- LY说澳洲教育强调尊重个人,我觉得这个很好,以后自己的思维也要有这种优良意识,尊重个人的自由意志.
- finance我感觉是个无底洞,想学的话需要找个fintech好点,自学的话需要大量simplify.
pru-0
paper - 看了一下finance尖端,发现真正的predicting不是curve fitting,
1. creative feature:
而是利用当时的economic indicator, 和对company的各种创意observant
a. coffee company - 可以看各国天气状态
b. service company - 可以看credit card 记录,或者各大网站sentiment
c. shipping company - 可以看deck的高度
这里nlu 和visual 都用上了
2. pair trading
利用statistical arbitrage 等用各种pair trading 来hedge 风险,这些pair 需要有cointegration, 就是想同类型(same industry, same size, 像credit card he master card)
3. new/exotic market
alpha is not well explored yet, 但是我的经验是大部分是骗人的。
4. expectation
公司的sales, lower than expectation, 会影响股市
另外 python 上
1. draw reverse chart
plt.plot(stockA['adj_close'])
plt.gca().invert_xaxis()
2. cointegration test
import statsmodels # More info: https://www.statsmodels.org/stable/tsa.html#tsa-tools
from statsmodels.tsa.stattools import coint
#TDL4: Run Cointegration Test on Your 2 stocks
score, pvalue, _ = coint(stockA['adj_close'],stockB['adj_close']) #TDL4
print(pvalue) # the lower the better. <0.05 means the 2 stocks are statistically cointegrated.
print(_)
3. 在做pair trading 前必须normalize
Normalise the data so that the first data point is 1 (Dollar hedging)
No comments:
Post a Comment