Title: CHAPTER ONE Performance Leakage and Value Discounts on the TSX
1CHAPTER ONEPerformance Leakage and Value
Discounts on the TSX
- Lawrence Kryzanowski and Skander Lazrak
- Concordia University Brock University,
respectively
2Reading Questions
- What costs should an investor consider when
executing a transaction? - What type of trader is most concerned about the
quoted spread and how is it calculated? - What type of trader is most concerned about the
effective spread and how is it calculated? - What metric is available to measure price impact
and how is it calculated? - What information is obtained from examining the
quoted depth?
3Reading Questions Contd
- Are the various spread measures normally
distributed for the security types on the TSX? - Which security types have the lowest and highest
relative spreads, on average, on the TSX? - Describe how trades are classified as buyer- or
seller-initiated when this needs to be inferred
from quote trade data and what is the accuracy
of doing so? - Describe the time-series behavior of the number
of trades over time? - Discuss three implications of the findings of
this study?
4Performance Leakage and Value Discounts on the TSX
- Performance affected by trade execution quality
- Trade costs commissions, fees, execution
opportunity costs - Execution quality depends upon investors
investment style and trading demands
5Execution Quality
- Imputed from data since not observable
- Difference between actual trade execution price
fair value estimate - Pre- and post-trade measures
- Metrics include quoted, effective realized
spreads, quoted depth, number of trades trading
volume (number of shares dollar value)
6Quoted Spreads and Depth
- Quoted Spread
- Important for (e.g., momentum) traders whose
informational advantage becomes stale quickly - Quoted (Inside) Depth
- Important for traders whose trade sizes exceed
the quoted inside depth
7Effective Realized Spreads Price Impact
- Effective Spread
- Important for (e.g., value) traders whose
informational advantage does not become stale
quickly - Realized Spread
- A measure of the market-makers profit if s/he
rebalances inventory shortly after making the
trade - Price Impact Difference between effective
realized spreads
8Studied Sample
- All 2,300 listed securities on TSX with data for
first 2 months of 2008 - Security types common shares (CDN USD)
preferreds units warrants asset-linked notes - Examine all valid trades and quotes
- Over 197 million quotes 25 million trades
9Execution Quality Estimates
- Nearly all distributions are right skewed so that
the Median lt Mean - For all three relative spread measures
- Lowest averages and ? Common shares with prices
gt 5 and NT_NO - Highest averages and ? Warrants USD
- Considerable variation intra inter security
type - E.g., intra quoted spreads for common shares with
prices gt 5 range from 0.03 to 16.45
10Execution Quality Estimates Contd
- Quote depth averages ?
- Lowest highest Warrants Common gt5
- Traded volume (number shares) averages ?
- Lowest highest NT_NO Common lt5
- Traded volume () averages ?
- Lowest highest NT_NO (except Warrants for
median) Common gt5 - Number of trades averages ?
- Lowest highest NT_NO Common gt5
11Execution Quality Estimates Contd
- Intertemporal variation appearance in top 3 ranks
for the 6 liquidity measures - Warrants USD frequently appear
- Common gt5 never appears
- Common lt5 has solo appearance
- Units appears in top 3 for 2 measures
- NT_NO relatively higher for all six liquidity
measures.
12Conclusion
- Net benefits of trade impacted by material
variation in potential actual trade execution
costs intra inter security type
intertemporally (including global shocks) - Valuation discount estimates by security type
important for asset valuators investors for
nongranular fund holdings - Value discount or performance drag estimates
important for assessing reported portfolio
performance, particularly under adverse market
conditions
13CHAPTER TWOInformed Trading in Parallel Auction
and Dealer Markets The Case of the London Stock
Exchange
- Pankaj J. Jain
- Christine Jiang
- Thomas McInish
- Nareerat Taechapiroontong
14Reading Questions
- At one time the London Stock Exchange was losing
trades to other European exchanges. Why? - Describe the main characteristics of the London
Stock Exchanges SEATS and dealer markets. - Who can trade on the SEATS and dealer markets and
why? Describe a type of order that you might
decide to route to each market? - Which trading platform within the London Stock
Exchange has a higher permanent price impact and
why? - Which trading platform within the London Stock
Exchange has the ability to screen out informed
traders and why?
15Reading Questions
- How have the dealers obligations to provide
liquidity on the London Stock Exchange changed
over time? - What do you know about transparency (pre- and
post-trade) on the London Stock Exchange? - Provide your opinion about the various optimal
institutional design features of a stock
exchange. - Which trading platforms on the London Stock
Exchange are associated with lower temporary
price impact and total price impact? - List some key characteristics of a trade that
determine its price impact.
