Professional Documents
Culture Documents
Management
Sahand Haji Ali Ahmad
Alpha Signals
• These are the signals that predict the winners/losers ( have predictive
power over the returns and as such can predict the winners/losers)
• Could have predictive power over returns over different periods
• Some sort of daily data most likely has predictive power over very
short term windows, while a quarterly signal most likely helps explain
the quarterly returns
Well-known Signals
Accounting Factors (sourced from Compustat, Worldscope, Factset,…):
• Value ( Price/Earning, Book/Price,..)
• Growth( Revenue Growth rate, Earnings growth rate,…)
• Size
• Quality(Balance sheet quality, Debt to Equity,….)
• Profitability( Dividend yield, Earnings/Revenue,….)
• Dividend increase/decrease
• Liquidity
• MOAT (Morningstar-rated) equities
Technical Factors:
• Low-Volatility Anomaly
• Momentum ( Price momentum, Industry momentum,…)
• 20-day mean-reversion factor(Jegadeesh,…)
Example, Quality Factor
• A lot of divergence over the definition of a Quality factor
• Could even include Piotrski’s F-score including 9 factors
• Quality: Higher revenue and cash generation, more stable growth,
conservative yet effective capital structure, run by prudent managers
• 3 main categories:
• Profitability Generation: unique offering, Competitive position,…
• Earnings Quality: Earnings persistence, Accounting red flags, …
• Financial Robustness: effectiveness of capital structure, ability to
remain solvent, capacity to sustain growth
Quality
• Profitability Generation: Return on Equity, Return on Assets, Gross
Profit/Assets
• Earnings Quality: Accruals Ratio, change in Cash flow from
Operations& Net income
• Financial Robustness: Operating leverage, Financial leverage, Current
ratio(Current assets/Current liabilities)
• Selected: Return on Equity, Balance Sheet Accruals Ratio, Financial
Leverage
Quality
• Return on Equity: 12-month income scaled by the company book value
• Persistence of Return on Equity
7.0
3.5
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
Time
Transient Impact Temporary Impact Permanent Impact
( ) ( )
Total impact over the trade period
2
METRIC = S + v T + v T
• Where S is the average bid-ask spread, is the volatility, v is the trade rate and T is the
trade duration
• Decomposes the cost into: Instantaneous Impact, Transient Impact, Permanent Impact
14
Machine Learning in Regime Shift Detection
• There are many ways to detect regime shift, such as Rodionov Regime
Switch,…
• One particularly popular way is to use Hidden Markov Models
together with Filtering of asset price time-series to create a Markov
Switching Regime Shift model
• Smoothening the time-series, and splitting it into 2-4 regimes ( bull,
bear, and possibly transition states)
Using Clustering for regime detection
Using Trend Filtering and Markov Switch
model
Regime-Switching multi-factor model for
asset returns
• Using a regime switching multi-factor model for macro asset prices,
and using BIC criterion to minimize, 4 regimes for macro assets are
detected.
• Equilibrium transition probabilities are calculated. These 4 states,
represent whether it is stable bull, stable bear, transition bull to bear
and transition bear to bull and can be very useful in deciding:
1) what should be the proportions of cash and equity
2) regarding equity portfolio, where should the weight be tilted
to(Value, Growth or momentum)
Some of Yin Luo stuff
Event Driven Trading
• Activist Investing(13D)
• CEO/CFO Turnover
• Dividend Policy Changes
• Complicated Firms
Activist Investing
Activist Investing stats
• Average monthly excess returns ranging from 88bps to 181bps after
adjusting for market, value, size, and price momentum.
Activist Investing
• One-month after the commencement of activism, a backtested portfolio of
targets yielded a market-adjusted return of 3.9%. After controlling for market,
size, value, and industry, the excess return was 3.0%.
• Twelve-months after the disclosure of activist involvement, a backtested portfolio
of targets produced an annualized excess return of 11.7% after controlling for
market, size, value, and momentum.
• Research finds no evidence that by utilizing Form 13F filings and taking positions
in pre-disclosed targets prior to the commencement of activism, increased excess
returns.
• Pre-activism, the targets tended to be (i) small-capitalization (ii) slow-growing (iii)
low payout (iv) financially healthy.
• Post-activism, the targets were characterized by having (i) higher financial
leverage (ii) lower cash holding (iii) higher payout (iv) no improvement to their
financial operations.
CEO Departure
CEO Turnover
Dividend Policy Change
Dividend Policy Change
Complicated Firms
• They lag the big changes in their representative industries
• Strategy: Identify companies where their underlying industries are
experiencing the strongest performance over the past month
• FASB Statement 14 requires that public companies provide information
regarding their reportable segments including the segments industry and
revenue
• Simple Firms - firms that derive greater than 80% of their total sales from
one industry segment
• Complicated Firms - firms where there largest industry segment accounts
for less than 80% of their total sales
• We weight simple industry returns proportionate to the company’s sales
in each industry in an attempt to identify where there have been
information shocks over the past month
Conglomerate Strategy Performance
Equities Classification and Statistical Arbitrage
• GICS
• AHL paper
• publications
Extract all I can from Talks(Quantopian,…)