2.2. Metodología y procedimientos
2.2.7. Técnicas de procesamiento y análisis de datos
As shown in the literature review chapter, international stock markets are highly
correlated with each other. While there is a great deal of controversy over the sources of the market comovement, the time varying properties and the benefits from global
diversification, researchers tend to agree that the returns of the local stock markets are influenced by the performances of the larger markets, such as the U.S., especially during
market downturns. In particular, the opening prices from local markets reflect overnight news from international markets, while the closing prices from the local markets reflect
country-specific information released during the day. This assumption is the basis of the trading strategy proposed in this thesis.
A profitable trading strategy cannot be developed using international stock markets comovement alone because stock markets are efficient enough to eliminate
profits from international lead-lag relationships between countries in the same time zone. Even though an Asian trader who is in a different time zone may want to exploit
the U.S. market returns as a leading indicator, the overnight return of the U.S. market is merely ex-post information with the local market already reflecting this information in
the opening price. Hence, profits cannot be guaranteed by simply following the U.S. market.
The short-term momentum effect is a necessary, but not sufficient, assumption for developing a profitable trading strategy. Momentum effect is the concept that past
winners (losers) will remain winners (losers) in the future. It is a market neutral strategy as a momentum investor holds both long and short positions at the same time. The
strategy proposed in this thesis is different from a momentum strategy because it aims to forecast daily market direction. However, the source of the momentum effect which
is underreaction or overreaction to information is employed to develop a day-trading strategy proposed here. The basic assumption of the trading strategy proposed in this
thesis is that market participants underreact to the information in the short term, but overreact in the long term.
Jegadeesh and Titman (1993) state that stock markets underreact to information about short-term prospects due to delayed stock price reactions. According to Hong
and Stein (1999), information diffuses gradually across the population, and stock prices underreact in the short run. The underreaction, resulting in the continuation of short-
term returns, generating positive autocorrelation in returns which suggests a strategy which follows this trend can be effective. Zhang (2006) investigates behavioural bias
which is traders’ underreaction to new information. He asserts that information uncertainty results in short-term price continuation. De Bondt and Thaler (1985)
interpret the trend in a different way. They provide empirical evidence of overreaction to unexpected news. Daniel et al. (1998) explain that overreactions come from investors’
overconfidence about the precision of their private information. Jegadeesh and Titman (2001) argue that delayed overreactions to information make momentum strategies
profitable.
The main argument of the trading strategy proposed in this thesis is that
opening prices reflect the U.S. market returns as overnight information, but smart traders influence the opening prices in the local markets considerably irrespective of the
U.S. overnight returns. The term “smart traders” or “informed traders” includes investors who obtain all relevant information promptly (whether it is public or private)
and judge the impact of information correctly, thus trading financial assets in a rational way. Their monopolistic forecasting power is incorporated into prices gradually (Kyle,
1985). The opening price in most stock markets is determined by a 30 minute to one hour blind auction. At this stage, investors adjust their portfolio according to their
judgement on the impact of overnight information. If investors think the U.S. market return is the best proxy for overnight information, they would decide the buying or
selling price in accordance with the U.S. market returns. For example, if the S&P 500 fell by one percentage point the previous night, a trader who wants to buy a stock in the
opening auction in the Japanese market is expected to place a buying order at one percent below the previous day’s closing price. The trader would judge a fair value of an
opening price using historical evidence of the comovement between the U.S. and Japanese stock markets. So, if there is a perfect correlation between the two markets,
there would be no reason to buy a stock at more than one percent below the previous day’s closing price on the basis of the U.S. market return which is the source of
overnight information.
What if the trader has a more important source of overnight information than
the U.S. performance which others do not have? What if the trader more accurately judges that the overnight U.S. news is positive news for the Japanese market despite a
drop of the U.S. market index? What if the trader correctly interprets how much the U.S. news affects the Japanese market and concludes that only a 0.5 percent drop is
reasonable? The trader would then be willing to buy at higher than negative one percent. If the trader’s judgement is right, we call the trader a “smart trader” or an “informed
trader”. In the same way, a smart trader can sell at lower than negative one percent of the previous closing price if the trader has negative information on the Japanese market
which is not necessarily connected with the U.S. market return. The trader might also judge that the U.S. market decline of one percent will have an even greater influence on
the Japanese market. In that case, the trader would be willing to sell at lower than negative one percent and will not place a buying order at higher than negative one
In our research, we use the local close-to-open returns 1 reflecting the arrival of overnight U.S. information. Hence,
, ∙ 1 ∙ Eq1) ∙ 1 ∙ , Eq2) where is a positive or negative effect of daytime information and is the sensitivity
of the local market to the U.S. market returns. For simplicity, we naively assume that
is equal to 1 and is constant in time.3
Equation 1 and 2 assume the overnight U.S. market return is the only factor to
influence the opening price in the local market. However, we know there are fluctuations in the observed opening prices which may differ from the theoretical prices
predicted by the U.S. market return due to the presence of smart traders. We exploit the smart traders’ forecasting power to generate a trading model, and this factor is added to
Equation 1 and 2.
∙ 1 ∙ Eq3) ∙ 1 ∙ ∗ Eq4)
where π is the buying or selling power of smart traders, which can be positive or
negative. The daytime information ( ) is replaced by ( ∗ ) in Equation 4 because the
predictive power of smart traders is already included in . Smart traders can predict the
market direction based on the private information they have, therefore they are willing
to buy higher (or sell lower) than the level determined by overnight information. If a trader only has the U.S. market performance as overnight information, the trader would
expect the local market to open at a level roughly reflecting the U.S. market return.
3
However, if smart traders have additional information about stock prices, they will gladly bid up the price (or sell lower) when they have positive (or negative) information.
Thus, it is reasonable to suggest that π reflects the predictive power of smart traders. The proposed trading strategy uses the sign of π: if the sign is positive, a buy
signal is generated at opening time, and vice versa. For example, if the U.S. market increased by 1% and the local market increased by 1.5%, then π would be positive 0.5%
and, a buy signal would be generated. In the same way, if the U.S. market increased by 1% and the local market increased by only 0.5%, then π would be negative 0.5% and a sell