16Objectives and Contributions
- Test whether intensity of trader anonymity is
correlated with trading with informed traders
(adverse selection) - Use permanent price impact (PPI) of trades to
gauge information content of orders in 2 parallel
markets - Provide evidence based on unique structure of the
LSE. - Compare adverse selection problem between
parallel anonymous Auction market and
non-anonymous voluntarily Dealer market. - Fully time-synchronized markets and no
firm-specific differences (same firms)
16
17Previous Studies on Trader Anonymity
- Survey Institutional investors prefer to trade
in anonymous automated execution systems that
provide low disclosure of identity of the company
submitting the orders. - Economides and Schwartz (1995) and Schwartz and
Steil (1996) - Theory Negotiated dealer market serves as a
screening device to eliminate informed trades. - Seppi (1990) and Pagano and Roell (1992)
- Professional non-anonymous relationship between
specialist and brokers reduces the adverse
selection problem. - Benveniste et al. (1992)
- Off-exchange dealers are likely to cream skim
order flow and divert informed orders to
on-exchange market. - Easley, Kiefer and OHara (1996)
- Upstairs dealer market facilitates searching and
matching of order flow. - Seppi (1990), Burdett and OHara (1987) and
Grossman (1992)
17
18 Order flow of SETS stocks on the London Stock
Exchange
18
19Comparison of SETS and Dealer
markets in 2000
19
20Data selection and processing
- Main data source London transaction data service
- Compustat global file used for Market
capitalization - SETS stocks FTSE 100 or FTSE 250 in 2000
- Trading days gt80 days in 2000
- Sample stocks 177
- Delete 28 stocks for methodological problem
- Final sample 149
- Trade Reports File contains
- Trade direction (buy or sell)
- Trade location (SETS or Dealer)
- Code that identifies each counterparty, but
there no information concerning their actual
identity - Standard trade and quote filters are applied
20
21Methodology
- Keim and Madhavan (1996) and Booth et al.
(2002)s price impact measures - Permanent price impact BSln (PA/PB) inform.
content - Temporary price impact BSln (PT/PA) liquidity
cost - Total price impact BSln (PT/PB)
- Note BS is buy/sell indicator PB,PA,PT are
before, after, and trade prices
21
22Fig. 1. Cumulative average returns around large
GBP trades. We identify the 5 of trades that
have the greatest GBP value. We label each of
these trades, in turn, as trade 0. For each
trade 0, we identify the twenty previous trades,
trades -1 through -21, and the subsequent 21
trades, trades 1 through 21. We calculate the
return for each trade from -20 to 20 as the
difference in the log of the trade price minus
the log of the previous trade price. These
returns are averaged and cumulated beginning with
trade -20. Mean values of cumulative average
returns are plotted above.
22
23Table 4. Information Differences on SETS and
Dealer
Permanent Price Impact Permanent Price Impact Temporary Price Impact Temporary Price Impact
Independent variables Independent variables Coefficient t-statistics Coefficient t-statistics
Intercept -0.2224 -1.74 1.3699 12.19
SETS 0.2849 21.62 -0.1956 -16.92
Cap 0.0008 0.08 -0.0024 -0.26
Price 0.0274 2.00 -0.1060 -8.82
Volatility 0.1031 4.90 0.1176 6.37
Freq -0.0399 -4.77 -0.0180 -2.45
Size 0.0311 2.28 -0.0948 -7.93
Adj. R2 0.7062 0.7836
F-value 119.97 175.58
23
24Conclusions
- Regulators of the London Stock Exchange
accomplished their goals of providing efficient
markets by offering alternative trading venues. - Dealers compete effectively with SETS
- More number of trades on SETS
- Larger trade size on dealer market
- Price impact measures suggest that SETs trades
have larger information content - Dealers effectively screen out informed traders
or charge them more for providing liquidity
24
25CHAPTER THREEMomentum Trading for the Private
Investor
- Alexander Molchanov
- Philip Stork
- Massey University
26Reading Questions
- Is there momentum in the worlds largest shares
returns? - Are these momentum profits economically
meaningful? - Do the momentum returns depend on ranking and
holding periods? - How different is the magnitude of the momentum
effect in various regions? - Is momentum a disappearing phenomenon?
- Is trading volume a useful variable in
determining momentum profits? - How robust are momentum profits to trading costs?
27Motivation
- Momentum effect is one of the best-known return
predictability patterns - Jegadeesh and Titman (1993) Past winners keep
outperforming past losers - Behavioural finance has several potential
explanations (Daniel, Hirshleifer and Subramanyam
(1998), Hong and Stein (1999)) - Concensus is yet to be reached
- Simplicity of a momentum trading strategy makes
it attractive to a private investor
28Data
- We use only the largest shares in the most liquid
markets - Transaction costs, bid/ask spreads and lending
fees are likely to be lower for these shares - They are more likely to come into private
investors spotlight (Barber and Odean, 2008) - Constituents of five major indices, covering
years 1992 2008 - Dow Jones Euro Stoxx 50, DJIA 30, Dow Jones
Nordic 30, SP/ASX 50, SP/TSX 60 - Survivorship bias is (mostly) removed
29Momentum Profits
- Top five winners are bought and bottom five
losers sold short - Both ranking and holding periods are varied from
one to twelve months - One month is skipped between ranking and holding
periods - Momentum returns are observed in all markets, but
the statistical significance varies
30Momentum Profits
- Australia and Canada exhibit highest momentum
returns both in terms of magnitude (up to 31)
and statistical significance (T-Value up to 4.31) - For other markets, even though returns are mostly
positive, significance is not reached - Momentum returns generally increase with an
increase in ranking period length
31Robustness Checks
- Is momentum a disappearing phenomenon?
