Upbit MCP Server

from fastmcp import Context import httpx import numpy as np from typing import Literal from config import API_BASE async def technical_analysis( market: str, interval: Literal["minute30", "minute60", "minute240", "day"], ctx: Context = None ) -> dict: """ 특정 마켓에 대한 기본적인 기술적 분석을 수행합니다. Args: market (str): 마켓 코드 (예: KRW-BTC) interval (str): 시간 간격 (minute30, minute60, minute240, day) Returns: dict: 기술적 분석 결과 """ if ctx: ctx.info(f"{market} {interval} 기술적 분석 수행 중...") # 캔들 데이터 조회 if interval.startswith("minute"): url = f"{API_BASE}/candles/{interval}" else: url = f"{API_BASE}/candles/{interval}s" params = { 'market': market, 'count': '100' # 100개 캔들 사용 } async with httpx.AsyncClient() as client: try: res = await client.get(url, params=params) if res.status_code != 200: if ctx: ctx.error(f"캔들 데이터 조회 실패: {res.status_code}") return {"error": f"캔들 데이터 조회 실패: {res.status_code}"} candles = res.json() # 종가, 고가, 저가 추출 closes = np.array([float(candle["trade_price"]) for candle in candles]) highs = np.array([float(candle["high_price"]) for candle in candles]) lows = np.array([float(candle["low_price"]) for candle in candles]) volumes = np.array([float(candle["candle_acc_trade_volume"]) for candle in candles]) # 배열 순서 뒤집기 (최신 데이터가 마지막에 오도록) closes = np.flip(closes) highs = np.flip(highs) lows = np.flip(lows) volumes = np.flip(volumes) # 단순 이동 평균선 (SMA) sma5 = np.mean(closes[-5:]) if len(closes) >= 5 else None sma10 = np.mean(closes[-10:]) if len(closes) >= 10 else None sma20 = np.mean(closes[-20:]) if len(closes) >= 20 else None sma50 = np.mean(closes[-50:]) if len(closes) >= 50 else None # RSI 계산 (14일) if len(closes) >= 15: delta = np.diff(closes) gain = np.where(delta > 0, delta, 0) loss = np.where(delta < 0, -delta, 0) # 첫 번째 평균 게인/로스 계산 avg_gain = np.mean(gain[:14]) avg_loss = np.mean(loss[:14]) # 나머지 값 계산 for i in range(14, len(delta)): avg_gain = (avg_gain * 13 + gain[i]) / 14 avg_loss = (avg_loss * 13 + loss[i]) / 14 if avg_loss == 0: rsi = 100 else: rs = avg_gain / avg_loss rsi = 100 - (100 / (1 + rs)) else: rsi = None # 볼린저 밴드 (20일) if len(closes) >= 20: middle_band = np.mean(closes[-20:]) std_dev = np.std(closes[-20:]) upper_band = middle_band + (2 * std_dev) lower_band = middle_band - (2 * std_dev) else: middle_band = upper_band = lower_band = None # MACD (12, 26, 9) if len(closes) >= 26: ema12 = np.mean(closes[-12:]) # 간단한 구현을 위해 SMA 사용 ema26 = np.mean(closes[-26:]) macd_line = ema12 - ema26 if len(closes) >= 35: # MACD 신호선을 위한 9일 추가 signal_line = np.mean([ema12 - ema26 for ema12, ema26 in zip( [np.mean(closes[i-12:i]) for i in range(len(closes)-8, len(closes)+1)], [np.mean(closes[i-26:i]) for i in range(len(closes)-8, len(closes)+1)] )]) macd_histogram = macd_line - signal_line else: signal_line = macd_histogram = None else: macd_line = signal_line = macd_histogram = None # 스토캐스틱 (14일) if len(closes) >= 14: low_14 = np.min(lows[-14:]) high_14 = np.max(highs[-14:]) if high_14 - low_14 > 0: k_percent = ((closes[-1] - low_14) / (high_14 - low_14)) * 100 else: k_percent = 50 # 가격 범위가 없는 경우 # 간단한 구현을 위해 %D를 3일 평균으로 계산 if len(closes) >= 16: d_percent = np.mean([ ((closes[-3] - np.min(lows[-16:-2])) / (np.max(highs[-16:-2]) - np.min(lows[-16:-2]))) * 100, ((closes[-2] - np.min(lows[-15:-1])) / (np.max(highs[-15:-1]) - np.min(lows[-15:-1]))) * 