See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/328936755

Leveraging a call-put ratio as a trading signal

Article in Quantitative Finance · November 2018

DOI: 10.1080/14697688.2018.1538563

CITATIONS READS
8 5,436

2 authors:

Patrick Houlihan German Creamer

Stevens Institute of Technology Stevens Institute of Technology
4 PUBLICATIONS 25 CITATIONS 69 PUBLICATIONS 883 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

NYC Computational Economics & Complexity Workshop View project

All content following this page was uploaded by German Creamer on 06 May 2019.

The user has requested enhancement of the downloaded file.

 Leveraging a call-put ratio as a trading signal

Patrick Houlihan, Germán G. Creamer

Stevens Institute of Technology, Hoboken, NJ
[email protected], [email protected]

Abstract

We examine whether a put-call ratio, derived from a unique set of market data, can be used to

predict directional moves in asset prices during various market conditions between March 2005

and December 2012. Our findings show: 1) specific market participant’s options trading volume is

a predecessor to asset price movements, and 2) portfolios based on the put-call ratio adjusted for

four factors Carhart model and transaction costs exhibit abnormal excess returns.

‘It amazes me how people are often more willing to act based on little or no data than to use data that is a challenge
to assemble.’
― Robert J. Shiller

Keywords: Anomalies in Prices; Portfolio Management; Technical Trading; Financial Forecasting; Investment
Management; Options; Behavioral Finance.
I. Introduction

Consumers invest in goods or services to fulfill particular needs and to gain an expected

utility. If the demand is not satisfied or no utility is gained, they will most likely never purchase

such goods or services again. Given many products and services, consumers may discover

superior alternatives by trial and error or through research. To be more knowledgeable,

consumers can turn to informative online reviews which influence purchasing decisions (Chen et

al. 2008; Vermeulen et al. 2009; Zhu et al. 2010) and some consumers take heed of reviews that

line up with their beliefs (Mullainathan et al. 2005). Understanding purchasing decisions or

behavior can provide an investor with information that warrants an investment in an asset. For

example, famed portfolio manager Peter Lynch observed long lines at multiple locations of The

Limited retail store (Lynch and Rothchild, 2000), and if he had acted on those observations, he

would have seen a 100x return a few years later. A well-informed investor will purchase a

reasonably valued asset based on the present value of discounted cash flows (Cochrane 2009)

which can be influenced by investors’ decisions. In an ideal world, the strong form of the

efficient market hypothesis (EMH) (Fama, 1970) stipulates a fairly valued asset factor in all

public and private information. However, Fama et al. (1969) showed that stocks experienced

periods of abnormal returns prior to publically available company announcements, suggesting

informed investors had access to either superior research (Karpoff et al. 2010; Christophe et al.

2010) or had private access to inside information (Agrawal et al. 2014; Diamond et al. 1987) as

their investing behavior led (Easley et al. 2008) other investor types, pointedly uninformed

investors.

The question is: how can informed investor behavior be observed or measured and

leveraged as a trading signal? In this paper, we focus on a practical approach by creating such a
measure, a trading signal that contains a component of investor behavior. We start by providing

an in-depth literature review of the current landscape of sentiment, the data sources sentiment is

derived from, and how it is measured and leveraged by investors in trading strategies. Next, we

formulate a trading signal, call-put ratio, by leveraging a similar derivate dataset that Pan et al.

(2006) studied to use in trading simulations. Our results also adjust returns for risk and

momentum.

Additionally, as a contribution to this research, we include actual transaction costs (as a

function of shares bought and sold) to capture a more realistic alpha. Pan et al. (2006) perform

daily cross-sectional regressions, whose dependent variable is the next day four-factor risk-

adjusted return, and independent variables are based on a put-call ratio, firm size and the Easley

et al. (1996) and Easley et al. (2002) information based probability ratio, PIN. While these

models performed well in a theoretical context by exhibiting statistical significance in explaining

asset returns, model parameters were determined using the entire data set to formulate the slopes

and t-statistics for the respective regression models. However, in a practical trading scenario, we

do not have the luxury to use the entire data set to perform a regression as we would be data

snooping using future data. However, we also conduct a regression analysis and compare its

results with our trading simulations. Another contribution of our research is the use of a more

recent data set that includes pre, during and post-financial crisis, bringing us through varying

market conditions that were not tested in previous research. We will show our trading signal for

a trader, when used in short, long or hybrid strategies, yields abnormal returns when compared to

holding the benchmark. Also, this research adjusts returns for risk, momentum and actual

transaction costs (as a function of shares bought and sold) to capture a more realistic alpha.
II. Literature Review

