Inefficient Markets Flashcards
(18 cards)
Requirements for efficiency
All investors are always rational.
Investors’ errors are uncorrelated.
There cannot be arbitrage.
Trading on Noise
Can this generate a profit?
People may trend on ‘noise’ as if it were correct information - they are wrong to expect a profit from this.
Why is noise trading done?
Noise is misinformation.
Noise trading is generally done for reasons other than information and pricing, e.g., business reasons.
Advantages of noise trading
Noise trading is not a bad thing because it provides liquidity and offers counterparts for informed traders.
* People don’t want to trade with those more informed than them.
* Noise traders will trade with informed traders.
Is noise trading good or bad for market efficiency?
Introduces variability, can hide the true value and produces disturbances in the market.
But also provides liquidity and offers the opportunity for nore informed traders to make the market more efficient.
Fundamental uncorrelated IID noise
The noise is tied to genuine information relevant to asset valuation, but it still introduces randomness in how prices respond because fundamentals are not perfectly observable or measurable.
Non-fundamental uncorrelated IID noise
This refers to price movements that are completely unrelated to fundamental information, often driven by noise traders, sentiment, liquidity needs, or mechanical trading strategies.
Does not contain any information about the asset’s intrinsic value.
Noise vs Sentiment
Noise is opinion on value unrelated to fundamental information, i.e., misinformation.
Sentiment is correlated noise - this has the potential power to move markets.
This implies that price movements can be driven by misinformation rather than information.
How can sentiment, noise trading, and positive feedback in markets be exploited?
By smart-money traders.
An arbitrage opportunity realises with over-(under-) valuation.
Smart traders anticipate this and buy (sell), contributing even more to the build-up of sentiment.
Prices go up (down), and this positive-feedback creates an illusion.
Uninformed traders may fall prey of this illusion and buy (sell).
Smart money traders exit the trade early and make positive profit.
Noise traders are late and lose money when the arbiyrage closes.
Can we test for market efficiency?
There are different ways to test for efficiency.
Under the random walk hypothesis, autocorrelations are zero so variance ratio should be 1 for market efficiency.
The efficient market hypothesis implies that returns are not predictable by conditioning on past prices (lagged returns), but there is evidence that such implication of the hypothesis does not hold.
There is evidence that the sign of the correlation depends on the time horizon:
* Short term (one month): reversal
* Medium term (several months): momentum
* Long term (years): reversal
Estimating Beta in the CAPM
Estimating beta means to find the historical statistical link between an asset’s excess return and the overall market excess return (the market risk premium or equity premium).
Beta is the composition of correlation and risk. The CAPM expresses a trade-off in risk and expected return.
Do prices follow the CAPM or should we extend the model?
If prices follow CAPM, there is little to do in terms of trading strategies.
Otherwise, we can extend the model by adding factors like size, vaue, etc.
Once the (extended) model is statistically significant, it becomes a ‘prediction’ model.
Banz 1981 Model
- Size factor (market cap)
Smaller firms (lower market cap) tend to have higher average risk-adjusted returns than larger firms.
Fama and French (1993) Model
Three factor model:
* Market price
* Size
* Value (e.g., book to market or price-to-earnings)
Lower P/E does better, i.e., value stocks outperform Glamour (growth) stocks.
How to use factor models to derive a trading signal?
Take a model and estimate it.
Compare the model’s price to the market price.
Construct a trading signal.
Trade.
If factors are tradeable assets…
If factors are tradeable assets and short selling is possible, we can trade the whole prediction model.
Construct a trading model where you can trade the explanatory factor, e.g., the market in CAPM.
Limits of trading a model
- Transaction costs
- Assets listed on different exchanges with different times
- Short-selling is expensive and can be difficult.
- Have to close the position to cash-in a gain
- Risk remains - not possible to hedge perfectly
- The model may lose prediction power.
