trader-hand-skill

交易专家知识

参考知识

1. 技术分析指标参考

RSI（相对强弱指数）

Formula: RSI = 100 - (100 / (1 + RS))
Where:  RS = Average Gain / Average Loss over N periods (default N = 14)

Step-by-step calculation:
  1. For each period, compute change = Close(t) - Close(t-1)
  2. Gains = max(change, 0), Losses = abs(min(change, 0))
  3. First average: simple mean of first 14 gains/losses
  4. Subsequent: AvgGain = (PrevAvgGain * 13 + CurrentGain) / 14  (Wilder smoothing)
  5. RS = AvgGain / AvgLoss
  6. RSI = 100 - (100 / (1 + RS))

Worked example (14-period):
  Avg Gain over 14 periods = 1.02
  Avg Loss over 14 periods = 0.68
  RS = 1.02 / 0.68 = 1.50
  RSI = 100 - (100 / (1 + 1.50)) = 100 - 40 = 60.0

解读：

• RSI < 30：超卖区域（潜在买入信号）
• RSI > 70：超买区域（潜在卖出信号）
• RSI = 50：中性——价格动能均衡

RSI 高级信号：

信号	描述	强度
看跌背离	价格创出新高，RSI 创出更低的高点	强烈的反转警示
看涨背离	价格创出新低，RSI 创出更高的低点	强烈的反转警示
看涨失败摆动	RSI 跌破 30，反弹，回升至 30 以上，并突破此前 RSI 高点	非常强烈的买入信号
看跌失败摆动	RSI 升破 70，回落，反弹至 70 以下，并跌破此前 RSI 低点	非常强烈的卖出信号
区间转换	上升趋势中 RSI 在 40-80 之间波动；下降趋势中在 20-60 之间波动	趋势确认

**最佳实践：**切勿将 RSI 作为唯一信号。应结合趋势方向（移动平均线）和成交量。在强劲趋势中，RSI 可能会长时间处于超买或超卖状态。

---

MACD（移动平均收敛发散指标）

MACD Line   = EMA(12) - EMA(26)
Signal Line = EMA(9) of MACD Line
Histogram   = MACD Line - Signal Line

EMA formula: EMA(t) = Price(t) * k + EMA(t-1) * (1 - k)
Where: k = 2 / (N + 1)
  For EMA(12): k = 2/13 = 0.1538
  For EMA(26): k = 2/27 = 0.0741

Worked example:
  EMA(12) = 155.20
  EMA(26) = 152.80
  MACD Line = 155.20 - 152.80 = 2.40
  Previous Signal Line = 1.80
  Signal Line = 2.40 * (2/10) + 1.80 * (8/10) = 0.48 + 1.44 = 1.92
  Histogram = 2.40 - 1.92 = 0.48 (positive = bullish momentum increasing)

解读：

信号	条件	强度
看涨交叉	MACD 上穿信号线	中等买入信号
看跌交叉	MACD 下穿信号线	中等卖出信号
零轴看涨交叉	MACD 上穿零轴	趋势转为看涨
零轴看跌交叉	MACD 下穿零轴	趋势转为看跌
柱状图扩大	柱体变高	动能加速
柱状图收缩	柱体变短	动能减弱，可能即将反转
看涨背离	价格创出新低，MACD 创出更高的低点	强烈的反转信号
看跌背离	价格创出新高，MACD 创出更低的高点	强烈的反转信号

---

布林带

Middle Band = SMA(20)
Upper Band  = SMA(20) + 2 * StdDev(20)
Lower Band  = SMA(20) - 2 * StdDev(20)
Bandwidth   = (Upper - Lower) / Middle
%B          = (Price - Lower) / (Upper - Lower)

Worked example:
  SMA(20) = 150.00
  StdDev(20) = 3.50
  Upper = 150.00 + 2 * 3.50 = 157.00
  Lower = 150.00 - 2 * 3.50 = 143.00
  Bandwidth = (157.00 - 143.00) / 150.00 = 0.0933 (9.33%)
  Current price = 155.00
  %B = (155.00 - 143.00) / (157.00 - 143.00) = 12/14 = 0.857
  Interpretation: Price is 85.7% of the way from lower to upper band — near upper band

布林带关键信号：

信号	条件	含义
收窄	带宽处于 6 个月低位	波动性收缩，大幅波动即将到来
收窄后向上突破	收窄后价格突破上轨	强烈的看涨突破
收窄后向下突破	收窄后价格跌破下轨	强烈的看跌突破
沿上轨运行	价格紧贴上轨，且中轨上升	强劲的上升趋势——切勿做空
沿下轨运行	价格紧贴下轨，且中轨下降	强劲的下降趋势——切勿买入
均值回归触轨	价格触及外轨，%B 反转	可能回归至中轨
W 底	价格两次触及下轨，第二个低点的 %B 更高	看涨反转形态
M 顶	价格两次触及上轨，第二个高点的 %B 更低	看跌反转形态

---

VWAP（成交量加权平均价格）

VWAP = Cumulative(Typical Price * Volume) / Cumulative(Volume)
Typical Price = (High + Low + Close) / 3