- Sample is split in two periods 1992 1999 and
2000 2008 - Returns decrease in Europe, US and Canada (in the
latter case remaining economically meaningful) - Returns increase greatly in Australia and remain
virtually unchanged in Nordic stocks - Do momentum returns depend on the size of
momentum portfolio? - Returns increase for extreme winners/losers, but
at a cost of greater volatility
32Volume Filter
- Momentum returns have been linked to trading
volume (Conrad, Hameed and Niden (1994), Lee and
Swaminathan (2000)) - Shares which show extreme volume changes (10 or
-50) are filtered out - Volume filter works for small-size portfolios
- For larger portfolios, the difference is negative
and small
33Private Investor Trading
- The following transaction costs are assumed
- Four times 0.3 and twice the bid-ask spread of
0.3, a total of 1.8 per six months - Stock lending fee of 1 and miscellaneous costs
of 0.4 - Total transaction costs of 5 per annum
- The actual costs are likely to be less than our
estimates - Canadian and Australian momentum strategies are
robust to trading costs and yield more than 10
per annum
34Conclusions
- Momentum effect is one of the better-known market
anomalies - We document significant momentum returns, robust
to a number of tests - Momentum profits are significant in Australia and
Canada, net of trading costs
35CHAPTER FOURTrading in Turbulent Markets Does
Momentum Work?
- Tim A. Herberger
- Daniel M. Kohlert
- University of Bamberg
36Reading Questions
- Do momentum strategies involving NYSE stocks
deliver positive abnormal returns in the period
from 1994 to 2009? - Are some momentum strategies involving NYSE
stocks more profitable than others in the period
from 1994 to 2009? - How do transaction costs affect profits of
momentum strategies involving NYSE stocks in the
period from 1994 to 2009? - Do profits of momentum strategies involving NYSE
stocks depend more on poorly performing loser
stocks or on well performing winner stocks in
the period from 1994 to 2009?
37Momentum Strategy
- Trading strategy that buys stocks that have
performed well in the previous J months
(winners) and sells stocks that have performed
poorly in the same period (losers) - Based on different combinations of ranking
periods (J) and holding periods (K) - J and K usually between 3 to 12 months
- Self-financing due to long/short approach
38Empirical Evidence (Selection)
- Jegadeesh and Titman (1993) First report of
positive and statistically significant returns of
momentum strategies for the U.S. stock market - Moskowitz and Greenblatt (1999) Strong
contribution of industry momentum to momentum in
stock returns - Rouwenhorst (1998) Positive market-adjusted
abnormal returns for a sample of 12 European
stock markets - Chui et al. (2000) Positive market-adjusted
abnormal returns for a sample of 8 Asian stock
markets - Glaser and Weber (2003) Positive market-adjusted
abnormal returns for the German stock market
39Data Set
- Monthly stock prices over the period from
12/31/94 to 05/31/09 for all stocks listed on the
New York Stock Exchange (NYSE) excluding - American Depository Receipts (ADRs)
- Real Estate Investment Trusts (REITs)
- Closed-end funds
- Delisted stocks
- Weighted NYSE index used as market proxy
- Source Datastream
40Methodology
- Three strategies
- Short-term (J 3, K 3)
- Medium-term (J 6, K 6)
- Long-term (J 12, K 12)
- Winner portfolio Best-performing percent of NYSE
stocks - Loser portfolio Worst-performing percent of NYSE
stocks - Overlapping portfolios
- Portfolio formation one month after ranking
41Methodology
- Gross return
- Market-adjusted abnormal return
- Market-adjusted abnormal return after transaction
costs -
42Results
Average Monthly Returns of Momentum Portfolios in the Period from 1994 to 2009 P1 Low-return loser portfolio, P2 High-return winner portfolio, J Ranking period (in months), K Holding period (in months), P2 P1 Zero-cost momentum portfolio, t-statistics in parentheses Average Monthly Returns of Momentum Portfolios in the Period from 1994 to 2009 P1 Low-return loser portfolio, P2 High-return winner portfolio, J Ranking period (in months), K Holding period (in months), P2 P1 Zero-cost momentum portfolio, t-statistics in parentheses Average Monthly Returns of Momentum Portfolios in the Period from 1994 to 2009 P1 Low-return loser portfolio, P2 High-return winner portfolio, J Ranking period (in months), K Holding period (in months), P2 P1 Zero-cost momentum portfolio, t-statistics in parentheses Average Monthly Returns of Momentum Portfolios in the Period from 1994 to 2009 P1 Low-return loser portfolio, P2 High-return winner portfolio, J Ranking period (in months), K Holding period (in months), P2 P1 Zero-cost momentum portfolio, t-statistics in parentheses Average Monthly Returns of Momentum Portfolios in the Period from 1994 to 2009 P1 Low-return loser portfolio, P2 High-return winner portfolio, J Ranking period (in months), K Holding period (in months), P2 P1 Zero-cost momentum portfolio, t-statistics in parentheses
Ranking Period (J) Holding Period (K) Holding Period (K) Holding Period (K)
Ranking Period (J) Portfolio 3 6 12
