is the mql5 cloud network worth paying for optimization or use my own cores

[richardlee]

running big optimization passes on my EA and my desktop takes hours per run. the mql5 cloud network offers to throw hundreds of agents at it for a fee. is it actually worth it, or do i just leave my machine grinding overnight and save the money?

[thomas_x]

cloud is worth it only when your bottleneck is genuinely compute and your test is sound. the trap is people pay for massive cloud optimization to brute-force thousands of parameter combinations, which is just industrial-scale curve fitting. faster overfitting is still overfitting. fix the methodology before you rent the horsepower.

[Oliver]

this. a focused optimization over a few meaningful parameters runs fine on your own cores overnight. if you need hundreds of cloud agents, your parameter space is probably too big, which is a design smell, not a hardware problem. narrow the search first.

[Chris]

i use cloud occasionally for walk-forward analysis where i genuinely need many independent windows tested, that workload parallelises honestly and the cost is justified. for a single grid search on five parameters, my own machine overnight is plenty. match the spend to whether the work is genuinely parallel and genuinely needed.

[danielw_]

renting hundreds of cores to optimize a retail EA is like putting a turbo on a shopping trolley. the limiting factor was never your cpu, it was the idea.

[fadedlights]

own cores overnight, fix the method first.