Worked example (first 3 bars of the day):
  Bar 1: TP = (101+99+100)/3 = 100.00, Vol = 10,000 -> cumTP*V = 1,000,000
  Bar 2: TP = (102+100+101)/3 = 101.00, Vol = 15,000 -> cumTP*V = 2,515,000
  Bar 3: TP = (103+101+102)/3 = 102.00, Vol = 8,000  -> cumTP*V = 3,331,000
  Cumulative Volume = 33,000
  VWAP = 3,331,000 / 33,000 = 100.94

用法：

• 机构基准：价格 > VWAP 时买方占优；价格 < VWAP 时卖方占优
• 日内支撑位/阻力位：上升趋势中，VWAP 充当动态支撑位；下降趋势中，充当阻力位
• 入场过滤条件：仅在价格回落至 VWAP 时买入（不要追逐大幅延伸的行情）
• 标准差：VWAP +1/-1 和 +2/-2 标准差带可用作止盈目标
• 每日重置：切勿将 VWAP 延续至下一个交易时段——它是日内指标

---

移动平均线

SMA(N) = (Close_1 + Close_2 + ... + Close_N) / N
EMA(N) = Close * (2/(N+1)) + PrevEMA * (1 - 2/(N+1))

Key Moving Averages:
  EMA(9)   — very short-term trend (scalping, day trading)
  EMA(20)  — short-term trend
  EMA(50)  — medium-term trend
  SMA(100) — intermediate trend
  SMA(200) — long-term trend (institutional benchmark)

关键交叉信号：

交叉	名称	含义	可靠性
50 MA > 200 MA	黄金交叉	看涨趋势反转	高（滞后约 2 周）
50 MA < 200 MA	死亡交叉	看跌趋势反转	高（滞后约 2 周）
9 EMA > 21 EMA	快速看涨交叉	短期动能转变	中等
价格 > 200 SMA	高于长期趋势	看涨市场状态	极高
价格 < 200 SMA	低于长期趋势	看跌市场状态	极高

移动平均线带（20/50/100/200 MA 均呈发散状）：表明趋势非常强劲。当所有均线按顺序排列（上升趋势中为 20 > 50 > 100 > 200）时，趋势的可靠性很高。

---

ATR（平均真实波幅）

True Range = max(High - Low, |High - PrevClose|, |Low - PrevClose|)
ATR(14) = Simple or Wilder Moving Average of True Range over 14 periods

Worked example:
  Today: High = 105, Low = 101, PrevClose = 102
  TR = max(105-101, |105-102|, |101-102|) = max(4, 3, 1) = 4
  If ATR(14) was 3.50 yesterday:
  ATR(14) = (3.50 * 13 + 4) / 14 = (45.50 + 4) / 14 = 3.536

实际应用：

用例	公式	示例
止损位设置	入场价 - 2 * ATR	入场价 $100，ATR $2.50 -> 止损位 $95.00
止盈目标	入场价 + 3 * ATR	入场价 $100，ATR $2.50 -> 目标价 $107.50
仓位规模	风险金额 / ATR	$200 风险 / $2.50 ATR = 80 股
波动率筛选	ATR > 阈值	仅在 ATR > 日均值时交易（避免市场沉寂）
移动止损	最高收盘价 - 3 * ATR	随着价格上涨锁定利润

---

成交量分析

OBV (On-Balance Volume):
  If Close > PrevClose: OBV = PrevOBV + Volume
  If Close < PrevClose: OBV = PrevOBV - Volume
  If Close = PrevClose: OBV = PrevOBV

Volume Rate of Change: VROC = (Volume - Volume_N_ago) / Volume_N_ago * 100

成交量确认规则：

价格走势	成交量	解读
价格上涨	成交量增加	强劲看涨——走势可信
价格上涨	成交量减少	涨势疲弱——可能反转
价格下跌	成交量增加	强劲看跌——恐慌抛售或破位
价格下跌	成交量减少	跌势疲弱——可能接近底部
突破	成交量 > 20 日均值的 150%	突破得到确认——执行交易
突破	成交量 < 均值	可能是假突破——等待或反向交易
成交量高潮	成交量极度飙升（均值的 3 倍以上）	可能是耗竭／反转点

---

支撑位与阻力位

斐波那契回撤位：

After a move from Low (L) to High (H):
  23.6% level = H - (H - L) * 0.236
  38.2% level = H - (H - L) * 0.382
  50.0% level = H - (H - L) * 0.500
  61.8% level = H - (H - L) * 0.618  (Golden Ratio — strongest level)
  78.6% level = H - (H - L) * 0.786

Worked example (move from $80 to $120):
  Range = $40
  23.6% = 120 - 40 * 0.236 = 120 - 9.44  = $110.56
  38.2% = 120 - 40 * 0.382 = 120 - 15.28 = $104.72
  50.0% = 120 - 40 * 0.500 = 120 - 20.00 = $100.00
  61.8% = 120 - 40 * 0.618 = 120 - 24.72 = $95.28  (most likely bounce)
  78.6% = 120 - 40 * 0.786 = 120 - 31.44 = $88.56

枢轴点（标准）：

PP = (High + Low + Close) / 3
S1 = 2 * PP - High
S2 = PP - (High - Low)
R1 = 2 * PP - Low
R2 = PP + (High - Low)

