Strategy optimization glossary.

Optimization basics

Optimization cycle

One trial in an optimization run. TradingTune sets a single combination of parameter values, runs the backtest, and records the result. The free tier allows 10 cycles per day, and Pro removes the daily limit.

Parameter

An adjustable input of a strategy, such as a moving-average length or a stop-loss percentage. Optimization searches for the parameter values that perform best on your data.

Parameter range (min, max, step)

The set of values the optimizer is allowed to try for a parameter, defined by a minimum, a maximum, and a step size between them. Wider ranges and finer steps create a much larger search space.

Backtest

A simulation of how a strategy would have performed on historical price data. Every optimization cycle runs one backtest. See real backtests compared to buy and hold . A backtest measures the past only and does not guarantee future results.

Robustness and validation

In-sample vs out-of-sample

In-sample data is the period you optimize on. Out-of-sample data is a later period the optimizer never saw, used to check whether tuned settings generalise. Learn how to use this to avoid overfitting .

Walk-forward

A validation approach where you optimize on one window, test on the period right after it, then roll both windows forward and repeat. Consistent out-of-sample performance across each fresh window is strong evidence of a durable edge.

Overfitting (curve-fitting)

Tuning a strategy so tightly to one slice of history that it learns the noise rather than a repeatable pattern. Overfit settings look great in the backtest and tend to fail live. Our guide to avoiding overfitting covers the warning signs and the fixes.

Result metrics

Net profit

The total profit or loss a parameter combination produced over the backtest period. The headline metric in the results table, but best read alongside drawdown, Sharpe, and trade count.

Win rate

The share of trades that closed in profit. A high win rate can still lose money if the losses are large and the wins small, so read it together with net profit and drawdown.

Sharpe ratio

A measure of return earned per unit of volatility. A higher Sharpe ratio means the gains came more smoothly, which usually indicates a more comfortable, more robust strategy to trade.

Max drawdown

The largest peak-to-trough drop in equity over the backtest. It represents the worst-case loss on paper, and a smaller drawdown is often worth accepting a slightly lower net profit for.

Trade count

How many trades a combination took during the backtest. Too few trades and the other metrics are mostly luck, so prefer results backed by a meaningful sample size.

Optimization methods

Brute force

An exhaustive search that tries every combination of every parameter value. It guarantees the best result within your ranges, but is only practical for a few parameters. See the brute force method .

Simulated annealing

A probabilistic search that explores widely at first, then settles toward an optimum as it cools, letting it escape shallow local optima on bumpy landscapes. See the simulated annealing method .

Bisection

A fast, deterministic search that brackets the best region of each parameter with a coarse sweep, narrows in with golden-section steps, then polishes the top values. See the bisection method .

TPE (Bayesian)

Tree-structured Parzen Estimator, a Bayesian sampler that builds a model of which settings work and spends cycles where the payoff is most likely. It is TradingTune's default. See the TPE Bayesian method .

Where to go next

Now that the vocabulary is clear, see how to choose a strategy optimizer against the criteria that matter, or learn more about TradingTune and the local-first model behind it.