Hu (2014) shows that imbalances between option volume and underlying volume are

predecessors of future stock returns. Pan et al. (2006) also show that volume, for specific

traders, was determined to contain information about future prices. This latter study had access

to a unique data set that provided new buyer option contract volume (behavior) broken out by

various traders. Unique put-call ratios were derived by using each trader. The data (1990-2001)

was analyzed using a univariate regression, where the independent variables are the

corresponding put-call ratios, firm size and PIN, and the dependent variable is the next day risk-

adjusted return. The results showed stocks with low put-call ratios, derived from a trader (full-

service), outperformed stocks with high put-call ratios by 40 basis points, on the next day and

1% over the following week. The premise here is that informed, full-service (akin to the

customer trader described later in this article) investors trading the underlying stock instead of

index options are privy to firm-specific related news rather than market-wide news. This data

enables one to observe specific trader behavior and also to capture investor sentiment. 1 Using

indicators derived from index option data, Han (2004) showed the bull-bear spread (investor

intelligence), affect S&P index option prices and are conformist. Also, index options track the

performance of an entire index. Although, significant put-call imbalances blend themselves with

overall market-wide sentiment instead of a particular company sentiment. Table I lists some of

the most popular market data derived indicators. Much of the research above is focused on

abnormal returns before news releases. However, this does not preclude the uninformed investor

from participating, as information present in news diffuses into assets prices over a period of

time according to Chen et al. (2017), Hou (2007), Maheu et al. (2004), and Chan (2004), offering

1
Company specific indicators act contrarian in nature when they peak in value (Lamont et al. 2004; Tsuji 2009;
Simon et al. 2001)
investors potential trading gains that are also due to continuation, reversal, overreaction and

underreaction effects (Frazzini 2006).

Investor behavior differs under varying market conditions, as during bear markets.

Periods of extreme fear or high put-call ratios are prime buying opportunities due to substantial

levels of cash reserves. In bear periods like these, when a positive news is released, a massive

influx of money pours back into the markets (Simon et al. 2001), and asset prices experience

robust gains. Conversely, during bull markets, when a negative news is injected into a saturated

bull market, this can cause a substantial outflow of money causing significant price decreases.

This paper evaluates if various call-put ratio strategies, derived from trader option

volume, are proxies of investor sentiment and anticipate price changes. We use company-

specific call-put ratios computed from derivative data provided by the International Securities

Exchange (ISE) to determine bullish/bearish market directions. This ratio is like the put-call

ratio which helps explain pricing variations that are not captured by various fundamental factors

(Bandopadhyaya et al. 2011), implying that the put-call ratio is capturing a component of

investor behavior.

[TABLE I]

III. Data

We studied the S&P 500 constituents using the stock prices of the University of

Chicago’s Center for Research in Security Prices (CRSP) and options data provided by

International Securities Exchange Holdings (ISE Holdings) between May 2005 and December

2012. The ISE dataset consists of specific asset daily options volume data filtered by the

following: transaction type, trader types, and option type all used to compute the ISE call-put

ratio (Figure I).

 [FIGURE I]

Customer traders, for open buy order types on both call and put option types, dominate

option volumes, with 66.69% and 62.99%, respectively, for the entire available data set (see

Table II). Pan et al. (2006) showed that the full-service trader, akin to our customer trader, also

dominated the option volumes, and was shown to be the most reliable predictor of future gains.

[TABLE II]

We run the Bai-Perron (1998) regression test across the timeframe of our data set to

identify structural breaks in the S&P 500 exchange-traded fund (SPY ETF) series. The analysis

yielded the following three periods: before or the beginning of the financial credit crisis of the

late 2000s (May 2005 to May 27th, 2008), the financial crisis (May 27th, 2008 to September 13th.