Worked example (prev day: High=155, Low=148, Close=152):
  PP = (155 + 148 + 152) / 3 = 151.67
  S1 = 2 * 151.67 - 155 = 148.33
  S2 = 151.67 - (155 - 148) = 144.67
  R1 = 2 * 151.67 - 148 = 155.33
  R2 = 151.67 + (155 - 148) = 158.67

---

2. 蜡烛图形态

单根蜡烛形态

形态	信号	实体	影线	所需背景
十字星	犹豫不决	开盘价 = 收盘价（或接近）	两侧均有长影线	出现在支撑位／阻力位 = 反转
锤子线	看涨反转	较小，位于蜡烛顶部	下影线 > 实体的 2 倍	必须出现在下跌趋势底部
倒锤线	看涨反转	较小，位于蜡烛底部	上影线 > 实体的 2 倍	出现在下跌趋势底部，需确认
射击之星	看跌反转	较小，位于蜡烛底部	上影线 > 实体的 2 倍	必须出现在上涨趋势顶部
上吊线	看跌反转	较小，位于蜡烛顶部	下影线 > 实体的 2 倍	出现在上涨趋势顶部（形状与锤子线相同）
光头光脚阳线	强劲延续	完整绿色实体，无影线	无	买盘压力强劲
光头光脚阴线	强劲延续	完整红色实体，无影线	无	卖盘压力强劲
纺锤线	犹豫不决	居中的小实体	两侧影线等长	趋势可能正在失去动能
蜻蜓十字星	看涨反转	开盘价 = 收盘价 = 最高价	仅有长下影线	出现在支撑位 = 强烈反转信号
墓碑十字星	看跌反转	开盘价 = 收盘价 = 最低价	仅有长上影线	出现在阻力位 = 强烈反转信号

多根 K 线形态

形态	信号	描述	可靠性
看涨吞没	向上反转	大型绿色 K 线完全吞没前一根红色 K 线	在支撑位处较高
看跌吞没	向下反转	大型红色 K 线完全吞没前一根绿色 K 线	在阻力位处较高
早晨之星	看涨反转	红色 K 线、小实体／十字星（带跳空），以及大型绿色 K 线	非常高
黄昏之星	看跌反转	绿色 K 线、小实体／十字星（带跳空），以及大型红色 K 线	非常高
三白兵	强烈看涨	连续三根大型绿色 K 线，每根收盘价都更高	非常高
三乌鸦	强烈看跌	连续三根大型红色 K 线，每根收盘价都更低	非常高
看涨孕线	潜在反转	先是一根大型红色 K 线，随后是一根实体包含在前者实体内的小型绿色 K 线	中等（需要确认）
看跌孕线	潜在反转	先是一根大型绿色 K 线，随后是一根实体包含在前者实体内的小型红色 K 线	中等（需要确认）
镊子底	看涨反转	两根 K 线在支撑位处具有相同的最低价	高
镊子顶	看跌反转	两根 K 线在阻力位处具有相同的最高价	高
刺透形态	看涨反转	红色 K 线之后，绿色 K 线开盘价低于前者最低价，收盘价高于前者实体的 50% 位置	中等偏高
乌云盖顶	看跌反转	绿色 K 线之后，红色 K 线开盘价高于前者最高价，收盘价低于前者实体的 50% 位置	中等偏高

---

3. 风险管理公式

仓位规模（固定比例法）

Position Size (shares) = Account Risk Amount / (Entry Price - Stop Loss Price)
Account Risk Amount    = Portfolio Value * Risk Per Trade %

RULE: Never risk more than 1-2% of portfolio on a single trade.

Worked example:
  Portfolio Value = $10,000
  Risk Per Trade  = 2% ($200)
  Entry Price     = $100.00
  Stop Loss       = $95.00 (based on 2x ATR below entry)
  Risk per share  = $100.00 - $95.00 = $5.00
  Position Size   = $200 / $5.00 = 40 shares
  Position Value  = 40 * $100 = $4,000 (40% of portfolio)

  CONCENTRATION CHECK: If position value > 10% of portfolio, reduce size.
  Adjusted: max position = $1,000 / $100 = 10 shares
  Adjusted risk = 10 * $5.00 = $50 (only 0.5% of portfolio — acceptable)

凯利准则（最优下注规模）

Kelly % = W - ((1 - W) / R)
Where:
  W = win rate (decimal)
  R = average win / average loss ratio (reward-to-risk)

Worked example:
  Win rate: 60% (W = 0.60)
  Average win: $300, Average loss: $200
  R = 300 / 200 = 1.5
  Kelly = 0.60 - (0.40 / 1.5) = 0.60 - 0.267 = 0.333 (33.3%)

  Full Kelly is too aggressive for real trading. Use fractions:
  Half-Kelly  = 0.333 / 2 = 16.7% of portfolio per trade
  Quarter-Kelly = 0.333 / 4 = 8.3% of portfolio per trade (recommended)

  If Kelly is negative, the system has NEGATIVE expectancy — do not trade it.

风险价值（VaR）

Parametric VaR = Portfolio Value * Portfolio Volatility * Z-score * sqrt(Time Horizon)

Z-scores:  90% confidence = 1.282
           95% confidence = 1.645
           99% confidence = 2.326

Worked example (daily VaR, 95% confidence):
  Portfolio = $10,000
  Daily volatility (stddev of daily returns) = 2.0%
  VaR = $10,000 * 0.02 * 1.645 * sqrt(1) = $329.00
  Meaning: 95% confident daily loss will not exceed $329.

