Summary

This project is to create a self use stock selection and portfolio optimization strategy. In general, investment is the combination of art and science. Traditionally, people believe that an efficient portfolio should be exposed to as diversified instruments as possible. For example, from mean-variance perspective, variance of stocks and bonds are negatively correlated so they should be included with almost the same weight to diversify risks. However, regardless of effectiveness of mean-variance portfolio construction, most individual investors are not risk averse but loss averse. And the loss is not short term but mid/long term loss. For that reason, instrument selection is crucial to the performance in a relatively long time run. I agree with the idea that value investment should be embraced by most investors since technical analysis is time consuming. You may not want to quit your job and be a day trader who doesn’t create concrete value for the society.

Currently, there are thousands of instruments over the market, equities, fixed income, various derivatives, etfs, mutual funds, etc. My belief is, simple is the best. If you don’t feel the necessity to add that instrument to your portfolio, don’t add it. I summarized a few lessons I learned from the market: 1.

Instrument Selection

There are a few guidelines I followed to add instruments to my portfolio:

  1. You should be familiar with that company. For example, I go to Starbucks frequently, I use Nvidia’s GPU in my home machine, I travel with Southwest, etc. All of companies create great product. In mid-long term, their growth is proposing.

  2. The current market is mainly supported by the Federal Reserve and municipal policies. The value of all stocks are generally over-prices. To mimic the market,

  3. Information will be digested as soon as possible and that’s the reason why you may find counterintuitive phenomenon in the market

  • Technology
    • U.S.
      1. Amazon, Apple
      2. Uber
    • China
      1. Bilibili, Alibaba
      2. Gensheixue(Online learning)
  • BioTech
    1. Gilead Science, Crisper
  • Defense
    1. Lockheed Martin

In most of the periods between 2016-2020, stocks are highly

This code chunk was implemented to calculate risk parity portfolios. In most cases, when the entire market is gradually up, there is no need for individual investors to hold the cash or cash equivalent, if you care more about longer term profit rather than next month total portfolio net value. For that reason, investment target is to maximize midterm to long term total return. Based on this hypothesis, leveraged ETFs with risk parity

Portfolio Optimization

I do think portfolio optimization is useful. But we cannot simply give the framework the complete set of stocks which contains all stocks over the market. By intuitions, we may infer some stocks will outperform others in short/midterm length. For example, I can assure with high possibility that Amazon will outperform underarmor in the next one year. This strategy helps me to construct my portfolio.

But there are two rules which cannot be broken: 1. Diversify portfolio as much as possible 2. Never short the market

I used a Python package to conduct portfolio optimization

The mathematical formula for this optimization task is stated in below.

## Loading stock data from local file C:/Users/guanhua.huang/github/victorhuangkk.github.io/_scripts/stockdata_from_2015-05-20_to_2020-05-20_(64c75c1355d1ccd2ca3f2c5b4ec6e61c).RData

Backtest

To verify the risk

# check performance summary
backtestSummary(bt)$performance

##                   risk parity portfolio tangency portfolio
## Sharpe ratio                  0.8258296          0.7168205
## max drawdown                  0.3736189          0.4018382
## annual return                 0.2031849          0.1911637
## annual volatility             0.2460373          0.2666828
## Sterling ratio                0.5438292          0.4757231
## Omega ratio                   1.2044897          1.1786037
## ROT (bps)                  4578.6200190        656.9454952

# plot cumulative returns chart
backtestChartCumReturns(bt)


# plot assets exposures over time
backtestChartStackedBar(bt, portfolio = "risk parity portfolio", legend = TRUE)

Conclusion

Based on this short report, we can summarize that risk parity strategy do work in some scenarios. And depends on how risky we want to develop the portlfio,