2010) and the recovery period (September 13th, 2010 to December 31st, 2012).

The market experienced varying conditions: extreme, average, and low volatility. Figure

II compares the SPY ETF daily returns during the most volatile month (October 2008), and a

typical month (October 2005) to highlight the varying levels of volatility. Analysis of these

periods helps determine if any ISE call-put ratio contains more information during a specific sub-

period, reflective of market conditions, and naturally captures any trader’s shift in behavior

under varying conditions.

[FIGURE II]

One assumption for using the data from ISE is that neither exchange fragmentation nor

concentration exists, which affect the price discovery process. To lessen the perceived impact of

this assumption, O’Hara et al. (2011) showed that fragmentation did not cause an inefficient

marketplace. On the contrary, fragmentation increased competition between the exchanges and

helped induce competitive spreads and trade execution.

IV. Methodology

Following Pan et al. (2006), opening buy data for equity options was used to derive the

specific trader type daily call-put ratio by calculating the ISE call-put ratio (ISE ratio) utilizing

the opening buy order transaction types in the following formula:

LONG CALLS 𝑇𝑇𝑇𝑇 (Opening Position)

ISE = (1)
LONG PUTS 𝑇𝑇𝑇𝑇 (Opening Position)

where:
𝑇𝑇𝑇𝑇 = trader specific call volume
𝑇𝑇𝑇𝑇 = trader specific put volume

We used various option volume filters (total daily contract volume) to determine if

certain thresholds of option contract volume would yield a higher Sharpe ratio (Sharpe 1975).

Pan et al. (2006) used a minimum volume of 50 contracts to construct put-call ratios to prevent

small quantities from dictating the ratio values. We used a similar methodology but expanded it

to include volumes (number of option contracts) of 50, 100 and any value to formulate the call-

put ratios. Also, either the volume of put options or call options must meet the filter value to

qualify, not both. Using this filtering methodology captures varying levels of trade conviction

and their respective impacts on the Sharpe ratio.

In total, we used the ISE call-put ratio to define the following parameters for our

simulations:

Strategies:

- L ISE: Short stocks at or below the 10%, 20% and 30% percentile of the ISE call-put
ratio (L 10%, L 20%, and L 30%).
- H ISE: Long position for stocks at or above the 90%, 80% and 70% percentile of the ISE
call-put ratio (H 90%, H 80%, H 70%)
- HL ISE: Long/short position for stocks at or above (or below) the 90%/10%, 80%/20%
and 70%/30% percentiles respectively (HL 90-10%, HL 80-20%, HL 70-30%)
Filters: The strategies were separately simulated with different call and put option volume cut-
offs:
- 50: Option contract volume sizes >= 50
 - 100: Option contract volume sizes >= 100
- All: Any option contract

Traders:
- Customer: Option trade volume for traders acting on behalf of discount and full-service
customers.
- Broker-Dealer: Option trade volume for traders acting on behalf of institutional clients.
- Proprietary: Option trade volume for proprietary traders acting on behalf of their firm.
- Professional: Option trade volume for Non-Registered Broker Dealer traders whose daily
average is at least 390 trades (high-frequency traders). Information about professionals is
available only since October 2009.

Periods:
- May 2005 to December 31st, 2012
- May 2005 to May 27th, 2008
- May 27th, 2008 to September 13th, 2010
- September 13th, 2010 to December 31st, 2012

Assuming the ISE to be a conformist indicator, a low (L ISE) and a high ISE (H ISE)

values result in taking a short and long position equally invested at the closing bid and ask prices

respectively on the day after (t+1) the event day (t). Event day is when an asset’s ISE ratio

qualifies for each respective trading strategy. These short and long positions are liquidated at the

closing ask and bid prices respectively for the following day (t+2). We refer to this particular

strategy as the one-day long-short strategy.

The securities traded are constituents of the S&P500 index and dependent upon meeting

the specific simulation criteria: L ISE, H ISE, and HL ISE. We restricted our sample to the

S&P500 constituents as our benchmark is the S&P500 Composite Index and we wanted to avoid

the unstable behavior of small caps or outliers that may have occasional extraordinary results.