100, k_percent ]) else: d_percent = None else: k_percent = d_percent = None # 거래량 분석 avg_volume = np.mean(volumes) current_volume = volumes[-1] volume_ratio = current_volume / avg_volume if avg_volume > 0 else 0 # 지지/저항 레벨 (단순화된 방식) pivots = {} if len(closes) >= 20: pivot_point = (highs[-1] + lows[-1] + closes[-1]) / 3 r1 = 2 * pivot_point - lows[-1] r2 = pivot_point + (highs[-1] - lows[-1]) s1 = 2 * pivot_point - highs[-1] s2 = pivot_point - (highs[-1] - lows[-1]) pivots = { "pivot": pivot_point, "r1": r1, "r2": r2, "s1": s1, "s2": s2 } # 분석 결과 요약 analysis_result = {} current_price = closes[-1] # 이동평균선 신호 if sma5 and sma20: if sma5 > sma20: analysis_result["ma_signal"] = "상승 추세 (황금 교차)" elif sma5 < sma20: analysis_result["ma_signal"] = "하락 추세 (죽음의 교차)" else: analysis_result["ma_signal"] = "중립" # RSI 신호 if rsi is not None: if rsi > 70: analysis_result["rsi_signal"] = "과매수" elif rsi < 30: analysis_result["rsi_signal"] = "과매도" else: analysis_result["rsi_signal"] = "중립" # 볼린저 밴드 신호 if upper_band and lower_band: if current_price > upper_band: analysis_result["bb_signal"] = "과매수 (상단 돌파)" elif current_price < lower_band: analysis_result["bb_signal"] = "과매도 (하단 돌파)" else: analysis_result["bb_signal"] = "중립 (밴드 내)" # MACD 신호 if macd_line is not None and signal_line is not None: if macd_line > signal_line: analysis_result["macd_signal"] = "매수 신호" elif macd_line < signal_line: analysis_result["macd_signal"] = "매도 신호" else: analysis_result["macd_signal"] = "중립" # 스토캐스틱 신호 if k_percent is not None and d_percent is not None: if k_percent > 80 and d_percent > 80: analysis_result["stoch_signal"] = "과매수" elif k_percent < 20 and d_percent < 20: analysis_result["stoch_signal"] = "과매도" elif k_percent > d_percent: analysis_result["stoch_signal"] = "상승 중" elif k_percent < d_percent: analysis_result["stoch_signal"] = "하락 중" else: analysis_result["stoch_signal"] = "중립" # 종합 신호 signals_count = len(analysis_result) buy_signals = sum(1 for signal in analysis_result.values() if "매수" in signal or "상승" in signal) sell_signals = sum(1 for signal in analysis_result.values() if "매도" in signal or "하락" in signal) oversold_signals = sum(1 for signal in analysis_result.values() if "과매도" in signal) overbought_signals = sum(1 for signal in analysis_result.values() if "과매수" in signal) if signals_count > 0: if buy_signals / signals_count > 0.6 or oversold_signals >= 2: analysis_result["overall_signal"] = "매수 고려" elif sell_signals / signals_count > 0.6 or overbought_signals >= 2: analysis_result["overall_signal"] = "매도 고려" else: analysis_result["overall_signal"] = "중립 관망" return { "market": market, "interval": interval, "current_price": current_price, "indicators": { "sma": { "sma5": sma5, "sma10": sma10, "sma20": sma20, "sma50": sma50 }, "rsi": rsi, "bollinger_bands": { "upper": upper_band, "middle": middle_band, "lower": lower_band }, "macd": { "line": macd_line, "signal": signal_line, "histogram": macd_histogram }, "stochastic": { "k": k_percent, "d": d_percent }, "volume": { "current": current_volume, "average": avg_volume, "ratio": volume_ratio }, "pivots": pivots }, "analysis": analysis_result } except Exception as e: if ctx: ctx.error(f"기술적 분석 수행 중 오류 발생: {str(e)}") return {"error": f"기술적 분석 수행 중 오류 발생: {str(e)}"}