  Weekly VaR = $329 * sqrt(5) = $329 * 2.236 = $735.65
  Monthly VaR = $329 * sqrt(21) = $329 * 4.583 = $1,507.81

夏普比率

Sharpe = (Rp - Rf) / StdDev(Rp) * sqrt(252)
Where:
  Rp = mean daily portfolio return
  Rf = daily risk-free rate (Treasury yield / 252)
  StdDev(Rp) = standard deviation of daily returns
  252 = trading days per year (annualization factor)

Worked example:
  Mean daily return = 0.10% (0.001)
  Annual Treasury yield = 5.0% -> daily Rf = 0.05/252 = 0.000198
  StdDev of daily returns = 0.80% (0.008)
  Daily Sharpe = (0.001 - 0.000198) / 0.008 = 0.100
  Annualized Sharpe = 0.100 * sqrt(252) = 0.100 * 15.875 = 1.59

  Ratings:
    < 0.5  = Poor (not compensated for risk)
    0.5-1.0 = Acceptable
    1.0-2.0 = Good
    2.0-3.0 = Very Good
    > 3.0   = Excellent (verify — may indicate overfitting)

索提诺比率（仅考虑下行风险）

Sortino = (Rp - Rf) / DownsideDeviation * sqrt(252)
DownsideDeviation = sqrt(mean(min(Ri - Rf, 0)^2))

Better than Sharpe because it only penalizes downside volatility, not upside.
Sortino > 2.0 is considered very good.

最大回撤

For each point t in equity curve:
  Peak(t)     = max(Equity[0..t])
  Drawdown(t) = (Peak(t) - Equity(t)) / Peak(t) * 100%
  MaxDrawdown = max(Drawdown(t)) for all t

Worked example:
  Equity curve: $10,000 -> $12,000 -> $9,600 -> $11,500
  Peak at $12,000
  Drawdown at $9,600 = (12,000 - 9,600) / 12,000 = 20.0%
  Max Drawdown = 20.0%

Recovery Factor = Total Net Profit / Max Drawdown
  If total profit = $3,000, MaxDD = $2,400 -> RF = 3,000/2,400 = 1.25

Calmar Ratio = Annual Return / Max Drawdown
  If annual return = 25%, MaxDD = 20% -> Calmar = 1.25 (target > 1.0)

盈利因子

Profit Factor = Gross Winning Trades / Gross Losing Trades

Worked example:
  10 winning trades totaling $5,000
  8 losing trades totaling $3,200
  Profit Factor = 5,000 / 3,200 = 1.5625

  Ratings: < 1.0 = losing system, 1.0-1.5 = marginal, 1.5-2.0 = good,
           2.0-3.0 = very good, > 3.0 = excellent (verify with enough trades)

每笔交易的期望收益

Expectancy = (Win% * AvgWin) - (Loss% * AvgLoss)

Worked example:
  Win rate: 55%, Average win: $150, Average loss: $100
  Expectancy = (0.55 * 150) - (0.45 * 100) = 82.50 - 45.00 = $37.50/trade
  Over 100 trades: expected profit = $3,750

  Minimum for a viable system: Expectancy > 0 with at least 30 sample trades.

风险／回报比

R:R = (Target Price - Entry Price) / (Entry Price - Stop Loss Price)

Worked example:
  Entry = $100, Stop = $95, Target = $112
  R:R = (112 - 100) / (100 - 95) = 12 / 5 = 2.4:1

  Minimum acceptable R:R = 1.5:1
  With 40% win rate and 2:1 R:R: Expectancy = 0.40*2 - 0.60*1 = +0.20 (profitable!)
  With 40% win rate and 1:1 R:R: Expectancy = 0.40*1 - 0.60*1 = -0.20 (losing!)

---

4. Alpaca 交易 API 参考

身份验证

# Paper trading (ALWAYS start here)
BASE_URL="https://paper-api.alpaca.markets"

# Live trading (only after paper validation)
# BASE_URL="https://api.alpaca.markets"

# Data API (same for both paper and live)
DATA_URL="https://data.alpaca.markets"

# Auth headers (required on every request)
HEADERS="-H 'APCA-API-KEY-ID: $ALPACA_API_KEY' -H 'APCA-API-SECRET-KEY: $ALPACA_SECRET_KEY'"

账户信息

# Get account details
curl -s "$BASE_URL/v2/account" $HEADERS
# Key fields: id, status, equity, cash, buying_power, portfolio_value,
#   pattern_day_trader (bool), daytrade_count, last_equity

获取当前持仓

# All positions
curl -s "$BASE_URL/v2/positions" $HEADERS
# Returns array: symbol, qty, side, avg_entry_price, current_price,
#   unrealized_pl, unrealized_plpc, market_value, cost_basis

# Single position
curl -s "$BASE_URL/v2/positions/AAPL" $HEADERS

下单

# Market order (fills immediately at best available price)
curl -s -X POST "$BASE_URL/v2/orders" $HEADERS \
  -H "Content-Type: application/json" \
  -d '{"symbol":"AAPL","qty":"10","side":"buy","type":"market","time_in_force":"day"}'

