Options Trading, Binary Options

Option Trading: Information or Differences of Opinion? Siu Kai Choy and Jason Wei∗ Joseph L. Rotman School of Management University of Toronto 105 St. George Street Toronto, Ontario, Canada, M5S 3E6 First version: March 16, 2009 Current version: April 23, 2009 Abstract This paper investigates the motive of option trading. We show that option trading is mostly driven by differences of opinion. Our findings are different from the current literature that attempts to attribute option trading to information asymmetry. We present three specific findings. First, cross-section and time-series regressions reveal that option trading is significantly explained by differences of opinion. While informed trading is present in stocks, it is not detected in options. Second, option trading around earnings announcements is speculative in nature and mostly dominated by small, retail investors. Third, around earnings announcements, the preannouncement abnormal turnovers of options seem to predict the post-announcement abnormal returns. However, once we control for the pre-announcement returns, the predictability completely disappears. Keywords: option trading, differences of opinion, informed trading, speculation, earnings announcements.

JEL classification: G10, G12 and G14. ∗Email contacts: siu.choy05@rotman.utoronto.ca and wei@rotman.utoronto.ca. Jason Wei gratefully acknowledges the financial support from the Social Sciences and Humanities Research Council of Canada. Option Trading: Information or Differences of Opinion? Abstract This paper investigates the motive of option trading. We show that option trading is mostly driven by differences of opinion. Our findings are different from the current literature that attempts to attribute option trading to information asymmetry. We present three specific findings. First, cross-section and time-series regressions reveal that option trading is significantly explained by differences of opinion. While informed trading is present in stocks, it is not detected in options. Second, option trading around earnings announcements is speculative in nature and mostly dominated by small, retail investors. Third, around earnings announcements, the pre-announcement abnormal turnovers of options seem to predict the post-announcement abnormal returns. However, once we control for the pre-announcement returns, the predictability completely disappears. Keywords: option trading, differences of opinion, informed trading, speculation, earnings announcements. JEL classification: G10, G12 and G14. 0 1. Introduction Option trading has been steadily increasing. According to CBOE, in 2007, a total of 4.7 million transactions were made on options, representing a total dollar volume of $609 billion. These numbers more than doubled their counterparts a decade ago. If the market is complete, the trading volume of options should be indeterminate since one can create an equivalent option position by trading the underlying stock and a risk-free bond. Why the increase in trading volume then? An obvious answer is the presence of transaction costs which prohibits the perfect replication of options. Another potential reason is the leverage advantage identified by Black (1975). He argued that the leverage in options can attract informed traders attempting to exploit their private information. This line of thinking has motivated theoretical modelling and empirical testing of informed trading in options. Although supporting evidence is found in some studies (e.g., Amin and Lee, 1997; Easley, O’Hara and Srinivas, 1998; Cao, Chen and Griffin, 2005; and Pan and Potehsman, 2006), evidence to the contrary also exists (e.g., Stephan and Whaley, 1990; Vijh, 1990; Chan, Chung and Johnson,1993; and Chan, Chung and Fong, 2002). The overall empirical results on informed option trading are mixed at best.

The main driving force of option trading has yet to be identified. We take up this challenge in the current paper. We argue and empirically demonstrate that opinion dispersion is the main driver for option trading.1 We arrive at our conclusion based on three pieces of empirical evidence. First, using data from OptionMetrics for the period of January 1, 1996 to December 31, 2006, we regress option turnovers on various proxies for information asymmetry and differences of opinion. Both the cross-section and time-series regressions reveal that option trading is significantly explained by opinion dispersion. While informed trading is present in stocks as is speculative trading, informed trading is largely absent in options. Second, the trading patterns around earnings announcements further confirm the presence of speculative trading and the absence of informed trading in options. Option trading increases significantly around earnings 1In this article, we use interchangeably the terms “differences of opinion”, “opinion dispersion” and “disagreement.” We also use the term “speculation” when called for. Different opinions could arise due to differential interpretations of public information (e.g., Kandel and Pearson, 1995) while speculation doesn’t have to originate from information. Regardless, all the aforementioned terms are meant to be antonyms of “information” in our context, which in turn refers to private information. 1 announcements and the increase is almost entirely attributable to smaller, retail investors engaging in speculative trades.

To begin, we need to establish our priors regarding the impact of various proxies on the turnover. The priors on the 11 proxies for differences of opinion are relatively straightforward: More disagreements lead to more trading. We therefore anticipate a positive sign for Alog num analyst, Asidedness, ADISP and Avolatility in both (3.1) and (3.2). As for information asymmetry, we need to first understand the time-series trading properties for a single stock in the presence of information. As shown by Chae (2005) and others, the turnover of stocks tends to go down before scheduled information events such as earnings announcements and go up before unscheduled events such as M&A announcements. The opposite reactions are mainly due to the behavior of discretionary liquidity traders. They postpone transactions before scheduled public announcements in order to avoid trading with informed traders, causing the turnover to go down; for unscheduled events, by definition, the discretionary liquidity traders have no knowledge of them and unwittingly become the counterparts of the informed traders, causing the turnover to go up. As a result, we are less likely to detect cross-section differences in trading due to public information events since they tend to affect all stocks at the same time and in similar ways. For instance, macro-economic news will have simultaneous impacts on all stocks; the fiscal year of most firms coincides with the calendar year and as a result the earnings announcements tend to come out at the same time; and so on.

