Glossary

Malgot has a small, precise vocabulary. Learning these terms once makes every other page easier.

System & structure

• Trading system — the deployable unit: one or more strategies (sleeves) plus a broker and signal allocator, sized against a starting capital.
• Strategy — a Python class (Malgot.Strategy) that emits signals each bar. The decision logic.
• Sleeve (trading-system-strategy) — one strategy's slot inside a system. Parameters and positions are scoped per sleeve, so two sleeves using the same indicator stay fully independent.
• Indicator — a reusable computed series (SMA, RSI, Z-Score…), linked to a strategy and read back by name.
• Observer — records one value per bar for charting (equity, exposure, cash ratio…). Your window into why the curve moved.
• Metric — a single end-of-run number (CAGR, Sharpe, max drawdown…).

Execution & trading

• Signal — a strategy's intent to trade: Malgot.Signal(symbol, side, size), optionally with order type and risk legs (stop-loss, take-profit).
• Signal allocator — turns raw signals into correctly sized orders against available capital.
• Order — an instruction sent to the broker (market, limit, stop, trailing…), with a lifecycle: pending → executed / cancelled / rejected.
• Trade — a completed round-trip (entry + matching exit), carrying realised PnL. One trade can span several orders.
• Position — current holding in a symbol, keyed by ticker. Negative quantity = a short.
• Broker — the cost & rules model: commission, slippage, short-margin requirement, short-borrow rate. Simulation and (live) real-broker variants share the interface.

Money

• Cash — free balance; grows on sell fills, shrinks on buy fills. Short-sale proceeds land here.
• Equity / portfolio value — cash + Σ(quantity × price). Shorts (negative quantity) reduce it. This is the true account value and the basis for the equity curve and CAGR.

Runs & data

• Quick Sim — a fast single-strategy backtest with one execution per system.
• Execution — one simulation run. Its execution_id isolates that run's time-series and result artefacts from every other.
• Trading state — the full recorded state of a run (cash, positions, orders, signals, trades, series) — what the detail view reads.

Workflow terms

• In-sample (IS) / out-of-sample (OOS) — data used to choose parameters vs. data reserved to test them honestly.
• Walk-forward — repeatedly optimize on IS, validate on the next OOS window, roll forward. See Optimization.
• Overfitting — fitting past noise so well that the system predicts nothing.
• Plateau vs. spike — a broad region of good parameters (robust) vs. a lone good point surrounded by bad ones (luck). See Sensitivity.