Skip to content
TradingTune

Random

Uniform random sampling. A surprisingly tough baseline.

Quality
Medium
Speed
Fast
Budget
O(N)
Add to Chrome, it's free

How it works

Random search is the honest baseline. Each cycle it picks a brand new, fully independent set of values from your ranges. No memory, no learning, no pattern.

It sounds naive, but it is genuinely hard to beat for a quick read on a new strategy, and it never gets stuck the way a structured search can.

Think of it as the control group: if a smarter method cannot beat random on your strategy, that tells you something useful about the landscape.

Under the hood

Every parameter is sampled independently each cycle: a step-aligned uniform draw for numerics and a 50/50 coin flip for booleans. There is zero correlation between cycles.

Bergstra and Bengio showed in 2012 that random search beats grid search for high-dimensional tuning, because it does not waste cycles re-testing the same value of one parameter. TPE in turn beats random by learning, which is why TPE is the default.

TPE itself falls back to random before it has enough history to model, so random is also the warm-up that bootstraps the smarter method.

Memory
None
Numeric draw
Uniform, step-aligned
Boolean draw
50 / 50

When to use it

Reach for it when

  • A fast, unbiased first read on a new strategy
  • A baseline to measure smarter methods against

Reach for something else when

  • When you want to converge on the best settings (use TPE)
  • Expensive cycles where sample efficiency matters

Want to see optimization in action? Browse our real strategy backtests vs buy and hold on NVDA, TSLA, and AAPL.

New to this? Read how to run your first optimization or learn how to choose the right optimization method for your strategy.

Related methods