3 Loser (P1) -0.0167
Winner (P2) 0.0038
Winner Loser (P2 - P1) 0.0205
(t-stat) (4.16)
6 Loser (P1) -0.0187
Winner (P2) 0.0032
Winner Loser (P2 - P1) 0.0219
(t-stat) (5.93)
12 Loser (P1) -0.0091
Winner (P2) -0.0064
Winner Loser (P2 - P1) 0.0027
(t-stat) (1.12)
43Results
Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses Abnormal Monthly Returns of Momentum Portfolios Using the Average NYSE Return as the Market Proxy in the Period from 1994 to 2009 P Return of momentum portfolio, J Ranking period (in months), K Holding period (in months), M Average market proxy return, P M Abnormal return, Panel A Zero transaction costs, Panel B Transaction costs .2 percent, Panel C Transaction costs .5. Panel D Transaction costs 1, t-statistics for monthly return differences in parentheses
Ranking Period (J) Panel A Panel A Panel A Panel B Panel B Panel B Panel C Panel C Panel C Panel D Panel D Panel D
Ranking Period (J) Holding Period (K) Holding Period (K) Holding Period (K) Holding Period (K) Holding Period (K) Holding Period (K) Holding Period (K) Holding Period (K) Holding Period (K) Holding Period (K) Holding Period (K) Holding Period (K)
Ranking Period (J) 3 6 12 3 6 12 3 6 12 3 6 12
3 Market Return (M) 0.0026 0.0026 0.0026 0.0026
Momentum Return (P) 0.0209 0.0181 0.0140 0.0071
Abnormal Return (P - M) 0.0183 0.0156 0.0114 0.0045
(t-stat) (-3.31) (-2.82) (-2.08) (-0.82)
6 Market Return (M) 0.0015 0.0015 0.0015 0.0015
Momentum Return (P) 0.0223 0.0210 0.0189 0.0154
Abnormal Return (P - M) 0.0208 0.0195 0.0174 0.0139
(t-stat) (-5.10) (-4.77) (-4.27) (-3.41)
12 Market Return (M) 0.0021 0.0021 0.0021 0.0021
Momentum Return (P) 0.0030 0.0023 0.0013 -0.0004
Abnormal Return (P - M) 0.0009 0.0002 -0.0008 -0.0025
(t-stat) (-0.30) (-0.07) (0.28) (0.87)
44Results
- Significantly positive abnormal market-adjusted
returns for short- and medium-term momentum
strategies - Medium-term strategy delivers highest performance
- Significant abnormal returns for short- and
medium-term strategies after transaction costs of
.2 and .5 percent, and for medium-term strategy
after 1 percent transaction costs - No significant returns at all for long-term
strategy - Profits mainly driven by loser portfolios
45Conclusion
- Generation of superior returns using momentum
strategies possible over the period from 12/31/94
to 05/31/09 - Despite, or precisely because of turbulent market
environment
46CHAPTER FIVEThe Financial Futures Momentum
- Juan Ayora and Hipòlit Torró
- Universitat de València
47Main Objectives
- Testing the existence of the momentum effect in
financial futures markets. - Researching the influence of futures volatility,
trading volume, and open interest on momentum
strategy performance.
48Testing Momentum Effect (I) Procedure
- Formation periods (F) and holding periods (H) are
set at 1, 3, 6, 12 and 24 months. - In each F period, futures are grouped in three
portfolios P1, P2 and P3 - P1 contains past winners
- P3 contains past losers
- P2 contains the remaining contracts.
- We then compute portfolio returns for each
holding period. - Finally, the momentum portfolio return is
calculated as P1-P3. That is, by taking long
positions in the past winning contracts, and
short positions in the past losing contracts. - The momentum effect exists if the average return
of this portfolio is positive and significantly
different from zero.
49Testing Momentum Effect (II) Results
Return averages and t-statistics (above)
50Momentum, Volatility, Trading Volume, and Open
Interest
- Bessembinder and Seguin (1993) results
- Positive relationship between trading volume and
volatility. - Negative relationship between open interest and
volatility. - Cross-effect an increase in open interest
reduces trading volume influence on volatility. - Abnormally high trading volume associated with
positive shocks in prices. - Next, we analyse implications on momentum
returns.
51Momentum Returns and Futures Volatility
Return averages and t-statistics (above)
52Momentum Returns and Trading Volume
Return averages and t-statistics (above)
53Momentum Returns and Open Interest
Return averages and t-statistics (above)
54Momentum Returns, Open Interests, and Volume
Return averages and t-statistics (above)
55Conclusions
- Momentum strategies for 6 and 12 months produce
significant returns. - Momentum effect is more persistent in highly
volatile futures contract portfolios. - Momentum effect is more persistent in futures
portfolios with low relative increase in trading
volume. - Momentum effect is more persistent in futures
portfolios with high relative increase in open
interest - When open interest and traded volumes are allowed
to interact, the momentum effect is more
persistent in futures portfolios with high
relative increase in traded volume, and low
relative increase in open interest. - 6 and 12 month momentum strategies returns are
shown to be abnormal - Risk-adjusted momentum returns using CAPM and
Fama and French three-factor model only partially
explain momentum strategy returns. - Bootstrap test shows that mean and Sharpe ratios
are above the percentile 99.5 of the empirical
distribution.
