Binance Futures Strategy Analysis
Overview
Use this skill to produce strategy decisions for Binance U-margined perpetual contracts from public market data only. Do not place orders, do not call any private API, and do not use spot market data.
Workflow
- 1. Read the canonical strategy source at
../../../src/main/resources/system-prompt.txt. - Read
references/system-prompt-coverage.md before analysis. It mirrors the canonical rules in skill-friendly form and includes the analysis-only output adaptation. - Read
../../../docs/core-feature-spec.md only if you need the project's original execution flow and field semantics. - Read
references/analysis-mode.md before deciding, because this skill runs in analysis-only mode instead of live execution mode. - Normalize user input symbols to Binance USDT perpetual symbols:
-
BTC ->
BTCUSDT
-
ETH ->
ETHUSDT
-
BTCUSDT stays
BTCUSDT
- 6. Run
python3 scripts/binance_futures_snapshot.py <SYMBOL...> with the user symbols. - Use the script output as the only market data source unless the user gives additional account context.
- Apply the strategy rules from
system-prompt.txt exactly. Do not simplify the decision logic just because this skill is analysis-only. - Build the internal decision object in the canonical schema first, then render it as a human-readable structured report unless the user explicitly asks for raw JSON.
Analysis Mode Rules
- - Default to analysis-only assumptions when the user gives only symbols:
-
equity_usdt = 10000
-
free_usdt = 10000
-
positions = []
-
cooldown_ok = true
- - Replace those defaults if the user provides account equity, free balance, cooldown status, or current positions.
- Because there is no live position context by default:
- Use
BUY,
SELL, or
HOLD.
- Do not emit
ADD,
REDUCE, or
CLOSE unless the user explicitly supplies an existing position.
- - Still fill every required field from the canonical schema:
-
ts
-
instId
-
action
-
position_mode
-
size_usdt
-
order_type
-
stop_loss_px
-
exit_plan
-
reason
-
confidence
-
leverage
-
cooldown_ok
- - Use
instId in BTC-USDT style in the internal canonical decision. The script's binance_symbol is only for data fetching. - Set
order_type to MARKET. - Keep
reason in simplified Chinese and make it cite actual sequence evidence from the script output. - Respect the canonical leverage ladder, confidence buckets, sizing rules, and exit-plan rules even when the final answer is prose.
Default Output Format
Default to a structured Chinese analysis report, not raw JSON.
Include these sections in order unless the user explicitly asks for a different format:
- 1. INLINECODE40
- State the final recommended action:
BUY /
SELL /
HOLD.
- State whether the recommendation is immediate action or wait-for-confirmation.
- 2. INLINECODE44
- Explain the 2H strategic direction.
- Explain the 15m execution structure.
- Mention 5m only as supporting context.
- 3. INLINECODE45
- Explain why the chosen action is valid.
- Explain why the opposite direction is not preferred.
- Explain why now is or is not the right timing.
- 4. INLINECODE46
- List near support, key support, near resistance, and key resistance.
- Distinguish short-term levels from higher-timeframe levels.
- 5. INLINECODE47
- Explain what price action would upgrade to
BUY.
- Explain what price action would upgrade to
SELL.
- Explain what invalidates the current bias.
- 6. INLINECODE50
- Give stop-loss logic, target logic, and management logic in plain Chinese.
- Mention if current risk/reward is insufficient.
If the user explicitly asks for JSON, return the canonical JSON array instead.
How To Read The Script Output
Focus on these sections:
- Use for strategic trend class:
STRONG_UP,
MODERATE_UP,
MODERATE_DOWN,
STRONG_DOWN.
- Check price structure, EMA alignment, MACD histogram, RSI, ADX, swing highs/lows, reversal candles, volume anomalies, and volume profile.
- Use for execution quality.
- Check breakout confirmation, pullback hold, EMA21 behavior, ATR-based stop distance, recent swing levels, and risk/reward estimates.
- Use only as short-term context. Do not let it override the 2H strategic filter.
- Use for cost evaluation.
- Use spread and slippage estimates to judge whether execution cost is acceptable.
- Use as supporting context, not as a standalone trigger.
Decision Discipline
- - Enforce the 2H directional filter first.
- Require confirmed 15m structure before any
BUY or SELL. - Prefer
HOLD when confirmation is missing, risk/reward is poor, ATR expansion is too aggressive, or cost is unattractive. - For counter-trend setups, allow only the weaker reconnaissance logic that the canonical prompt permits.
- Always produce an
exit_plan with numeric tp and sl, plus a Chinese inv management rule. - If the data is incomplete or inconsistent, output
HOLD. - Internally evaluate all canonical controls:
- trend class
- entry confirmation
- counter-trend exception handling
- leverage bucket
- nominal size
- worst-case risk
- exit-plan priority
- cooldown
- simultaneous-position cap
- confidence bucket
Coverage Check
This skill is not allowed to rely on a partial summary anymore. Before answering, make sure your reasoning covers all canonical rule groups listed in references/system-prompt-coverage.md.
Script
- - Main helper: INLINECODE70
- Purpose:
- Fetch Binance U-margined perpetual public data
- Compute the indicators and structural helpers required by the strategy
- Output a compact JSON snapshot for one or more symbols
Output Contract
- Return a structured Chinese market analysis report.
- Include a single explicit final recommendation per symbol:
BUY,
SELL, or
HOLD.
- Include detailed reasons, support and resistance, invalidation conditions, and next-step triggers.
- Return one decision object per requested symbol in a single JSON array.
- Keep the field order exactly aligned with the canonical prompt.