# Limit order (fills only at your price or better)
curl -s -X POST "$BASE_URL/v2/orders" $HEADERS \
  -H "Content-Type: application/json" \
  -d '{"symbol":"AAPL","qty":"10","side":"buy","type":"limit","time_in_force":"gtc","limit_price":"150.00"}'

# Stop order (triggers market order when stop price hit)
curl -s -X POST "$BASE_URL/v2/orders" $HEADERS \
  -H "Content-Type: application/json" \
  -d '{"symbol":"AAPL","qty":"10","side":"sell","type":"stop","time_in_force":"gtc","stop_price":"145.00"}'

# Stop-limit order (triggers limit order when stop price hit)
curl -s -X POST "$BASE_URL/v2/orders" $HEADERS \
  -H "Content-Type: application/json" \
  -d '{"symbol":"AAPL","qty":"10","side":"sell","type":"stop_limit","time_in_force":"gtc","stop_price":"145.00","limit_price":"144.50"}'

# Trailing stop (dynamic stop that trails price by dollar or percent amount)
curl -s -X POST "$BASE_URL/v2/orders" $HEADERS \
  -H "Content-Type: application/json" \
  -d '{"symbol":"AAPL","qty":"10","side":"sell","type":"trailing_stop","time_in_force":"gtc","trail_percent":"5"}'

# Bracket order (entry + stop loss + take profit as one atomic order)
curl -s -X POST "$BASE_URL/v2/orders" $HEADERS \
  -H "Content-Type: application/json" \
  -d '{
    "symbol": "AAPL",
    "qty": "10",
    "side": "buy",
    "type": "limit",
    "time_in_force": "day",
    "limit_price": "150.00",
    "order_class": "bracket",
    "stop_loss": {"stop_price": "145.00"},
    "take_profit": {"limit_price": "165.00"}
  }'

# OCO order (one-cancels-other: stop loss OR take profit, whichever hits first)
curl -s -X POST "$BASE_URL/v2/orders" $HEADERS \
  -H "Content-Type: application/json" \
  -d '{
    "symbol": "AAPL",
    "qty": "10",
    "side": "sell",
    "type": "limit",
    "time_in_force": "gtc",
    "limit_price": "165.00",
    "order_class": "oco",
    "stop_loss": {"stop_price": "145.00"}
  }'

订单参数参考：

参数	值	备注
side	buy、sell
type	market、limit、stop、stop_limit、trailing_stop
time_in_force	day、gtc、ioc、fok	day = 收盘时取消，gtc = 取消前持续有效
order_class	simple、bracket、oco、oto	bracket = 入场 + 止损 + 目标价
qty	字符串数字	股票以整股计
notional	字符串美元金额	qty 的替代参数（零碎股）

管理订单

# List open orders
curl -s "$BASE_URL/v2/orders?status=open" $HEADERS

# Get specific order
curl -s "$BASE_URL/v2/orders/{order_id}" $HEADERS

# Cancel specific order
curl -s -X DELETE "$BASE_URL/v2/orders/{order_id}" $HEADERS

# Cancel ALL open orders
curl -s -X DELETE "$BASE_URL/v2/orders" $HEADERS

平仓

# Close entire position in a symbol
curl -s -X DELETE "$BASE_URL/v2/positions/AAPL" $HEADERS

# Partially close (sell 5 of 10 shares)
curl -s -X DELETE "$BASE_URL/v2/positions/AAPL?qty=5" $HEADERS

# EMERGENCY: Close ALL positions
curl -s -X DELETE "$BASE_URL/v2/positions" $HEADERS

市场数据（Alpaca 账户免费提供）

# Latest quote (bid/ask)
curl -s "$DATA_URL/v2/stocks/AAPL/quotes/latest" $HEADERS

# Latest trade (last fill)
curl -s "$DATA_URL/v2/stocks/AAPL/trades/latest" $HEADERS

# Historical bars (OHLCV) — daily
curl -s "$DATA_URL/v2/stocks/AAPL/bars?timeframe=1Day&start=2024-01-01&limit=100" $HEADERS

# Intraday bars — 5-minute
curl -s "$DATA_URL/v2/stocks/AAPL/bars?timeframe=5Min&start=$(date -d 'today' +%Y-%m-%d)&limit=78" $HEADERS

# Multi-symbol snapshot
curl -s "$DATA_URL/v2/stocks/snapshots?symbols=AAPL,MSFT,GOOGL" $HEADERS

# Crypto bars
curl -s "$DATA_URL/v1beta3/crypto/us/bars?symbols=BTC/USD&timeframe=1Day&limit=30" $HEADERS

# Crypto latest quote
curl -s "$DATA_URL/v1beta3/crypto/us/latest/quotes?symbols=BTC/USD,ETH/USD" $HEADERS

市场时钟与日历

# Is market open right now?
curl -s "$BASE_URL/v2/clock" $HEADERS
# Returns: timestamp, is_open (bool), next_open, next_close

# Upcoming market calendar
curl -s "$BASE_URL/v2/calendar?start=$(date +%Y-%m-%d)&end=$(date -d '+7 days' +%Y-%m-%d)" $HEADERS

加密货币交易说明

• 交易符号使用斜杠格式：BTC/USD、ETH/USD、SOL/USD、DOGE/USD
• 全天候交易（不受市场交易时间限制）
• 允许使用零碎数量（例如，BTC 可用 "qty": "0.001"）
• 模拟交易与实盘交易的运作方式相同
• 美元计价的加密货币订单使用 notional："notional": "100.00" 可买入价值 $100 的加密货币