56CHAPTER SIXOrder Placement Strategies in
Different Market Structures A Primer
- Giovanni Petrella
- Catholic University (Milano, Italy)
56
Handbook of Trading, edited by Greg N. Gregoriou
57Reading Questions
- When should you submit a limit order?
- When should you place a market order?
- Will an aggressive trader use limit orders?
Explain your answer. - What is the winner's curse for a limit order
trader? - What effect does the bid-ask spread have on the
decision to take or offer liquidity? - Explain the factors affecting the probability of
limit order execution. - What is a liquidity event?
- What is the difference between efficient and
inefficient return volatility? - Which is the type of order that you would use
when the stock price is mean reverting? Explain
your answer. - Which is the type of order that you would use if
a news is expected to arrive? Explain your answer.
58Order Placement Strategy
- The order placement strategy refers to
- the type of orders (i.e., to trade via limit
orders, market orders or a combination of both), - the size of the orders,
- and the timing of the orders
- The order placement strategy depends on the
relative merits and costs of limit orders and
market orders - What are the costs and returns from placing limit
orders?
58
Handbook of Trading, edited by Greg N. Gregoriou
59Costs of Trading by Limit Orders
- Risk of adverse informational change (also known
as ex post regret or cost of being bagged or
winners curse) - Heads You Win
- The market moves against the limit order trader
(thats why I will regret the execution) - Bearish news has caused the price of the stock to
fall and my buy limit order executes - Bullish news has caused the price of the stock to
rise and my sell limit order executes - Risk of limit order not executing (i.e., non
execution cost) - Tails I Lose
- Bullish news has caused the price of the stock to
rise and my buy limit order does not execute - Bearish news has caused the price of the stock to
fall and my sell limit order does not execute - So, why did I place that limit order?
59
Handbook of Trading, edited by Greg N. Gregoriou
60Benefits of Limit Orders A Better Price
- Limit order traders hope to get better prices
- Buyers who submit limit orders hope to buy at the
bid - If they had submitted a buy market order instead,
they would pay the ask price (which is higher
than the bid) - Sellers who submit limit orders hope to sell at
the ask - If they had submitted a sell market order
instead, they would receive the bid price (which
is lower than the ask) - but they do not always realize their hopes
- Limit order traders receive better prices only if
their order actually trades - If the market moves away from their limit price,
they may never trade - If they still want to trade, they will have to
chase the price by raising their bid or
lowering their offer - This would make the final price actually worse
than the price that they would have obtained had
they used market orders
60
Handbook of Trading, edited by Greg N. Gregoriou
61 When a Liquidity Event Occurs!
- A liquidity event is the arrival of a trader on
the other side of the market who is buying
liquidity (and is not an informed trader) - I posted a buy limit order and an impatient
seller arrived - I posted a sell limit order and an impatient
buyer arrived - Alternatively, you can look a the same event as
mean reversion in the pricing process - I posted a buy limit order and the price first
goes down and then up - I posted a sell limit order and the price first
goes up and then down - Again, why did I place that limit order?
- Because I expected that sufficient mean reversion
would offset the costs that might result from
informational change
61
Handbook of Trading, edited by Greg N. Gregoriou
62Order Placement Strategy in a Continuous
Order-Driven Market
- Should I submit
- A market order?
- A limit order?
- If a limit order, how should I price it?
- The decision is made with respect to
- Gains from trading
- A buy (sell) market order would pay (receive) a
higher (lower) price - A buy (sell) limit order would pay (receive) a
lower (higher) price - Probability of a limit order executing
- A market order would be executed with certainty
- A limit order would not be executed with certainty
62
Handbook of Trading, edited by Greg N. Gregoriou
63Factors Affecting the Probability of Order
Execution (I)
- Price aggressiveness
- For buy orders, the higher is the limit price,
the higher is the probability of execution - For sell orders, the lower is the limit price,
the higher is the probability of execution - Demand/supply equilibrium
- The larger is the size of the depth on the buy
(sell) side, the lower is the probability of
execution for a buy (sell) limit order - The market order arrival rate
- The larger is the arrival rate of market order on
the opposite side of the market, the higher is
the probability of execution for a limit order
63
Handbook of Trading, edited by Greg N. Gregoriou
64Factors Affecting the Probability of Order
Execution (II)
- Duration of the order
- The longer is the duration of the order, the
higher is the probability of execution - Volatility
- The larger is price volatility, the higher is the
probability of execution for a limit order - This does not imply that order execution will be
profitable, this just means that higher
volatility entails higher probability of
execution - Execution is profitable for a limit order trader
if the volatility that triggered the limit order
execution is temporary - Execution is unprofitable for a limit order
trader if the volatility that triggered the limit
order execution is permanent
64
Handbook of Trading, edited by Greg N. Gregoriou
65Implications for Trading Strategies in a
Continuous Order-Driven Market
