Skip to content
TradingTune
Read the neighborhood Advanced

Parameter sensitivity analysis

The single best-scoring combination is often the most dangerous one to trade. What you actually want is a setting whose neighbours also work, because in live trading you never land exactly on the peak. Sensitivity analysis is how you tell a sturdy plateau from a knife-edge spike.

Add to Chrome, it's free

The peak that is not real

Optimizers report the single best combination, and that combination is frequently a spike: a lonely high point surrounded by mediocre or losing neighbours. It scores well because it happened to thread the noise of the backtest perfectly. Trade it and you will not reproduce that thread, because the future is a different sequence. A spike is overfitting wearing the costume of a great result.

A broad plateau versus a lonely spike. The plateau keeps working when the parameter drifts; the spike collapses the moment it does.

Plateau over spike

A plateau is a region of parameter space where many nearby settings all perform well. It means the strategy is exploiting something real and durable, not a coincidence at one exact value. When you must choose, prefer a slightly lower-scoring setting in the middle of a plateau over the single highest spike. You give up a little backtest profit and gain a lot of live robustness.

Tip

Score the neighbourhood

Instead of the single best metric, look at the average metric of a setting and its immediate neighbours. The setting with the best neighbourhood average is usually the one to trade.

Mapping the surface

To see the shape of the surface, sweep each parameter coarsely across its full range with sensible min, max, and step, then look at how the metric changes as the value moves. A robust parameter shows a smooth hump: performance rises, holds across several adjacent values, and falls gently. A fragile one shows a single spike with cliffs on both sides. TradingTune's sortable results table lets you order by metric and scan whether the top combinations cluster together (a plateau) or scatter (noise).

A coarse sweep to map sensitivity (JSON)

JSON

Wide ranges and coarse steps reveal the shape of the surface cheaply. Refine only inside the plateau you find, not around an isolated spike.

{
  "Fast MA Length": { "min": 5,  "max": 80, "step": 5, "enabled": true },
  "Slow MA Length": { "min": 50, "max": 300, "step": 25, "enabled": true },
  "ATR Stop Mult":  { "min": 1.0, "max": 6.0, "step": 0.5, "enabled": true }
}

From peak-hunting to robustness

The mindset shift is the whole point: stop hunting for the highest number and start hunting for the most stable region. A setting that survives small perturbations is a setting that has a chance of surviving the market. Pair sensitivity analysis with out-of-sample testing and walk-forward analysis and you have a genuine robustness toolkit rather than a backtest-flattering one.

Key idea

The one rule

Trade the middle of a broad plateau, not the single highest spike; a setting whose neighbours also work is a setting you can trust.

Key takeaways

  • The single best combination is often an overfit spike.
  • Prefer a plateau where nearby settings also perform well.
  • Score the neighbourhood average, not just the peak metric.
  • Sweep coarsely to map the surface, then refine inside the plateau.

Frequently asked questions

What is parameter sensitivity analysis?

Testing how a strategy's performance changes as each parameter moves around its optimized value. A robust parameter shows a broad plateau where nearby values also work; a fragile one shows a lonely spike.

Why is the best-scoring combination risky?

The single highest score is often a spike that threaded the noise of one backtest perfectly. The future is a different sequence, so the spike does not reproduce. A plateau setting is exploiting something more durable.

How do I find a robust setting instead of a spike?

Sweep each parameter coarsely to map the surface, then prefer a setting in the middle of a region where many neighbours also perform well, scoring the neighbourhood average rather than the single peak.

Share this guide
X LinkedIn Email

Related guides

All guides

Put it into practice

Install TradingTune and optimize any TradingView strategy right in your browser. Free tier, no API keys. A free account is required to run.

Add to Chrome, it's free