账户活动与历史记录

# Trade history
curl -s "$BASE_URL/v2/account/activities/FILL?after=2024-01-01" $HEADERS

# Portfolio history
curl -s "$BASE_URL/v2/account/portfolio/history?period=1M&timeframe=1D" $HEADERS
# Returns: timestamp[], equity[], profit_loss[], profit_loss_pct[]

---

5. 免费金融数据源

价格数据（通过 web_search + web_fetch 获取）

来源	URL 模式	可用数据
Yahoo Finance	finance.yahoo.com/quote/AAPL	实时报价、图表、财务数据、分析师评级
Google Finance	google.com/finance/quote/AAPL:NASDAQ	报价、新闻、相关股票、财报
CoinGecko	coingecko.com/en/coins/bitcoin	加密货币价格、市值、成交量、24 小时涨跌幅
CoinMarketCap	coinmarketcap.com/currencies/bitcoin/	加密货币价格、排名、市场主导率、供应量
MarketWatch	marketwatch.com/investing/stock/AAPL	报价、新闻、分析、期权数据
Finviz	finviz.com/quote.ashx?t=AAPL	技术面与基本面筛选器、图表
TradingView	tradingview.com/symbols/NASDAQ-AAPL/	图表、技术指标、社区观点

基本面数据

来源	URL 模式	可获取的数据
Macrotrends	macrotrends.net/stocks/charts/AAPL/apple/pe-ratio	市盈率、营收、利润率、历史数据
Simply Wall St	Web 搜索："AAPL simply wall st"	可视化基本面分析、公允价值
SEC EDGAR	sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=AAPL&type=10-K	官方 10-K、10-Q、8-K 文件
Earnings Whispers	earningswhispers.com/stocks/AAPL	盈利预期、超预期历史、日历
Stock Analysis	stockanalysis.com/stocks/AAPL/financials/	清晰的财务报表、比率
Wisesheets	Web 搜索："AAPL income statement"	电子表格格式的财务数据

情绪与另类数据

来源	URL	可获取的数据
CNN 恐惧与贪婪指数	money.cnn.com/data/fear-and-greed/	市场情绪指数 0-100（极度恐惧至极度贪婪）
CBOE VIX	Web 搜索："VIX index today"	波动率指数（>30 = 恐惧，<15 = 自满）
Finviz 热力图	finviz.com/map.ashx	按行业／规模划分的市场热力图
StockTwits	stocktwits.com/symbol/AAPL	社交情绪（看涨／看跌比例）
看跌／看涨期权比率	Web 搜索："CBOE put call ratio today"	期权情绪（>1.0 = 看跌，<0.7 = 看涨）
空头持仓	finviz.com/quote.ashx?t=AAPL -> Short Float	被卖空的流通股比例
内部人交易	openinsider.com/screener	CEO／CFO 买入／卖出模式

宏观经济数据

来源	URL	可获取的数据
FRED	fred.stlouisfed.org	利率、CPI、就业、GDP、M2、收益率曲线
Treasury.gov	treasury.gov/resource-center/data-chart-center/interest-rates/	每日美国国债收益率曲线
CME FedWatch	Web 搜索："CME FedWatch tool"	联邦基金利率概率
BLS	bls.gov/news.release/	就业状况、CPI、PPI
ISM	Web 搜索："ISM manufacturing PMI"	PMI（>50 = 扩张，<50 = 收缩）
Conference Board	Web 搜索："consumer confidence index"	消费者信心、领先指标
盈利日历	earningswhispers.com/calendar	即将到来的财报日期
经济日历	Web 搜索："economic calendar this week"	计划发布的数据

加密货币专属数据源

来源	URL	可获取的数据
CoinGecko	coingecko.com	价格、市值、交易量、DeFi TVL
DefiLlama	defillama.com	所有链上的总锁仓价值
Glassnode（免费版）	Web 搜索："bitcoin on-chain metrics"	链上分析（NUPL、MVRV、交易所资金流）
Bitcoin 恐惧与贪婪指数	alternative.me/crypto/fear-and-greed-index/	加密货币专属情绪指数 0-100
Ultrasound Money	ultrasound.money	ETH 供应量／销毁指标

---

6. 信心校准指南（超级预测）

校准原则（Philip Tetlock）

• 一个“70% 确信”的预测，正确率应约为 70%
• 大多数人过度自信：他们“90%”确信的预测，正确率只有约 70%
• 系统地记录预测，并比较预测频率与实际频率
• 逐步更新（每条新证据调整 2-5%），不要大幅调整

信心等级指南

等级	含义	所需证据	交易行动
20-30%	略微倾向	单一弱信号，数据有限	不交易 — 优势不足
40-50%	胜负难料，略有优势	信号相互矛盾，证据中等	不交易 — 如同抛硬币
55-65%	中等把握	多个信号一致，有历史先例	小仓位，止损范围放宽
70-80%	把握较大	多因素强烈一致，已发现催化因素	标准仓位规模
85-95%	把握极大	证据压倒性 — 应对自己保持怀疑	满仓，但绝不要全仓押注