- High volatility days or high volatility stocks
make convenient to use limit orders - You might even place limit orders on both sides
of the market to benefit from accentuated
intra-day volatility - When operating in this manner you resemble a
dealer (buy at the bid and sell at the ask) - Relatively patient traders place limit orders
- This is called Passive Trading Strategy (PTS)
- Large investors may prefer PTS because this
strategy does not cause market prices to change
(market impact) as the orders are naturally
absorbed by the market - Relatively eager traders place market orders
65
Handbook of Trading, edited by Greg N. Gregoriou
66Continuous Order-Driven Markets vs. Call Auctions
- Immediacy
- Waiting time is lower in continuous order driven
markets - Price discovery
- Prices are better in call markets then in
continuous markets - Prices are volatile in continuous markets because
of news (efficient volatility) and order
imbalance (inefficient volatility) - Traders get their orders executed depending on
the current market conditions at the time of the
submission - In a call auction, the terms of your trade will
depend on the market conditions at the end of the
pre-opening phase (which implies that you may get
more orders on the opposite side of the market,
and thus you may get better conditions)
66
Handbook of Trading, edited by Greg N. Gregoriou
67Continuous Order-Driven Markets vs. Call Auctions
- Trading volume
- Volume is higher in continuous markets
- Some traders that do not get execution in a call
market, may execute their trades in a continuous
market (with zero surplus) - If on the opposite side of the market, there are
impatient traders demanding liquidity - In short
- Continuous markets sacrifice surplus (i.e.,
expected profits) for immediacy - Call auction markets sacrifice immediacy for a
better price discovery process
67
Handbook of Trading, edited by Greg N. Gregoriou
68Implications for Trading Strategies in a Call
Auction
- There is no bid-ask spread in a call auction
because all executed orders clear at the same
price - Executed orders receive price improvement (or
positive surplus or positive expected profits) in
a call auction - Buy orders priced above the single clearing price
and sell orders priced below it receive the price
improvement - Orders priced exactly at the clearing price do
not receive price improvement - Traders are more aggressive in a call auction
- The limit price affects the probability of
execution, but it does not set the price
68
Handbook of Trading, edited by Greg N. Gregoriou
69CHAPTER SEVENProfitability of Technical Trading
Rules in an Emerging Market
- Dimitris Kenourgios
- University of Athens, Department of Economics
-
- Spyros Papathanasiou
- Hellenic Open University, School of Social
Sciences
70 Reading Questions
- Why the profitability of technical trading
strategies does not support the Efficient Market
Hypothesis? - Describe why and how traders use moving average
trading rules. - Why academics have main concerns regarding the
validity of technical analysis? - Describe how the bootstrap methodology overcomes
traditional statistical problems of financial
time series. - What is the reason for the use of a generalized
autoregressive conditional heteroskedasticity
(GARCH) model under the bootstrap methodology?
71Reading Questions
- Why it is important to examine the profitability
of technical trading rules after considering
transaction costs? - The main conclusion of this chapter is that
technical trading strategies win the buy and
hold strategy. Present and discuss the main
results which confirm this argument. - Are the reporting results consistent to the study
of Brock, Lakonishok and LeBaron (Journal of
Finance, 1992) and other existing empirical
research? Why? - Do the results of this chapter support or not the
use of technical analysis? - How traders would benefit reading this chapter?
72Purpose
- This chapter investigates the profitability of
technical trading rules in the Athens Stock
Exchange (ASE). - We compare various moving average trading
strategies against the buy and hold position in
the spirit of Brock, Lakonishok and LeBaron
(Journal of Finance, 1992), employing traditional
t-test and bootstrap methodology. - The profitability of technical trading rules
contradicts the Efficient Market Hypothesis
(EMH).
73Contribution
- Although the majority of the professional traders
and investors use technical analysis, empirical
research is still limited since most academics,
until recently, had not recognized the validity
of these methods. - It focuses on a less developed and efficient
stock market, given the existing paucity of
research in such markets. - In contrast to prior relevant studies, it
examines the profitability of technical trading
rules after transaction costs. - Statistical and more advanced econometric methods
can help traders to implement profitable
technical trading strategies.
74Literature (1)
- U.S. stock markets
- Brock, Lakonishok and LeBaron BLL (1992)
investigate two popular trading rules (moving
average and trading range break rule), utilizing
the Dow Jones Index from 1897-1986. - Kwon and Kish (2002) provide an empirical
analysis on technical trading rules (the simple
moving average, the momentum, and trading
volume), utilizing the NYSE value-weighted index
over the period 1962-1996. - Their results indicate that the technical trading
rules can capture profit opportunities over a
buy-hold strategy.
75Literature (2)
- Emerging and less developed stock markets
- Singapore Wong, Manzur and Chew (2003)
- Jordan Atmeh and Dobbs (2006)
- Greece Vasiliou, Eriotis and Papathanasiou
(2008a, 2008b) - Their results show that the application of
technical trading strategies provide substantial
profits.