However, as the extreme values of the ISE index are associated with very liquid stocks, our tests

show that the results are less optimistic or similar than the simulations without the restriction of

the S&P500 constituents.

 The same securities are not traded every day as it depends upon the imbalance of call

volume to put volume to qualify the ISE call-put ratio for specific simulation criteria. Figure III

shows the average number of daily stocks selected by the long and short positions. In average,

344 stocks are daily selected across all the positions and the maximum number of stocks per

position selected are 73 for the customer’s H 70% position. Transaction costs consistent with the

NYSE (www.nyse.com) equity per share charge of 0.0023 were incurred on both entry and exit,

open and close, twice per security. Also, closing bid and ask prices were taken into account for

the respective position type: long entry on closing ask, short entry on closing bid, long exit on

next day closing bid and short exit on next day closing ask. All the prices were adjusted by

shares outstanding for most distribution events and dividend payments. Formula (2) shows the

return for the short positions which include the interest payment on margin requirement (50%)

using the one-month Treasury bill rate (from Ibbotson Associates) plus a spread (about 7.3%)

that approximates the difference between this rate and the margin rates offered by large

companies such as Fidelity.

Ask – dividends paid – bid – margin interest – transaction cost

Return (short) = (2)
Initial margin requirement

We calculated the annual Sharpe ratio using the next day and the next two to five days’

excess returns with respect to the S&P 500 Composite Index after an event day.

[FIGURE III]

We may have overestimated our transaction costs as we assumed that the traders only

submit market orders; however, if the orders are executed as limit orders, then they may receive

a rebate instead of paying a fee according to the price policy of several exchanges. The

additional cost of collecting the detailed information of the ISE signal could be avoided by

investors that approximate this ratio with the CBOE statistics.

 We also calculated alpha or excess return with respect to the three and four factors

proposed by Fama and French (1993) and by Carhart (1997) respectively.

V. Results

Before delving into the main simulation results, we performed regression tests to justify

why we use extreme call-put ratios in this research. We run a multiple linear regression across

the entire period from May 2005 to December 2012 for all the traders and volume levels. We

used the post-event day cross-sectional returns and four factors risk-adjusted returns as the

dependent variables and dummy variables for stocks at or above the top decile (top 10%) and at

or below the lowest decile (bottom 10%) of the ISE call-put ratio. Additionally, we included the

return of the five previous days (R-5,-1) to account for time series and short-term reversal effects

as in Lo and MacKinlay (1990) (Table III).

[TABLE III]

Comparing the three groups by trading volume, the simulations with at least 100

contracts show the most promising results. In this group, most of the top 10% and bottom 10%

stocks according to the ISE ratio traded by all traders show a positive and negative relationship

respectively with both return and risk-adjusted return at a 1% significance level. The return of

the past five trading days shows a significant impact for proprietary and professional traders. As

the past five trading days’ return is an indicator of short-term reversals, it would have a

significant impact on traders with a short investment horizon as it is the case of professional

traders who are considered high-frequency traders. These results suggest that extreme ISE ratios

help explain the variability of post-event day returns justifying our use of extreme ISE ratios

from this point forward in this research as part of a conformist one-day long-short strategy: take
a long or short position after a stock is classified into any of the H ISE or L ISE categories

respectively.

Looking at the results of the trading strategies in Table IV-a, the average of the H ISE (H

90%, H 80%, and H 70%) simulations by strategy and volume yields positive annual Sharpe

ratios while the L ISE and hybrid strategies do not perform well. The highest values are observed

in the simulations with at least 100 contracts, which support findings that volume contains

information about future stock returns (Campbell et al. 1992, Easley et al. 1998, Cao et al. 2003,

Pan et al. 2006 and Johnson et al. 2012). The small volume is most likely coming from smaller

market participants and large volume from larger market participants. Typically, the larger the

market participant, the more informed they are (Easley et al. 1998 and Barclay et al. 1993) as

they have more resources at their disposal. In our analysis, we will concentrate in the strategies

based on the top, and bottom deciles (H 90%, L 10%, and HL 90-10%) with at least 100 options

contracts for either puts or calls as the H 90% strategy outperforms the rest.