交易预测的 Brier 分数

Brier Score = mean((predicted_probability - actual_outcome)^2)
actual_outcome: 1 if prediction was correct, 0 if wrong

Worked example (5 predictions):
  Pred 1: 80% confident -> correct (1)  -> (0.80 - 1)^2 = 0.04
  Pred 2: 60% confident -> wrong (0)    -> (0.60 - 0)^2 = 0.36
  Pred 3: 70% confident -> correct (1)  -> (0.70 - 1)^2 = 0.09
  Pred 4: 90% confident -> correct (1)  -> (0.90 - 1)^2 = 0.01
  Pred 5: 55% confident -> wrong (0)    -> (0.55 - 0)^2 = 0.30
  Brier Score = (0.04 + 0.36 + 0.09 + 0.01 + 0.30) / 5 = 0.16

  Ratings: 0.00 = perfect, < 0.15 = excellent, 0.15-0.25 = good,
           0.25 = coin flip, > 0.25 = worse than random

校准自检协议

累计 20 笔以上交易预测后，按置信度区间分组：

1. 你的 60% 置信度预测有约 60% 的时间是正确的吗？
2. 如果你的 60% 置信度预测有 80% 的时间是正确的，说明你的信心不足——应调高
3. 如果你的 80% 置信度预测有 55% 的时间是正确的，说明你过于自信——应调低
4. 每解决 50 笔预测后，重新校准你的置信度尺度

---

7. 交易心理与认知偏差

需警惕的偏差

偏差	描述	缓解方法
确认偏差	寻找能够证实自己论点的信息	始终先构建反方论据（对抗式辩论）
锚定效应	过分看重最先看到的数字（入场价、分析师目标价）	从基准率和当前数据开始分析，而不是旧价格
近因偏差	过分看重近期事件（上周的暴跌、上个月的上涨）	查看更长的时间范围——至少查看 6 个月和 1 年的图表
损失厌恶	亏损时持仓太久（“它会涨回来的”），盈利时过早卖出	使用机械式止损和止盈目标，并在入场前设定
过度自信	认为自己比实际情况更正确	跟踪 Brier 分数，使用 Kelly 比例，单笔交易下注绝不超过 2%
叙事偏差	动人的故事 = 好交易（往往并非如此）	关注量化数据，而不是故事。“好公司” != “好交易”
错失恐惧症（FOMO）	害怕错失机会，追涨买入	只在计划好的价位入场。市场每年开放 252 天
沉没成本效应	“我已经亏了这么多，现在不能卖”	每一刻都是新的决策。问自己：“如果今天以当前价格买入，我还会进行这笔交易吗？”
后见之明偏差	“我早就知道会这样”	在交易前记录具体预测，而不是事后记录
处置效应	为了“锁定利润”而过早卖出赢家，却继续持有亏损仓位	让赢家继续上涨（移动止损），在计划的止损位砍掉亏损仓位
赌徒谬误	“已经连续跌了 5 天，它肯定会反弹”	每一天都是独立的。趋势持续的情况比反转更常见
禀赋效应	高估自己已经持有的仓位	仿佛今天从零开始建仓一样评估持仓

纪律规则

1. 每笔交易入场前都要有书面计划：入场价、止损价、目标价、仓位大小、论点
2. 入场前写下你的理由——如果无法清楚阐述优势，就不要交易
3. 在下单时设定止损，而不是“记在脑子里”
4. 每周回顾交易日志——寻找盈利和亏损中的规律
5. 大幅盈利后（过度自信风险）和大幅亏损后（情绪风险）都要休息
6. 除非原始论点明确计划如此，否则绝不对亏损仓位向下摊平
7. 绝不把止损价设得比入场价更远（只能收紧，不能放宽）
8. 市场明天还会开盘——错过一笔交易不算亏损，但账户爆仓算

---

8. 投资组合构建

资产配置指南

风格	股票	加密货币	固定收益 / 现金	单一持仓上限
保守型	50-60%	0-5%	35-50%	5%
稳健型	60-75%	5-15%	10-35%	8%
激进型	70-85%	10-25%	5-20%	10%
投机型	50-70%	20-40%	5-10%	15%（必须严格止损）

行业分散

任何单一行业的占比最高为 30%：

• 科技、医疗保健、金融、可选消费、必需消费
• 能源、工业、公用事业、房地产、材料、通信服务

相关性认知

高度相关的持仓会放大风险。新增持仓前，请检查相关性：

配对	典型相关性	风险
AAPL + MSFT + GOOGL	0.7-0.9	大盘科技股过度集中
BTC + ETH + SOL	0.8-0.95	加密资产过度集中（走势趋同）
SPY + QQQ	0.9+	敞口几乎相同
股票 + 债券	-0.2 到 0.3	确实能分散风险
黄金 + 股票	-0.1 到 0.2	危机期间可对冲
VIX + SPY	-0.8	反向关系——以 VIX 对冲