76Methodology
- t-test we compare the mean returns of the
unconditional buy methodology with the returns of
the buy signals given by the moving averages and
the returns of the unconditional buy methodology
with the returns of the buy signals minus the
returns of the sell signals given by the moving
averages. - ? Null hypothesis No actual difference between
mean returns. - 2. Bootstrap methodology under GARCH(1,1)
model we compare the returns conditional on buy
(or sell) signals from the actual FTSE/ASE-20
data with the returns from simulated comparison
series generated by a GARCH model (500
replications). - ? Null hypothesis the trading rule excess return
calculated from the original series is less than
or equal to the average trading rule return for
the pseudo data samples.
77Data
- Daily closing prices (in logs) of the FTSE/ASE-20
index from 1/1/1995 to 31/12/2008. - The returns are calculated after transaction
costs. - We evaluate the performance of the following
moving average rules against the buy and hold
strategy 1-9, 1-15, 1-30, 1-60, 1-90, 1-120 and
1-150. - The first number in each pair indicates the days
in the short period and the second number shows
the days in the long period.
78Empirical Results (1)
- t- test results
- Rejection of the null hypothesis.
- The technical strategies win the buy and hold
strategy (FTSE/ASE-20). - The buy-hold strategy gives 5.109 annually
return, while the strategy of simple moving
averages 24.675.
79Empirical Results (2)
- Bootstrap results
- Acceptance of the null hypothesis.
- Most of the simulated GARCH (1,1) series generate
a mean return larger than that from the original
FTSE/ASE-20 index. - The technical strategies win the buy and hold
strategy (FTSE/ASE-20).
80Conclusions
- Making trading decisions based on moving average
rules lead to significantly higher returns than
the buy and hold strategy, even after transaction
costs. - Technical rules produce useful signals and can
help to predict market movements. - This contradicts the EMH since traders and
investors can gain significant abnormal returns. - Findings are consistent to the existing empirical
evidence.
81CHAPTER EIGHTTesting Technical Trading Rules as
Portfolio Selection Strategies
- Vlad Pavlov
- Stan Hurn
- Queensland University of Technology
82Reading Questions
- What is technical analysis? Outline how
moving-average rules operate. - Explain what is meant by the data-snooping bias.
- What is the Reality Check?
- How is an arbitrage portfolio constructed.
- Why is it difficult to assess the profitability
of technical trading rules?
83Literature
- Brock, Lakonishok and LeBaron (1992) - bootstrap
analysis of a few popular technical trading rules - Lo, Mamaysky and Wang (2000) - wide universe of
rules - Sullivan, Timmermann and White (1999) data
snooping bias in rules selection, Reality Check
test
84Problems
- Multitude of trading rules
- Infrequent signals
- requires large amounts of data
- Most studies use either one or a small number of
time series - need very long histories
85The Main Idea
- Apply trading rules to a large dataset of stocks
- move away from using a single index or currency
- Use trading rules to form portfolios
- from an arbitrage portfolio using buy and sell
signals
86The Data
- AGSM dataset comprising monthly observations on
prices and dividend payments for 6000 ASX stocks
from 1973 to 2008. - How to deal with missing observations?
- Mean returns become very sensitive to the
treatment of missing returns and exits for very
small stocks. - Empirical exercise limits attention to top 500
stocks. - Additional liquidity filters used to limit the
effect of missing observations.
87MA Trading Rules
- Compute two MAs (long and short)
- BUY signal
- short MA (t) gt long MA (t) and short MA (t-1) lt
long MA (t-1) - SELL signal
- short MA (t) lt long MA (t) and short MA (t-1) gt
long MA (t-1) - Arbitrage portfolio
- purchase all stocks which generate buy signals
in a particular month and finance this purchase
by shorting the stocks for which sell signals
are generated
88Example of MA Rule
89Empirical Results
- MA rules do appear to generate abnormally high
returns - The simple MA rule works as traditionally
implemented - Exponentially weighted MA rule requires
non-traditional contrarian implementation - Reconciliation of this result hinges on
recognition of relationship between the two
different types of MA rule. - Exponentially weighted rule actually magnifies a
small area of the possible outcomes generated by
the simple MA rule relating to simple MA rules
with averaging less than 12 months - Exponentially weighted MA rules may be more
appropriate because averaging periods greater
than a year generate very few trading signals.
90Bootstrap Results
91Conclusion
- The analysis in this paper does not lead to any
specific investment strategies based on MA
crossover rules. - In the Australian data over the period from 1973
to 2008, however, contrarian interpretation of MA
trading rules generate profits over a one month
horizon. - There is an obvious mode of the return
distribution which is very difficult to explain
by alluding to pure chance selection. Some
parameter subset of these MA rules appear to pick
a systematic factor in returns that is not one of
the well-known Fama-French factors.
92CHAPTER NINEDo Technical Trading Rules Increase
the Probability of Winning? Empirical Evidence
from the Foreign Exchange Market
- Alexandre Repkine
- Korea University
93Reading Questions
- What is the primary purpose of applying technical
analysis rules to trading financial instruments
and commodities? - What would be the realistic trading strategies
applied by the practicing traders? - Why are stop-loss and stop-limit orders important
for any trading strategy? - What would be the probability of a currency pair
moving by b points in the favorable direction
before moving by a points in the opposite
direction? - What would be an average return on a random-entry
strategy based on a stop-limit order at b points
and a stop-loss order at a points?
