Reference

Glossary

Data

Bar / OHLC — one time-slice of market data: open, high, low, close (plus volume).
Timeframe — the bar size, e.g. 5m = five-minute bars.
Dataset — one instrument at one bar size in the catalog, e.g. MES 5m.
Resample — building a coarser dataset from a finer one (1m → 5m). See Importing data.
RTH / ETH — regular trading hours (09:30–16:00 ET for US index futures) / extended (overnight) hours.
BarsExporter — the bundled NinjaTrader 8 add-on that exports bars as CSV. See Exporting data.

Genome / candidate — one strategy as the optimizer sees it: an entry-condition tree plus exit genes and parameter values.
Population — the set of candidates alive in one generation.
Generation — one full evaluate-select-breed cycle.
Crossover — breeding two parents by exchanging parts of their genomes.
Mutation — random perturbation of parameters or logic.
Elitism — copying the best candidates unchanged into the next generation.
Seed — the random-number starting point; same seed + same inputs = identical run.
Diversity — how genetically varied the population is (collapsing diversity = converging search).
Fitness — the score the optimizer maximizes. See Fitness & metrics.
Composite — the default fitness: a production blend of return, risk, and consistency.

In-sample (IS) — data the optimizer saw while evolving (green equity segments).
Out-of-sample (OOS) — fold windows the optimizer never touched (magenta).
Incubation — the newest ~10% of the dataset, held out entirely (yellow).
Walk-forward (WF) — tiling the data into rolling IS→OOS folds. A fold is one such tile.
WF efficiency — mean OOS/IS performance ratio; ~1.0 excellent, <0.5 the edge halves out-of-sample.
Consistency — the fraction of folds profitable out-of-sample.
Monte Carlo (MC) — re-running the trade sequence many times with shuffled order and random slippage to see the distribution of outcomes, not one path.
CVaR₁₀ — the mean of the worst 10% of Monte Carlo outcomes.
Parameter stress — jittering numeric parameters ±25% and measuring how much fitness survives. See Understanding validation.
Gate — a hard floor a candidate must clear. See Quality gates.
Verdict — the gate roll-up: Pass (all clear), Marginal (soft check failed), Kill (hard gate failed). See Verdicts & robustness.
Robustness — the 0–100 composite of gate ratio, WF efficiency, Monte Carlo, and param stress.
HBT — an in-house cross-check against a NinjaTrader-parity reference engine; "not configured" on standard installs.

Net PnL — dollars won minus dollars lost.
Profit factor (PF) — gross profit ÷ gross loss.
Expectancy — average P&L per trade.
Win rate — winning trades ÷ total trades.
Max drawdown (DD) — largest peak-to-trough equity drop.
Recovery factor — net profit ÷ max drawdown.
Sharpe / Sortino — return per unit of (downside) volatility.
R-multiple — a result expressed as a multiple of the initial risk; a 2R win made twice what the stop risked.
Equity mode — how the account base evolves in backtests: Fixed, Daily reset, or Compounding.

Trailing stop — a stop that follows price as the trade moves favorably.
Breakeven move — moving the stop to entry after a trigger.
Partials — scaling out of part of the position at a target.
Time stop — force-exit after a maximum number of bars in the trade.
Reversal exit — an indicator condition that flattens (or experimentally reverses) an open position. See Exit settings.

Spec (.vsb.json) — the portable, hash-stamped definition of a strategy. See Strategy specs.
Standalone export — the generated .cs inheriting NinjaTrader's own Strategy; no other files needed.
Repainting — an indicator revising its past values; excluded from the palette by construction.
Seat — one activated device under a license (two per license).
License token / heartbeat — the locally-verified proof of license (~3-week life) and the silent background renewal. See Licensing.
Prop firm mode — evaluating candidates inside a funded-account ruleset. See Prop firm mode.