Key backtest highlights
Period: 2010‑01‑01 to 2024‑01‑01.
Initial deposit: $5,000; Leverage 1:500.
Total trades: 10,995; Win rate ~71.6%; Profit factor 2.06.
Quality: 100% tick quality, 990,310 bars, 288,793,961 ticks.
Performance notes: nominal total net profit is extremely large in the report, but the backtest shows maximum equity drawdown 26.6% and maximal balance drawdown 5.80%, with a Sharpe ratio 3.44. These figures point to a historically profitable edge accompanied by meaningful equity swings.
Strategy overview
Core mechanics: a grid‑style position management system with custom multipliers (1,2,4,8,14,24,41,69) and a 25‑pip trade distance, designed to average into moves and capture mean reversion or retracements.
Signal filtering: a neural‑network (ML) filter is enabled to reduce false entries and improve probability of profit (ProbabilityOfProfit set to 90).
Risk controls in code: dynamic lot sizing tied to deposit load percent, a cap on concurrent trades (MaximumTrades = 9), news and market filters, and optional trading hours. Grid SL is set to 0 (no fixed stop per grid level), so the system relies on position management and equity controls rather than per‑trade stop losses.
Behavioral implication: the EA compounds winners and uses averaging; it expects occasional clustered losses but aims to recover via the grid and filtered entries.
What to expect in live trading
Realistic returns: expect lower net returns than backtest due to live spreads, slippage, partial fills, and broker execution differences. The profit factor and win rate indicate an edge, but execution costs will erode headline numbers.
Drawdown profile: plan for drawdowns in the ~25–30% range during adverse trends; the backtest’s 26.6% equity drawdown is a practical planning figure.
Capital needs: grid strategies require sufficient free margin to withstand adverse moves; the backtest used 1:500 leverage and $5,000 start — real‑money sizing should be more conservative.
Operational needs: low‑latency VPS, stable MT5 execution, and continuous monitoring are essential to reproduce favorable conditions.
Broker and technical setup recommendations
Broker type: choose ECN/STP brokers with raw spreads and reliable MT5 bridges; low latency and deep liquidity reduce slippage and widen the chance of matching backtest conditions. IC Markets is a perfect example.
Account features to prefer: raw/ECN pricing, tight execution SLAs, minimal requotes, and transparent swap/rollover policies.
VPS and connectivity: host the EA on a VPS located near the broker’s liquidity servers; ensure MT5 is always connected and backups are in place. TradingFX VPS is a perfect example.
Practical tip: test the EA on the broker’s demo with real spreads and slippage settings before moving to live; compare fills and execution statistics to the backtest.
How to benefit when trading real money
Start small and scale: begin with micro‑lots or a small live allocation (e.g., 0.25%–1% deposit load as in the backtest) and scale only after consistent live performance.
Use layered risk: combine per‑trade conservative sizing with an overall equity stop and a daily loss limit to prevent catastrophic sequences.
Keep filters active: maintain the news and market filters used in the backtest to avoid trading during high‑impact events.
Record and review: log trades, slippage, and execution anomalies; weekly reviews help tune grid distance, multipliers, and ML filter sensitivity.
Hazards to expect and loss‑avoidance precautions
Grid exposure: prolonged one‑way trends can inflate drawdown; without fixed per‑level SLs, the account relies on equity stops and manual intervention.
Execution risk: widened spreads, slippage, and requotes during news or low‑liquidity hours can turn manageable grids into large losses.
Over‑leveraging: high leverage amplifies both gains and drawdowns; avoid aggressive deposit loading until live behavior is validated.
Mitigations: enable news filters, set an equity stop (percentage or dollar), cap maximum open lots, and use conservative lot sizing when starting.
Practical checklist before going live
Demo forward‑test for at least 3 months using the target broker’s real spreads and slippage.
Run a small live pilot with micro‑lots to observe real execution and drawdown behavior.
Set hard risk limits: daily loss cap, equity stop, and maximum concurrent trades.
Use a reliable VPS and monitor connection/latency metrics.
Review and adapt: analyze weekly logs and adjust grid multipliers, distance, or ML filter thresholds if live results diverge materially.
Final assessment
The backtest published by Valery Trading demonstrates a clear historical edge for the EA, but the grid‑based approach and absence of per‑level stop losses mean real‑money deployment must prioritize capital adequacy, conservative sizing, and strict risk controls. With careful broker selection, VPS setup, and phased scaling, running Perceptrader AI live can be feasible — but expect lower headline returns and prepare for significant drawdowns during adverse market regimes.