94Reading Questions
- Can technical trading rules be used for entry so
as to increase the probability of winning and to
decrease the probability of losing? - How can one use pattern recognition techniques in
order to identify charting patterns of technical
analysis such as e.g. double bottom? - What are the differences between empirical and
theoretical probabilities of winning? - Is there a technical trading strategy that would
result in positive returns for all currency
pairs? - What would be the reason for the persistent
popularity of technical analysis among practicing
traders?
95The Value of Technical Analysis Rules
- Technical analysis rules are sometimes thought of
as a means of realizing consistently positive
returns - Academic literature has recognized the ability of
certain technical rules to increase returns, but
it also emphasized the unstable performance of
these rules - Our question is not about consistent returns, it
is about increasing the probability of winning - Does the application of technical trading rules
in the foreign exchange market increase the
traders probability of beating the market?
96Risk- and Profit-Management in Real-World
Trading Strategies
- A position is opened according to some technical
trading rule - Stop-limit orders specify the level of profit on
an open position at which the latter must be
closed (profits are earned) - Stop-loss orders specify the level of loss on an
open position at which the latter must be closed
(losses are incurred) - Stop-limit orders help traders run their profits,
stop-loss orders help traders limit their losses
97Efficient Market Hypothesis
- The efficient market hypothesis implies it is
impossible to earn positive returns on a
random-entry strategy with whatever levels of
stop-limit (b) and stop-loss order (a) specified
since the expected return in that case will be - What if the market entry (creating an open
position) is not random? What if one enters the
market using signals produced by technical
trading rules? - Hypothesis Technical analysis rules shift the
probability of winning up to the level of
, making the return due to the technical
analysis strategy positive - We use two technical analysis rules for entry
double bottom and bull flag
98The Double Bottom Pattern
1 2 3 4 5 6 7 8 9 10
1 0.5 -1 -1 -1 0.5 0.5 -1 -1 -1 0.5
2 1 0 -1 0 1 1 0 -1 0 1
3 0.5 0 -1 0 0.5 0.5 0 -1 0 0.5
4 0 0.5 -1 0.5 0 0 0.5 -1 0.5 0
5 0 1 -1 1 0 0 1 -1 1 0
6 -1 1 0 1 -1 -1 1 0 1 -1
7 -1 0.5 0 0.5 -1 -1 0.5 0 0.5 -1
8 -1 0 0.5 0 -1 -1 0 0.5 0 -1
9 -1 0 1 0 -1 -1 0 1 0 -1
10 -1 -1 0.5 -1 -1 -1 -1 0.5 -1 -1
- Shaded cells represent the double bottom pattern
- Figures in each cell represent the weights that
reflect the extent to which the actual foreign
exchange rates are fitting the double bottom
pattern - This charting template is fitting each
individually observed exchange rate into one of
its cells, so a fitting statistic can be
calculated - If the fitting statistic is large enough, entry
occurs
99Calculating the Fitting Value An Example
- Suppose we want to see whether the exchange rate
movements were fitting the double bottom pattern
during the past 100 days - Suppose we choose the exchange rate fluctuation
range between the values of 1 and 2 during the
past 100 days - Each column in our template would correspond to a
sub-period of ten days with the first row of the
template corresponding to the range of 1.92,
the second one to 1.81.9 and the tenth row
(which is the bottom row) corresponding to the
range of 11.1 - If hypothetically for the past 100 days the
exchange rate were fluctuating in the range of
1.92, then the value of the fit for each
10-day sub-period would be simply equal to the
template weight value, and the total fit for the
historical window would be equal to the sum of
the weights in the first row of the template
table
100Empirical Study Plan
- The sample period January 1st, 1999, through
January 31st, 2007 - Fit the two trading patterns within the 40 days
trading windows - Use data on the worlds ten major currency pairs
- For each currency pair and one of the two trading
rules, round the highest estimated fitting value
down to the nearest integer and use this value as
a signal to enter the market - Computing the difference between empirical and
theoretical probabilities of winning - Choose the currency pair (e.g. EUR/USD)
- Let and vary between one and one thousand
points with the increment of one point, which is
supplying us with one million strategies for each
currency pair and the technical analysis pattern - Compute the number of profitable entries, i.e.
the ones that result in the exchange rate
deviating by points in the profit-making
direction before deviating by points in the
loss-making direction. Call the share of such
profitable entries . This will be our empirical
probability of a profitable entry. - In case we find for a particular trading rule and
currency pair, the average expected empirical
return on this rule is statistically positive for
this currency pair
101Differences between Empirical and Theoretical
Probabilities of Realizing Positive Returns due
to Technical Analysis Application.
EURUSD GBPUSD NZDUSD USDCAD AUDUSD
Bull -1.54 -21.78 25.38 11.41 13.93
Doulbe Bottom 38.93 31.24 -14.96 11.08 1.07
USDCHF USDJPY GBPJPY GBPUSD EURCHF
Bull 23.90 25.00 10.00 17.62 -49.47
Double Bottom 26.15 50.00 2