On 14th July, Impulse went into one of the largest grids I've seen since launch. I want to talk about it openly, because I think transparency is what separates EAs you can trust from EAs you can't — and because the event has directly shaped a set of improvements that are now in active development.
What happened on the 14th
Impulse entered a long trade on Strategy 1 that kept extending as price kept falling without a meaningful retracement. The EA managed the position according to its rules and the grid was handled as designed, however I did not design Impulse to enter such large grids... it's designed for fast trades and small grids, and the entry criteria is meant to reflect that. The entry conditions that day allowed a basket to open into an environment where momentum quickly faded, leaving it exposed. The drawdown protection and risk framework did their job, and no one can foresee the future, but a grid that size represents a failure of the entry logic, not a success of the recovery logic. That distinction matters, and it's where all of my attention is now focused.
The medium risk account
The medium risk signal hit the 50% drawdown protection level I was running on it, and trading on that account has been stopped as a result. That is exactly what the protection is there for.
I've made the decision not to resume trading on the medium risk account, but I am keeping the signal online permanently for full transparency. Nothing hidden, nothing deleted, nothing quietly restarted. You can review the complete history, including the drawdown event, at any time. I'd rather show you an honest record than a curated one.
Active Signal: https://www.mql5.com/en/signals/2375861
Discontinued Signal: https://www.mql5.com/en/signals/2380200
What I'm changing in Impulse
The 14th July event has become the reference case for the next development cycle. Here's what's being worked on:
1. Tighter entry criteria for several strategies including S01. The single biggest lever. I'm hardening the conditions under which Impulse is allowed to open a basket in the first place, with the goal of preventing entries in the kind of low volume, mean reversion environment we saw on the 14th.
2. TP scaling as the basket grows. Rather than holding a static take-profit $ target regardless of grid depth, Impulse will have the option to scale its TP deeper as the basket extends. Deeper baskets will enable Impulse to increase profits on larger retracements.
3. Control panel improvements. The on-chart panel is getting a performance upgrade to minimise CPU resources.
4. Performance of grid closures. Some users have reported seeing slippage when Impulse closes a basket of trades, so this will help to minimise any slippage experienced in real world conditions.
All changes go through my usual process: implementation, then multi-year backtest regression against the current version across multiple brokers before anything is released. Nothing ships until it's ready.
The bigger picture
Events like this are uncomfortable, but they're also the most valuable data a grid strategy can get. Backtests tell you a lot; a live stress event tells you the truth. Impulse's core edge — its read on gold's trending behaviour — remains intact, and the improvements above are about making sure the EA only deploys that edge in conditions where it actually applies.
Thank you to everyone running Impulse for your patience and your messages — the feedback from live users has been genuinely useful in shaping these priorities. The next update will be worth the wait.
As always, if you have questions, reach out any time.
— Simon
