Planned features

A canonical, honest list of features the docs reference that aren't yet implemented. Each entry links to the phase that will ship it.

If you're reading this because a recipe or example mentioned something doesn't work yet — sorry. We over-described some surface in earlier drafts and we're catching up. This page is the single source of truth for "what's coming."

Phase 24 — risk-sizing primitives

Tracked in docs/backlog.md. ETA: short — well-defined parser/sizer work.

Feature	What you'll be able to write	Workaround today
SIZING N PCT RISK	BUY btc SIZING 1.0 PCT RISK STOP_LOSS AT btc.close - atr(btc, 14) * 2	Compute size manually: SIZING <usd> USD after running the math yourself
WARMUP N BARS	Explicit warmup declaration at the top of STRATEGY blocks	The compiler infers warmup from indicator periods automatically — usually enough
IS NULL / IS NOT NULL	WHEN ema(btc.close, 200) IS NOT NULL AND ...	Lean on the implicit null-propagation behaviour — comparisons with null evaluate to false, so rules silently won't fire during warmup
FLATTEN DSL action	THEN FLATTEN (closes every position across symbols)	Use CLOSE_ALL which already exists

Phase 25 — operator tooling + DSL extensions

Tracked in docs/backlog.md. Heavier scope; multiple sub-tasks.

Feature	What you'll get	Workaround today
qkt fetch <SYMBOL> --from --to CLI	Populate the local data store from a fetcher (Dukascopy, etc.) without leaving the shell	Use ./scripts/fetch-dukascopy.sh directly; or write data via the DataStore Kotlin API
qkt sweep CLI	Parameter grid-search wrapper around the existing BacktestSweep Kotlin harness	The harness exists today — use it programmatically (see src/main/kotlin/com/qkt/backtest/sweep/)
qkt walkforward CLI	Walk-forward validation wrapper	BacktestSweep + rolling time-windows manually
TRAILING_STOP BY <amount> wiring	The token exists in the parser; finishing the AST → action-compiler → broker dispatch path	Add a rule that closes when the trade drops below a moving high — manual trail
VWAP DSL registration	VWAP(stream, period) callable from .qkt files	The VWAP Kotlin class exists; needs IndicatorInput.TICK_SERIES plumbing first. Use SMA on stream.close as a proxy until then
per_strategy: risk-config block	Per-strategy risk rules (e.g. cooloff-after-loss for one strategy without affecting others)	Use daemon-wide rules in risk: rules: for now
IF_TOUCHED, STOP_LIMIT order types	Additional order shapes beyond MARKET/LIMIT/STOP	Compose manually with multiple rules + WHEN triggers

Phase 26a — pending-entry OCO DSL surface + clock accessors

Spec: docs/superpowers/specs/2026-05-11-phase26-pending-oco-and-clock-design.md. DSL surface only — hedge-straddle becomes authorable, parseable, and backtestable. Live MT5 runtime ships in 26b.

Feature	What you'll be able to write	Workaround today
OCO_ENTRY { BUY ..., SELL ... }	Two pending entries linked one-cancels-other; whichever fills, the other auto-cancels	None — the only way today is two independent Signal.Submit calls from Kotlin code with no link
NOW.hour_utc, NOW.minute_utc, NOW.weekday, NOW.date_utc, NOW.epoch_ms	Session-window gating from inside .qkt files	None — strategies can't read the wall clock at all from the DSL
TIF GTD UNTIL now + <duration>	Auto-expire pending orders after a relative window	Use TIF GTD UNTIL <absolute-epoch-ms> and compute the timestamp at strategy-author time (only works in offline contexts)
Mock broker OCO fidelity verified	Confidence that backtests of OCO_ENTRY strategies are deterministic and correct	None — StandaloneOCO mock-broker behavior is currently unverified

Phase 27 — conditional bracketed stacks

Spec: docs/superpowers/specs/2026-05-12-phase27-conditional-bracketed-stacks-design.md. Adds independent stacks during a live position's lifecycle, MFE-and-time gated, each with its own bracket. Unlocks the full hedge-straddle P&L profile (Phase 26 ports the pre-stack version which captures the strategy logic but misses ~148% of the P&L per the production backtest). Requires a real model change: multi-position-per-symbol tracking with per-position lifecycles.

Feature	What you'll be able to write	Workaround today
STACK_AT MFE >= N WITHIN M BRACKET { ... } (one clause per stack tier)	Independent micro-trades opened opportunistically when the primary winner proves conviction	None — qkt's STACK clause models pyramiding-into-trend with shared bracket and sequential triggering; hedge-straddle stacks need per-layer brackets, simultaneous firing, and conditional triggering during WINNER phase
LegBook replaces singular Position per symbol	Each stack tracked with its own broker ticket, leg id, and bracket	None — the position model assumes one net position per symbol
POSITION.<stream>.mfe accessor	Max favorable excursion since the position opened, available in DSL conditions	None — strategies can't read MFE from the DSL
OrderTypeCapability.MULTI_POSITION_PER_SYMBOL declaration	Compile-time error if a strategy uses STACK_AT against a netting-only broker (Bybit Spot, Bybit Linear in one-way mode)	None — runtime failures only

Phase 28+ — exploratory

These are ideas with no scheduled phase yet. Tell us if you'd use them (open an issue):

• Multi-leg positions per symbol (simultaneous long + short, separate brackets) — needed for legacy market-mode hedge-straddle; production hedge-straddle uses pending mode and doesn't need this
• position(stream).is_long / .is_short convenience accessors
• True per-symbol realized P&L (today, POSITION.<stream>.realized_pnl returns strategy-level)
• Block bootstrap in Monte Carlo (today: i.i.d. bootstrap; bad for clustered trades)
• Order-book-aware backtesting
• More broker integrations (Alpaca, IBKR, OANDA)

How to read this list

If an admonition in another doc page sent you here, you're trying to use a feature that's announced but not shipped yet. Stick to the workaround until the phase that ships it merges. The phase changelogs at /phases/ document each shipping wave.

If you find docs referencing a feature that should be on this page but isn't, open an issue. The aim is zero gap between docs and engine; we're aware this list shrinks toward empty.