Trade decision

Forecast drives the plan

Direction and suggested exits come from exactly the same anchored and aggregated forecast bars displayed on the chart, using the qualified one-hour horizon and automatic live-price anchor. Four-hour and 24-hour views remain exploratory and cannot generate model-led plans. Direction compares the displayed endpoint close with the latest traded candle close: (endpoint / traded close − 1) × 100. Positive means long, negative means short, and a finish within 0.01% is Flat. Independent sample-direction statistics do not override the displayed path.

No strategy selection or forming setup is required. Selected strategies are supporting context only: the panel counts fresh forming candidates that agree or oppose and lists their names, sides and research-coverage limitations in Plan details. Opposing candidates and untested strategy coverage do not block, reverse or rewrite the forecast plan. Strategies remain experimental; no strategy passed every original validation gate.

Suggested entry, stop and take profits

Entry is the latest traded close with one basis point of adverse entry slippage: traded close × (1 + side × 0.0001), where side = +1 long or −1 short. TP2 is the final displayed forecast close. TP1 is halfway between the entry reference and TP2; it is a staged planning reference, not a second independent forecast or a prescribed partial-exit allocation.

ATR14 uses Wilder-smoothed true range from at least 14 completed candles at the selected chart interval. For a long, stop = min(entry − ATR14, minimum displayed forecast low − 0.25 × ATR14). For a short, stop = max(entry + ATR14, maximum displayed forecast high + 0.25 × ATR14). Thus the stop sits beyond the adverse projected wick with a quarter-ATR buffer and at least one ATR of entry distance. Model path extrema and this buffer are an explicit experimental risk policy, not validated stop barriers or probability bounds. No forecast quantile is presented as a guaranteed loss limit.

References update with live price, new five-minute model inputs, horizon and interval changes; they are not frozen orders or a tracked position. The endpoint timestamp labels the forecast's actual final bar. A plan can have weak reward:risk: the measured net R is displayed without inventing a favorable target or shrinking the stop to force a ratio.

Show plan on chart toggles four labelled references: entry, stop, TP1 and TP2. They are drawn as chart primitives without adding candles, changing chart scale or moving the viewport. Targets use the gain color and dashed lines, the stop uses the loss color and a dotted line, and entry is neutral. Existing selected-strategy overlays remain separate annotations. Forecast-plan visibility resets when market, chart interval or horizon changes; fresh updates retain visibility and invalid/stale plans clear the lines.

When to wait

• The chart feed must be a fresh matching exchange feed, include the current candle and have valid continuous UTC-aligned history (Monday-aligned weekly bars). Receipt age must be within 45 seconds, with five seconds of future clock tolerance. Offline snapshots, gaps and stale candles provide no plan.
• Model-led planning requires the qualified one-hour forecast, the current completed five-minute input cutoff, a matching market/model, fresh exchange data and a still-future valid displayed endpoint. Longer-horizon forecasts are exploratory and do not generate model-led plans. Disabled, stale, expired or flat projections never create invented levels; model-alert settings do not enable a hidden strategy fallback.
• Positive ATR14 history and valid positive prices bracketing the entry are required. If assumed costs exceed the forecast reward, valid references remain inspectable with an explicit costs message, but sizing and chart-plan activation are unavailable. No strategy gate or fixed 1.5R threshold blocks an otherwise valid forecast plan.

Optional sizing and costs

Plan details includes an optional positive finite loss budget in USDT. With adverse stop/target fills equal to stop/target × (1 − side × 0.0001), estimated stop loss per unit = side × (entry − stop fill) + 0.0005 × (entry + stop fill). Net TP2 reward per unit = side × (target fill − entry) − 0.0005 × (entry + target fill). Net R = reward / loss, quantity = budget / loss, and notional = quantity × entry. Entry slippage is already embedded and is not charged twice.

The displayed target net assumes the whole quantity exits at TP2. It does not assume or optimize a TP1 partial allocation. TP1 may not cover costs on a small projected move. Five-basis-point fees and one-basis-point slippage per side are research assumptions, not the user's actual fee tier. Funding, gaps, variable spread/slippage, lot sizes, margin and liquidation are excluded; actual loss can exceed the budget. The budget is not persisted and resets on market/interval changes.

Validation boundary

The forecast and suggested exit policy are experimental and have not been validated as a profitable trading strategy. The chart's median OHLC path is not a measured future price range, and a projected wick does not bound actual downside. This panel suggests a transparent plan; it neither executes orders nor knows the user's fill or open position.