[TABLE IV]

An essential question of this part of our analysis is to understand why the short strategies

(L ISE) did not perform well when the dummy variable of the bottom 10% stocks had a negative

and significant coefficient in the regression analysis. We believe that implementing our trading

strategy for each of the following post-event five days helps to answer this question. Table V-a

shows that after day 1, there is a short-term reversal effect and the short and long positions have

positive and negative results respectively. An aggregated position from day 2 to day 5 (t+2 - t+5)

still leads to similar results. It is also possible that the impact of the low (L) ISE ratio is felt an

extra day after the high (H) ISE ratio. Considering either the reversal effect or the extra reaction

time of the L ISE ratio, we implemented a variation of our trading strategy delaying the

implementation of our short position one day while the long position remains the same. The
combined portfolio (HL ISE) would be the result of taking a long position only for the next day

(t+1) after the stocks are included in the H ISE groups, and taking a short position only two days

(t+2) after the stocks are included in the L ISE groups. We refer to this strategy as the two-days

long-short strategy (H: t+1, L: t+2). Both positions are maintained only for one day. The

implementation of this strategy in the first two days leads to a positive result for the H 90%, L

10% and the HL 90-10% portfolios of the broker-dealer, proprietary and professional traders as

Tables IV-b and V-b show. The extension of this strategy in the following four days (H: t+2, L:

t+3 … H: t+5, L: t+6), leads to the observed short-term reversal effect.

[TABLE V]

In Tables VI-a and VI-b we disaggregate these results by traders and periods and the two

group of strategies. The results are still consistent with the previous results and with the

regression analysis. The simulations with at least 100 contracts show better results than the rest.

Even more, the negative performance of the H 90% strategy of the broker-dealers traders with a

minimum of 50 contracts go away when the requirement increases to 100. Likewise, the

performance of the professional traders improves substantially in the same circumstances and

become the dominant trading group followed by the proprietary traders, and then by the broker-

dealers. The dominance of the professional traders, who are considered high-frequency traders

due to their high volume of daily transactions, can be explained as their fast reaction to any

investment opportunity as indicated by the ISE signal.

Additionally, all the traders but the customers show the best performance during the crisis

period (May 2008 – September 2010) and in less degree during the recovery period. The fact that

the results of the customers become positive during the recovery period indicates that the ISE

ratio is especially important in periods of high volatility such as during the financial crisis of the

late 2000s. In general, the two-days long-short strategy outperforms the one-day long-short
strategy and leads to positive and significant results for most of the cases even for the short (L

ISE) and the hybrid positions (HL ISE). As expected, the results of the hybrid strategies (HL

ISE) are equivalent to the combined performance of the L ISE and H ISE strategies for each

trader.

[TABLE VI]

We extended our analysis to control for the three factors of the Fama and French (1993)

model and the four factors of the Carhart (1997) model applied only to the two-days long-short

strategy as the one-day long-short strategy would have similar patterns generating negative

results for the short positions and some of the hybrid portfolios. The Carhart model includes the

three factors of the Fama-French model and the incremental impact of the momentum factor (see

Table VII). We also included different deciles to evaluate our strategies with a smaller and larger

group of stocks.

Simulation results further validate our previous results. In most of the cases, there are no

significant differences between the results of these two-factor models. Additionally, the

relaxation of the thresholds that leads to an increase in the number of stocks included in the top

and bottom groups (see Figure III) reduces the alpha values and the number of cases with

positive risk-adjusted alphas.

All the traders show significant positive alpha, especially during the crisis and recovery

period. The performance of the professional traders gets closer to the performance of the

proprietary and broker-dealers. This result confirms the value of information of these traders

especially during periods of high volatility while the customer traders’ transactions are more

informative during periods of low volatility, especially as they dominate the option volume for

both transactions type and option type (see Tables II and X).

[TABLE VII]
VI. Professional Trader

Professional trader data was available as of October 2009. These traders make at least

390 trades per day; therefore, we consider them to be high-frequency traders (HFT). Kirilenko et

al. (2017) in the study of the flash crash of May 6, 2010, define HFT as the top 7% transactional

traders on a daily basis. HFTs will close positions before market close. Additionally, HFTs are

not chasing long-term trends; rather they are chasing short-term trends and provide liquidity for

the marketplace.