Thus the cross-sectional differences in trading, if any, are most likely caused by unscheduled events. We can reasonably argue that the more severe the information asymmetry, the bigger its impact on trading. Putting all the above together, we anticipate that small firms, higher PIN’s and wider bid-ask spreads are associated with more trading cross-sectionally. Therefore we expect a negative sign for Alog size and a positive sign for AP IN , Astock pba, and Aoption pba. Panel A of Table 2 contains the regression results for stocks. First and foremost, all four proxies of differences of opinion have a positive coefficient in all versions of the regressions and are significant at the 1% level. Opinion dispersion unambiguously increases trading. As for trading due to information asymmetry, only the size variable shows the strongest support with a significant negative coefficient. The PIN variable has the right sign and the t-values are significant at the 1% level, though much smaller than those for the size variable. The spread variable is only marginally 12 positive in the multivariate regression. Taken together, the results demonstrate convincingly the strong, positive impact of opinion dispersion on turnovers, and the mild impact of information asymmetry. Incidentally, it is apparent from the R-squares of the univariate regressions that the variable “sidedness” proves to be a successful proxy for dispersion. It commands the highest explanatory power among all the variables save firm size.4 We now turn to options in Panel B of Table 2.5 To the extent that options are more advantageous than stocks due to their leverage (Black, 1975), we should see stronger effects of information asymmetry and opinion dispersion on turnovers.

Each period is then divided into 3-day segments and the segment-sum of stock returns and the segment-average of turnovers (in logarithm) are calculated. We subtract the control-period turnover from those of the other periods and call the resulting differences “abnormal turnovers”. For each period, we divide the raw-return domain into 22 mutually exclusive ranges. We then match each return-range across different periods and tabulate the corresponding abnormal turnovers. Table 5 presents the results. For brevity, we only report the abnormal turnover for the non-event period. The rest are pair-wise differences in abnormal turnovers, all relative to the non-event period: (Pre — Non), (At — Non), and (Post — Non). The abbreviation “Non” stands for non-event period while “Pre”, “At” and “Post” stand for pre-announcement, announcement and post-announcement periods, respectively. For stocks, the results are consistent with those in Kandel and Pearson (1995) and Chae (2005): There is generally a drop in turnover during the pre-announcement period and a sharp increase in the announcement and post-announcement periods. This is consistent with the view that discretionary liquidity traders withdraw from trading before the announcement and come back right after. The fact that the abnormal turnover is significantly high for the announcement and postannouncement periods even when the price movement is near zero (referring to the middle cells in the table) means that the trading is most likely caused by investors’ differential interpretations of the same information or opinion dispersion. If there is also information trading in options in the pre-announcement period, we should see a similar dip in the turnover. The table shows quite the contrary: The option turnover actually shoots up in the pre-announcement period.

Trading is the most active during the announcement period, but drops below the pre-announcement period level after the announcement. This is in sharp contrast with stocks whose post-announcement turnover is much higher than that of the pre- 18 announcement period. To appreciate the magnitude of increase in the option turnover, take the return cell of [-0.5% ≤ R < 0] for call options as an example. The abnormal turnover for the preannouncement and announcement periods are 0.439 and 1.327 respectively. Since the turnovers are in natural logarithms, the abnormal turnovers mean that the turnovers in the pre-announcement and announcement periods are 1.55 and 3.77 times that of the non-event period. In summary, the active trading of options starts well before the actual announcement, leading up to the climax during the announcement period and quickly dies off after the announcement, consistent with the patterns of opinion-based trading or speculation. Taken together, the results in Table 5 clearly demonstrate that, unlike stocks, options are used for speculation purposes around earnings announcements. One potential, alternative explanation for the option trading pattern is the hedging demand for options due to the large movements in stock prices around earnings announcements. However, the stratification by returns is to control for the price movement in the underlying stock. One may argue that it is the anticipation of price movements that causes the additional hedging demand. But this essentially is one type of opinion dispersion or speculation.

Similar turnover can be defined for calls and puts separately. Alog_scaled_ratio: logarithm of volume ratio between option trading and stock trading. Specifically, for stock i on day t, the variable is defined as 0.001 where option volume refers to the total daily volume of all options and is the the total stock volume divided by the total option volume over the entire sample for stock i. The scale variable brings the trading volumes of stocks and options to comparable magnitudes. Independent variables Alog_price: logarithm of stock price. Alog_size: logarithm of firm size calculated as stock price times the number of shares outstanding. Search for some

binary options

Be wise enough when you choose to go for a options trading. APIN: PIN (probability of informed trading, quarterly frequency). Astock_pba: proportional bid-ask spread for stocks, calculated as the dollar bid-ask spread divided by the mid-point of bid and ask quotes. Aoption_pba: proportional bid-ask spread for options, calculated as the dollar bid-ask spread divided by the mid-point of bid and ask quotes. Alog_num_analyst: logarithm of one plus the number of analysts following the firm. Asidedness: sidedness as defined in Sarkar and Schwartz (2009). In this paper, it is estimated as the correlation between the numbers of buyerand seller-initiated trades over 5-minute intervals within the day. ADISP: dispersion of earnings forecasts – standard deviation of earnings forecasts measured in dollars per share. Ascaled_DISP: dispersion of earnings forecasts scaled by the average stock price within the calendar year of the estimates. |Ascaled_change_DISP|: absolute value of the change in Ascaled_DISP. Avolatility: return squared. Return_positive: return of the week if positive and zero otherwise. Return_negative: return of the week if negative and zero otherwise.

____________________________