def kline_similarity(k, f, size):
c = [i[4] for i in k[:-1]]
c = [j / c[0] for j in c]
diff, dea, _macd = utils.macd(np.array(c))
r = []
for i in range(len(diff)):
a, b, c = 0, 0, 0
if i != len(diff) - 1:
continue
for j in range(size):
a += math.pow(f["diff"][j] - diff[i - size + 1 + j], 2)
b += math.pow(f["dea"][j] - dea[i - size + 1 + j], 2)
c += math.pow(f["macd"][j] - _macd[i - size + 1 + j], 2)
return math.sqrt(a + b + c)
def feature_benchmark(k, f, size):
p = config.benchmark
c = [i[4] for i in k[:-1]]
c = [j / c[0] for j in c]
diff, dea, _macd = utils.macd(np.array(c))
r = []
for i in range(len(diff)):
a, b, c = 0, 0, 0
if i < 100:
continue
for j in range(size):
a += math.pow(f["diff"][j] - diff[i - size + 1 + j], 2)
b += math.pow(f["dea"][j] - dea[i - size + 1 + j], 2)
c += math.pow(f["macd"][j] - _macd[i - size + 1 + j], 2)
r.append(math.sqrt(a + b + c))
r.sort()
return r[int(len(r) * (100 - p) / 100.0)]
def handle_data(context, data):
hour = context.current_dt.hour
minute = context.current_dt.minute
if hour == 14 and minute == 59:
# 测试移动平均线
MA5 = utils.ma(g.security, 5)
MA10 = utils.ma(g.security, 10)
MA20 = utils.ma(g.security, 20)
log.info("%s MA测试 ==> MA5=%s, MA10=%s, MA20=%s", g.security, f2(MA5), f2(MA10), f2(MA20))
# 测试是否多头排列
is_bull_market = utils.is_bull_market(g.security)
log.info('%s 是否多头排列测试 ==> %s', g.security, is_bull_market)
# 测试macd
diff,dea,macd = utils.macd(g.security)
log.info("%s MACD测试 ==> DIFF=%s, DEA=%s, MACD=%s", g.security, f2(diff[-1]), f2(dea[-1]), f2(macd[-1]))
# 获取股票的位置,默认为近两年
recent_gains, recent_decline = utils.stock_position(g.security)
log.info("==> %s股票的位置：近期涨幅=%.2f，近期跌幅=%.2f", g.security, recent_gains, recent_decline)
def feature(k, point, feature_size):
c = [i[4] for i in k[:-1]]
c = [j / c[0] for j in c]
diff, dea, _macd = utils.macd(np.array(c))
f = {}
for ki in range(len(c)):
if k[ki][0] == point:
break
f["diff"] = list(diff[ki - feature_size:ki])
f["dea"] = list(dea[ki - feature_size:ki])
f["macd"] = list(_macd[ki - feature_size:ki])
f["v"] = feature_benchmark(k, f, feature_size)
#print f["v"]
#print "".join(hashlib.md5(json.dumps(f)).hexdigest())
#print point
#print "".join(hashlib.md5(json.dumps(k[:-1])).hexdigest())
#print k[:-1]
#print point
#print ki - feature_size, ki
return f