再平衡规则

• 按日历：每季度再平衡一次（季度首个交易日）
• 阈值：任一资产配置偏离目标 > 5% 时再平衡
• 税务考量：优先通过新增投入而非卖出进行再平衡（应税账户）

---

9. 跨平台命令

Windows（PowerShell / Git Bash）

# Python might be `python` not `python3` on Windows
python -c "import json; ..."

# Use forward slashes in file paths or escape backslashes
# curl is available via Git Bash, PowerShell, or WSL

# Check if market is open (Windows Git Bash)
curl -s "$BASE_URL/v2/clock" -H "APCA-API-KEY-ID: $ALPACA_API_KEY" \
  -H "APCA-API-SECRET-KEY: $ALPACA_SECRET_KEY" | python -c "
import sys, json
d = json.load(sys.stdin)
print('OPEN' if d['is_open'] else 'CLOSED', '| Next:', d.get('next_open','') or d.get('next_close',''))
"

macOS / Linux

python3 -c "import json; ..."
# curl, jq typically available by default
# Use jq for JSON processing:
curl -s URL | jq '.equity'

不使用 jq 处理 JSON

# Pretty-print JSON
python3 -c "import sys,json; print(json.dumps(json.load(sys.stdin),indent=2))" < file.json

# Extract specific field
curl -s URL | python3 -c "import sys,json; d=json.load(sys.stdin); print(d['equity'])"

# Parse Alpaca positions into readable table
curl -s "$BASE_URL/v2/positions" $HEADERS | python3 -c "
import sys, json
positions = json.load(sys.stdin)
fmt = '{:<8} {:>6} {:>10} {:>10} {:>12} {:>8}'
print(fmt.format('Symbol','Qty','Entry','Current','P/L','P/L pct'))
print('-' * 60)
for p in positions:
    print(fmt.format(p['symbol'], p['qty'], float(p['avg_entry_price']),
          float(p['current_price']), float(p['unrealized_pl']),
          round(float(p['unrealized_plpc'])*100,2)))
"

# Calculate RSI from historical bars
curl -s "$DATA_URL/v2/stocks/AAPL/bars?timeframe=1Day&limit=30" $HEADERS | python3 -c "
import sys, json
data = json.load(sys.stdin)
closes = [float(b['c']) for b in data['bars']]
changes = [closes[i]-closes[i-1] for i in range(1, len(closes))]
gains = [max(c,0) for c in changes[-14:]]
losses = [abs(min(c,0)) for c in changes[-14:]]
avg_gain = sum(gains)/14
avg_loss = sum(losses)/14
rs = avg_gain/avg_loss if avg_loss > 0 else 999
rsi = 100 - (100/(1+rs))
print(f'RSI(14) = {rsi:.1f}')
"

---

10. 交易前检查清单

每次交易前，请确认以下所有事项：

PRE-TRADE CHECKLIST
====================
[ ] 1. TREND: What is the higher-timeframe trend? (Daily chart 200 SMA)
       - Trading WITH the trend? (preferred)
       - Counter-trend? (requires stronger signal + tighter stops)

[ ] 2. SIGNAL: What specific setup triggered this trade?
       - Indicator signal (RSI, MACD, etc.)
       - Pattern (candlestick, chart pattern)
       - Catalyst (earnings, news, sector rotation)

[ ] 3. ENTRY: Exact entry price or condition
       - Limit order at specific level? Market order on breakout?

[ ] 4. STOP LOSS: Exact stop price
       - Based on ATR (2-3x ATR from entry)
       - Below key support (long) or above key resistance (short)
       - NEVER wider than 2% of portfolio

[ ] 5. TARGET: Exact take-profit price
       - Risk/Reward at least 1.5:1 (preferably 2:1+)
       - At logical resistance (long) or support (short)

[ ] 6. POSITION SIZE: Calculated from risk management rules
       - Risk amount = Portfolio * 1-2%
       - Shares = Risk amount / (Entry - Stop)
       - Total position < 10% of portfolio

[ ] 7. CORRELATION CHECK: Does this overlap with existing positions?
       - Not adding to concentrated sector exposure
       - Total portfolio heat (sum of open risk) < 6%

[ ] 8. CATALYST CHECK: Any upcoming events that could gap through stops?
       - Earnings date? Fed meeting? CPI release?
       - If yes: reduce size or wait until after event

[ ] 9. MARKET CONTEXT: Is the overall market favorable?
       - Fear & Greed index level
       - VIX level (>30 = caution, <15 = complacency risk)
       - Market trend (SPY vs 200 SMA)

[ ] 10. CONFIDENCE: Rate 1-10 honestly
        - Below 6? Skip the trade
        - Record confidence for calibration tracking

---

11. 交易日志模板

{
  "trade_id": "T001",
  "date_opened": "2025-01-15",
  "date_closed": null,
  "symbol": "AAPL",
  "side": "long",
  "entry_price": 150.00,
  "stop_loss": 145.00,
  "target": 162.00,
  "position_size": 40,
  "risk_amount": 200.00,
  "risk_reward": 2.4,
  "setup": "Bullish engulfing at 50 EMA + RSI divergence",
  "confidence": 7,
  "market_context": "SPY above 200 SMA, VIX at 18, F&G neutral (52)",
  "pre_trade_thesis": "AAPL pulled back to 50 EMA support, RSI showing bullish divergence, earnings in 3 weeks should provide catalyst. Sector (tech) is leading.",
  "result": {
    "exit_price": null,
    "exit_reason": null,
    "pnl": null,
    "pnl_percent": null,
    "held_days": null,
    "lessons": null
  }
}

使用 memory_store 存储交易日志，以便跟踪和进行校准审查。