick a strategy

   Indicators, MA Crossover, or EMA Crossover. Each strategy has its own results table with the right columns for that family.

2. 2

   Filter

   Narrow by pair, timeframe, indicator (for indicator strategies), the minimum-trades threshold, and the Monte Carlo filter that keeps only setups clearing their simulated survival thresholds.

3. 3

   Sort and choose a row size

   Sort by total gain, win rate, Sharpe, Sortino, or reward-to-risk. View 100, 200, or 500 rows per query.

Methodology

How the data was built

Every row in the database is produced under the same conservative defaults.

• One-bar shift on every indicator.Each signal on bar i is computed using only bar i−1 and earlier, so the simulation does not see information that would not have been visible at decision time.
• Stop-loss assumed first on M1 intrabar ties. When the stop and target both fall inside the same minute bar, the engine assumes the stop printed first. That is the more conservative outcome, by design.
• Recorded fill order on M5 and slower. Higher timeframes carry a per-bar high_first flag, so the engine knows whether the bar's high or low printed first instead of guessing.
• Net of commission. Every gain figure subtracts a per-trade commission in pips before it is stored, so the figure is not a frictionless one.
• Entries on the next bar's open. Signals fire at bar close and execute at the open of the following bar, a one-bar delay that mirrors how a real trader would respond.
• The full parameter grid is included.Every stop-loss × take-profit × pair × timeframe combination is in the dataset, including the losers, so the underlying distribution is visible alongside the top of the list.

Pricing

Annual access, queries metered by result size

Each paid term includes 1,000 query credits that reset every 30 days. Credits are spent by result size: 100 rows costs 1 credit, 200 rows costs 2, and 500 rows costs 5. If you use the full allowance at 100 rows per query, that is up to 100,000 ranked result rows per 30-day cycle for screening and comparison. Larger row sizes return more data per query but use credits faster.

Auto-renewal can be cancelled at any time. Cancelling stops the next renewal; access continues through the end of the year already paid for. Because the value of the database is transferred once users see the data, we have a strict no-refund policy. Please read our No-Refund Policy for more information.