Projects in Quant Finance

1. Stock price prediction

Method: LSTM (Long Short-Term Memory)
Github link: click here

Description: For my bachelor’s in Data Science & AI, I have had to incorporate many different AI models. I decided to use the best model on stock price time series data. Specifically, on the data of all the stocks I own in my personal portfolio. See an example of google’s stock.

2. Sharpe ratio optimized portfolio

Method: Factor Model Mean-Variance Optimization
Github link: click here

Description: This project reaches a little further outside of my expertise. I learned a lot. I took the current portfolio of stocks owned by Berkshire Hathaway (originally 15+ stocks), and optimized the cash allocation to maximize for Sharpe ratio.

Quant/Finance/Investing Books that I've read.

And my favorite quotes / formulas

"While the market is not always efficient, it is frequently efficient. Beating the market is difficult."

Wesley Gray

PhD Finance

Final formula:
P_FS = Sum(FS_ROA, FS_FCFTA, FS_ACCRUAL, FS_LEVER, FS_LIQUID, FS_NEQISS, FS_ROA, FS_FCFTA, FS_MARGIN, FS_TURN))/10

P_FP = Franchise Power • Percentile (average (P_8yr_ROA, P_8yr_ROC, P_CFOA, MM) among all firms in the universe

QUALITY = .5 × P_FP + .5 × P_FS

See my summary of their checklist here.

"In applying analysis to the field of securities we encounter the serious obstacle that investment is by nature not an exact science"

Benjamin Graham

"Father of value investing"

"A large number of valuation proxies are computed from fundamental data. These factor can be combined as inputs into a machine learning valuation model to predict asset prices."

Stefan Jansen

Packt Publishing

"Be fearful when others are greedy, and greedy when others are fearful."

Charlie Munger

Berkshire Hathaway

"There is no such thing as inherent value for a business; it always depends on who's operating it."

McKinsey

& Company

"In my experience, it has been easy to spot a bubble after it is well under way... Making a profit is trickier. As Keynes said, the market can remain irrational longer than you can remain solvent."

Ed Thorpe

Mathematician (w.r.t. gambling and trading)

"There will come a time when even the most sophisticated deep neural network will make a fundamentally wrong prediction, just as the CDO models did in 2008" and my favorite part: model.compile(optimizer='adam',...)

Jannes Klaas

ML & Finance Expert