---
tags: finance
---

# Finding Alpha

- What is alpha? Outperform the market benchmark.
- Quality of alpha:
- The idea and expression are simple.
- The expression/code is elegant.
- It has a good in-sample Sharpe ratio.
- It is not sensitive to small changes in data or parameters.
- It works in multiple universes.
- It works in different regions.
- Example of alphas
- Statistical arbitrage
- You may lose some time but make money most of time.
- Those models from academia, even when they provide a foundation for applied research, are often incomplete or based on assumptions that are inconsistent with the real markets in which traders operate.
- An implication of the Efficient Market Hypothesis (EMH) is that prices will evolve in a process indistinguishable from a random walk.
- How to check if the market is not ideal like the textbook? A hypothesis is to test if the time series is a AR(1) process.
- Is ^twii satisfying EMH?
- Karl Popper (1934): it's an accepted scientific principle that no rule can really be proved; it can only be disproved.
- Trend-following strategy: 順勢交易，又稱右側交易。
- Cutting Losses: 止損
- Plan in advance how to get out of the strategy cheaply.
- Pursue multiple strategies simultaneously.
- Cut all strategies that fall outside expectations
- Alpha design
- Data sources
- Prices and volumes.
- Fundamentals.
- Macroeconomic data.
- Text: e.g FOMC
- Multimedia.
- Noise reduction
- Factor models.
- Relationship models (?)
- Alpha evaluation
- Information ratio
- Margin
- Correlation
- In-sample vs. out-of-sample and avoid overfitting
- Alpha logic (see chapter 5)


- About data source: AA single good data source can make a big difference in the performance of an alpha or a portfolio. Therefore, investment in data searching, data cleaning, and data processing is an important part of alpha creation.
- Information ratio: return rate / volatility, basically same formula as the Sharpe ratio.
- Information coefficient: the correlation between the predicted and actual values.
- Turnover: the longer the horizon, the greater the uncertainty.
- Bid-ask spread: 例如房價的 spread 很大，賣出自己目前持有的房子，幾乎無法用原價買回來。這個 spread 的主因是來自於流動性問題 liquidity (liquidity)?
- Multi-alpha: a portfolio manager generally needs to combine multiple diverse alphas, which result in a stronger, more informed prediction that is more likely to overcome transaction costs and other trading constraints.
- Information is changing all the time, but (1) not all types of information change at the same rate, and (2) not all information is particularly usefull.
- It is important to optimize the trade-off between the return and turnover.
- Alphas should be neutralized by industry.
- Turnover * IR = constant.
- An alpha with a lower correlation coefficient normally is considered to be adding more value to the pool of existing alphas.
- Pearson's linear correlation
- Temporal-based correlation (see p. 63).
- Generalized correlation (see p. 64).
- Alpha value correlation (see p.66):
- Position correlation.
- Trading correlation.
- Spearman’s rank correlation
- Kendall rank correlation
- How to control biases?
- Look-ahead bias: 前窺
- ... (continue chapter 10).
- [Bloomberg Billionaires Index](https://www.bloomberg.com/billionaires/)