An additional analysis was run to determine if an ISE call-put ratio derived from the

professional trader can be used as a proxy for future asset returns. Even though we have included

partial information of the professional traders in the previous tables, in this section, we

concentrate only in the period when their data was available (October 2009 to December 2012).

Professional traders show a positive and statistically significant alpha since October 2009

and during the recovery period across all the strategies (see Table VIII). Additionally, they also

show a positive Sharpe ratio based on excess return during the same periods and across the

different strategies but the L 10% strategy that is not significantly different from zero during the

crisis period (see Table IX).

[TABLE VIII]
[TABLE IX]

Comparing the trader composition (Tables II and X), one can see that customer traders

dropped from 63% to 58.5%. Post-February 2009 was a time when many customer traders were

reluctant to step back into the market, whereas market makers and professional traders, who

make a living off trading, continued to make a market and trade in any market condition. In the

current study, the ISE ratio is based on a daily aggregate of option volumes. Therefore, it would
not capture the intraday trading behavior of a professional-HFT trader. However, it shows the

informational value of the professional traders’ options volume.

[TABLE X]

VII. Discussion

A major question in the literature is the reversal effect of any news or major event. In our

study, the main event is when an asset’s ISE ratio qualifies for each respective trading strategy,

and our test is based on the return of the following two to five days as presented in Tables V-a

and V-b. As we already indicated, there is a short-term reversal effect, as most of the strategies

show a change of sign of their results or they become less profitable as the horizon increases. L

10% is the only strategy that still shows positive result from the second to the fifth day

implemented as part of the two-day long-short strategy. The long (H 90%) and hybrid strategy

(HL 90-10%) show negative results in most of the cases. We obtain similar results using the rest

of deciles introduced in this study, although we did not include them in the table to simplify our

presentation.

The results discussed above seem to indicate the presence of a short-term momentum

effect and that information is not instantly absorbed into an asset price but diffuses over time.

Hong et al. (1997) showed that two market participants, ‘news watchers’ and ‘momentum

traders,' attribute the most to diffusion of price information. This delay provides momentum

traders with an opportunity to trade once the news is injected into the marketplace (Jegadeesh et

al. 1993, Hong et al. 2000, Jegadeesh et al. 2001, Pan et al. 2006). Jegadeesh (1993) showed

reversals over months (before the mass adoption of the internet), and with this research, it is

observed that significant reversal effects are compressed to a much shorter time horizon.
 We are not assuming all investor types are equally informed because of recent

technological advances. However, information is absorbed faster into assets prices through

momentum effects, as the largest number of trades are attributable to customer traders. Also,

looking at the overall average cross-sectional returns, all strategies (average of all traders)

experienced reversals. The reversals appear to support past research which showed an

overreaction to news from both Barberis et al. (1998) and Daniel et al. (1998). Our signal was

not directly derived from a news event like the research above; however, abnormal option

volume is a predecessor to asset price volatility (Campbell et al. 1992, Easley et al. 1998, Cao et

al. 2005, Pan et al. 2006 and Johnson et al. 2012). It can be postulated that high and low call-put

ratios can be produced by informed investors taking a larger position on one side, put or call. It

appears that the professional traders take advantage of momentum effects in the very short-term.

We find that the proprietary trader, jointly with the professional trader, exhibited substantial next

day gains in the different time periods. In contrast to the results of Pan et al. (2006) who

conjectured that the proprietary trader was hedging and closing positions on the same day, this

trader may have implemented more sophisticated strategies with stronger option volume

requirements than those postulated by Pan et al. (2006). Since this original study was published,

electronic and algorithmic trading has become a norm, and financial information has become

highly accessible through so many different sources that the position of sophisticated traders tend

to converge and explore new directions as we observe in this study.

It appears that the change of direction observed with the L ISE strategy is the result of

underreaction seen by both Hong et al. (1999) and Daniel et al. (1998). The information

provided to the traders to determine trades was not fully absorbed on event day, as seen by the

continuation effects and it seems that the proprietary and professional traders take advantage of
this circumstance selecting some of the most relevant stocks that move the market (see Table XI

for a sample of top chosen stocks).

