Cross-checking someone else's H1 breakout on my own rig

Breakout entry example (XAUUSD daily, real data): buy when price breaks above the recent high.

The breakout_h1 strategy, a long-only Donchian breakout system, has been a central focus of my research. It has shown up independently in two separate research tracks. While one of them suggested several levers to boost performance, my rigorous cross-verification using the EA engine and M1 intraday stress testing tells a different story.

The verdict on “boosters”

The separate research track proposed adding a short-side sleeve and an overlay mechanism to reduce drawdowns. When I tested these on my EA infrastructure, the results were consistently negative:

• Short Sleeve: While suggested as a hedge, the short component was net-negative in both OOS (out-of-sample) and IS (in-sample) periods, with a correlation of -0.05 to the long side. In other words, it acted as a drag rather than a hedge and reduced the basket efficiency from 5.04 down to 3.80.
• Overlay: This was intended to reduce drawdown from 32% to 20%. Instead, it cut efficiency in half (from 9.40 to 5.04) and returned only 159% compared to the 284% seen when it was turned off. The drawdown remained largely unchanged. The takeaway is clear: Overlay OFF is the correct setting. Both the short sleeve and the overlay are rejected for deployment.

Hidden risks in M1 intraday testing

The most significant finding concerns the recommended risk settings. The separate research track suggested a risk of 0.3% to 0.5% per pair. However, when I rebuilt the account equity using 1-minute bars (M1 intraday), the strategy frequently hit a 5% daily loss limit. This was particularly evident during the USDJPY intervention on April 29, 2024. To keep the system safe for prop firm rules, the risk must be capped at approximately 0.15% per pair. In other words, the safe limit is 1/2 to 1/3 of the original recommendation. At the higher levels recommended by the separate track, the strategy is simply not safe for accounts with strict daily loss constraints.

Why the numbers differ

When comparing the separate track’s results (monthly +2.62%) to my EA engine results, I found a discrepancy in drawdown. My testing showed a raw drawdown of -25% to -38%. Initially, I suspected this was due to gold (XAUUSD) or recent market regimes, but the drawdown persisted even after removing gold or adjusting for the 2025 to 2026 period. The true culprit is the correlated drawdown across the seven currency pairs. During poor market regimes, all pairs move against the position simultaneously. This is a structural risk that the separate research track’s simplified model overlooked.

Strengthening the strategy

My research confirms that the core breakout_h1 strategy is sound, but it requires hardening to be viable. The refined configuration is:

1. Lower Risk: Cap risk at 0.15% per pair to pass M1 intraday stress tests.
2. Overlay: Keep it OFF.
3. Sleeves: Reject the short sleeve entirely. My EA Core System v1.2.0 effectively incorporates these lessons by using multi-timeframe diversification (H1/H4/D1) to lower intraday risk, rather than relying on flawed overlay or hedging logic. The strategy is now more robust and is focused on the trend-following kernel that actually delivers.

How this connects

This verification builds on earlier ones (what failed before and what I tried this time, comparisons between approaches).

• The long-only trend strategy I had written off came back at…
• Hunting exhaustively for a second source of returns
• Every way to raise monthly returns, listed and tested