If the user asks for BTC ETH, analyze both symbols separately in the same response.
币安合约策略分析
概述
使用此技能仅根据公开市场数据生成币安U本位永续合约的策略决策。不下达订单,不调用任何私有API,不使用现货市场数据。
工作流程
- 1. 读取位于../../../src/main/resources/system-prompt.txt的规范策略源文件。
- 在分析前先读取references/system-prompt-coverage.md。该文件以技能友好形式镜像规范规则,并包含仅分析模式的输出适配。
- 仅在需要了解项目原始执行流程和字段语义时,读取../../../docs/core-feature-spec.md。
- 在决策前读取references/analysis-mode.md,因为此技能以仅分析模式而非实时执行模式运行。
- 将用户输入的交易对标准化为币安USDT永续合约交易对:
- BTC -> BTCUSDT
- ETH -> ETHUSDT
- BTCUSDT 保持不变
- 6. 使用用户指定的交易对运行python3 scripts/binancefuturessnapshot.py 。
- 除非用户提供额外的账户上下文,否则仅使用脚本输出作为市场数据源。
- 严格应用system-prompt.txt中的策略规则。不得因本技能仅为分析模式而简化决策逻辑。
- 首先按规范模式构建内部决策对象,然后将其呈现为人类可读的结构化报告,除非用户明确要求原始JSON格式。
分析模式规则
- equity_usdt = 10000
- free_usdt = 10000
- positions = []
- cooldown_ok = true
- - 如果用户提供账户权益、可用余额、冷却状态或当前持仓,则替换这些默认值。
- 由于默认无实时持仓上下文:
- 使用BUY、SELL或HOLD。
- 除非用户明确提供现有持仓,否则不输出ADD、REDUCE或CLOSE。
- ts
- instId
- action
- position_mode
- size_usdt
- order_type
- stop
losspx
- exit_plan
- reason
- confidence
- leverage
- cooldown_ok
- - 内部规范决策中使用BTC-USDT格式的instId。脚本中的binancesymbol仅用于数据获取。
- 将ordertype设置为MARKET。
- reason使用简体中文,并引用脚本输出的实际序列证据。
- 即使最终答案为文字描述,仍需遵循规范的杠杆阶梯、置信度区间、仓位规模规则和退出计划规则。
默认输出格式
默认输出结构化中文分析报告,而非原始JSON。
除非用户明确要求不同格式,否则按以下顺序包含这些部分:
- 1. 结论
- 说明最终推荐操作:BUY / SELL / HOLD。
- 说明该推荐是立即执行还是等待确认。
- 2. 趋势判断
- 解释2小时战略方向。
- 解释15分钟执行结构。
- 仅将5分钟作为辅助上下文提及。
- 3. 关键理由
- 解释为何所选操作有效。
- 解释为何不偏好相反方向。
- 解释为何当前时机合适或不合适。
- 4. 支撑与阻力
- 列出近期支撑、关键支撑、近期阻力、关键阻力。
- 区分短期级别与更高时间框架级别。
- 5. 触发条件
- 说明何种价格行为会升级为BUY。
- 说明何种价格行为会升级为SELL。
- 说明何种情况会否定当前偏向。
- 6. 风控与计划
- 用中文说明止损逻辑、目标逻辑和管理逻辑。
- 若当前风险回报比不足,需提及。
如果用户明确要求JSON,则返回规范JSON数组。
如何读取脚本输出
重点关注以下部分:
- 用于战略趋势分类:STRONG
UP、MODERATEUP、MODERATE
DOWN、STRONGDOWN。
- 检查价格结构、EMA排列、MACD柱状图、RSI、ADX、摆动高低点、反转K线、成交量异常和成交量分布。
- 用于执行质量评估。
- 检查突破确认、回调支撑、EMA21行为、基于ATR的止损距离、近期摆动水平和风险回报估算。
- 仅作为短期上下文使用。不得覆盖2小时战略过滤器。
- 用于成本评估。
- 使用价差和滑点估算判断执行成本是否可接受。
- 作为辅助上下文使用,不作为独立触发条件。
决策纪律
- - 首先强制执行2小时方向过滤器。
- 在任何BUY或SELL之前,要求确认15分钟结构。
- 当确认缺失、风险回报比不佳、ATR扩张过于激进或成本不具吸引力时,优先选择HOLD。
- 对于逆势设置,仅允许规范提示中允许的较弱侦察逻辑。
- 始终生成包含数值tp和sl的exit_plan,以及中文inv管理规则。
- 如果数据不完整或不一致,输出HOLD。
- 内部评估所有规范控制项:
- 趋势分类
- 入场确认
- 逆势异常处理
- 杠杆区间
- 名义规模
- 最坏情况风险
- 退出计划优先级
- 冷却状态
- 同时持仓上限
- 置信度区间
覆盖检查
此技能不得再依赖部分摘要。在回答前,确保推理覆盖references/system-prompt-coverage.md中列出的所有规范规则组。
脚本
- - 主要辅助工具:scripts/binancefuturessnapshot.py
- 目的:
- 获取币安U本位永续合约公开数据
- 计算策略所需的指标和结构辅助信息
- 输出一个或多个交易对的紧凑JSON快照
输出约定
- 返回结构化中文市场分析报告。
- 每个交易对包含一个明确的最终推荐:BUY、SELL或HOLD。
- 包含详细理由、支撑与阻力、失效条件和下一步触发条件。
- 在单个JSON数组中返回每个请求交易对的一个决策对象。
- 字段顺序严格与规范提示保持一致。
如果用户询问BTC ETH,在同一响应中分别分析两个交易对。