EMA Cross Demo EA with ATR Stops - sistema esperto per MetaTrader 5

EMA Cross Demo EA (MetaTrader 5)

A small, clearly-structured demo Expert Advisor built to show a reliable EA development method, not a trading edge.

Logic (evaluated once per new bar, on CLOSED bars only - no repainting):

- BUY signal: fast EMA crosses above slow EMA

- SELL signal: fast EMA crosses below slow EMA

- On an opposite signal the open position is closed and the new one opened on the next tick of the new bar

Risk handling:

- Stop-loss = entry minus/plus ATR x InpSLATRMult

- Take-profit = stop distance x InpRewardRisk

- Stop/TP levels are pushed out to the broker's minimum stop distance where needed

- Lot size is normalized to the symbol's volume step, min and max

- Entries are skipped when spread is wider than InpMaxSpreadPoints

Works on both hedging and netting accounts (position lookup is done by magic number and symbol, not by netted position type).

Inputs: fast/slow EMA period, ATR period, SL ATR multiplier, reward:risk, lot size, max spread in points, magic number, slippage in points.

This is a deliverable example meant to show structure and method: closed-bar signals, a spread filter, lot normalization, and hedging/netting-safe position handling. It is not a trading recommendation. The attached equity chart is from a synthetic price series used only to exercise the code (gross P/L is positive, net P/L after commission is negative on this synthetic run) - it demonstrates the backtesting approach, not a profitable strategy.

A Python reference implementation of the same signal and backtest logic (for parity testing) is available on request.

Update: fixed a stop-distance issue where the ATR-based SL/TP could sit inside the broker's minimum stop distance on fast timeframes, causing order rejections. The distance is now floored at the stops/freeze level and checked against the correct closing-side price (Bid for a BUY, Ask for a SELL) instead of the entry price.

Update 2: the EA now checks available margin with OrderCalcMargin() before sending an order, and skips (with a log line) instead of sending an order the account cannot afford. This avoids broker-level "not enough money" rejections on very small test balances.