Swing EA to Scalping Transition: Should we focus on Optimization or Architectural Overhaul?

[Yusuf Uzun]

Everyone, greetings!

I've been working on a personal EA (Expert Advisor) for swing trading for quite some time now, and the results are really solid. However, when I try to adapt it for scalping, I just can't seem to achieve the same consistency and harmony.

Here are the main issues I'm facing and what I'm wondering about:

• Timeframe Impact: Simply lowering the timeframe doesn't produce the expected result on its own; market noise disrupts the strategy even though the logic remains identical.
• Optimization Dead End: When I run genetic optimization, the results look promising at first, but the parameters fail to deliver the desired stability across real-world market dynamics and different market regimes—it practically ends up overfitting.

In your opinion, should I stick with just parameter optimization, or should I go for a fundamental architectural overhaul on the coding side (such as entry/exit filters, spread/commission cost calculations, or indicator logic)?

I would love to hear the advice and insights of anyone who has experienced this transition before. Thanks in advance!

[Lungile Mbanjwa]

Hello everyone,

I think the answer depends on how the EA was originally designed and the strategy behind its development.

In my own experience, I once modified one of my existing EAs to introduce a scalping mode. However, the results and behaviour were not what I expected. I eventually removed the scalping mode, and the EA worked better with its original trading approach.

From my perspective, if an EA was initially developed for swing trading, it may be better to consider building a separate project specifically for scalping rather than trying to force a different trading style into the existing architecture.

Scalping and swing trading can have different entry conditions, trade management, stop-loss requirements and exit strategies. Therefore, the decision between optimization and an architectural overhaul depends on the original EA structure and the scalping strategy you want to implement.

Sometimes starting from scratch is a better solution.

Thank you, and I would be interested to hear how other developers approach this transition.

[Jevon Naldo Yoshield]

Price the cost at your scalping horizon before the next optimization run. That ratio decides whether you need optimization or an overhaul. I ran a pre-registered short-horizon FX rule over 2012-2025. Mean gross was +0.03 bp per trade against 1.72 bp round-trip cost, and the book lost 22.44% a year net. A sign-permutation test put it at the 75.70th percentile against a 99.5th threshold, so I closed it rather than re-parameterize. At a scalping horizon the gross move is small while the spread and commission do not shrink with it, so put your average move per trade at that timeframe against the round-trip cost.