[TABLE XI]

VIII. Conclusion

This paper shows that on a short-term horizon, following a conformist strategy according

to our original assumption, a trader could benefit from a two-days long-short strategy based on

high and low ISE ratios respectively. Hence, the ISE ratio acts like the put-call ratio which may

explain pricing variations that cannot be explained by several fundamental factors

(Bandopadhyaya et al. 2011), implying that it reflects other aspects of investors’ behavior.

Positive alpha, for both the three-factors Fama-French and the four-factors Carhart model, was

seen across all traders for the two-day long-short strategy, which further reinforces our original

assumption.

The predictive power of the H ISE and L ISE observed when we tested our trading

strategies is consistent with our initial regression analysis (see Table III) which yielded statistical

significance for the positive and negative coefficients associated with the dummy variables that

reflected the top and bottom 10% set of stocks as defined by the ISE ratio. This trading signal,

according to our simulations, handily beats the S&P 500 composite index and supports the

conformist assumption that L and H ISE signals should lead to a short and long position

respectively. However, as we also explored in this paper, the use of these indicators requires

studying the appropriate horizon for each signal as the short-term effect and the reversal effects

of the L ISE and H ISE signals might be different.

In the long run, markets generally do go up, which partially explains the short-term

reversal effects. It is possible that beyond the one or two-day horizon, the L ISE and H ISE
signals can both be used as short-term indicators for the qualifying stocks due to the reversal

effects as we explored in this paper.

These results support the assertion that various call-put ratio strategies derived from

trader option volume are proxies of investor sentiment as we do see significant reversal effects

when used in trading strategies. Future research will involve a close look at stocks whose ISE

values had substantial deviations away from expected ISE values. Also, the signal derived from

the ISE data will be combined with sentiment extracted from social media to determine if the two

features help explain price diffusion more than any stand-alone version. Lastly, we could also

extend this analysis to intraday data capturing a shorter-term price indicator. This will be an area

to explore for future research.

Acknowledgements

The authors thank Axel Vischer from Eurex, Jeffrey Soule and International Securities Exchange

Holdings for providing the data and for discussing initial versions of this research. The authors

also thank David Starer, Hamed Ghoddusi, and participants of the 5th. High-Frequency data

conference at Stevens Institute of Technology for suggestions and informal discussions about

this paper; Patrick Jardine for proof-reading the article, and to the Stevens Alliance for

Innovation for partially funding this work. The opinions presented are the exclusive

responsibility of the authors.

References

Agrawal, Anup, and Tommy Cooper. ‘Insider Trading Before Accounting Scandals.’ Journal of
Corporate Finance, vol. 34, issue C, (2015): 169-190

Arms, Richard. ‘The Arms Index (Trin Index): An Introduction to Volume Analysis.’
Marketplace Books, Columbia, MD. (1996).

Bai, Jushan, and Pierre Perron. ‘Estimating and Testing Linear Models with Multiple Structural
Changes.’ Econometrica, 66, no. 1 (1998): 47–78.

Bandopadhyaya, Arindam, and Anne Leah Jones. ‘Measures of Investor Sentiment: A

Comparative Analysis Put-Call Ratio Vs. Volatility Index.’ Journal of Business & Economics
Research, 6, no. 8 (2011): 27-34.

Barberis, Nicholas, Andrei Shleifer, and Robert Vishny. ‘A Model of Investor Sentiment.’
Journal of Financial Economics, 49, no. 3 (1998): 307–43.

Barclay, Michael J., and Jerold B. Warner. ‘Stealth Trading and Volatility: Which Trades Move
Prices?’ Journal of Financial Economics 34, no. 3 (1993): 281–305.

Batchelor, Roy, and Pami Dua. ‘Improving Macro-Economic Forecasts: The Role of Consumer
Confidence.’ International Journal of Forecasting, 14, no. 1 (1998): 71–81.

Billingsley, Randall S, and Don M Chance. ‘Put—call Ratios and Market Timing Effectiveness.’
The Journal of Portfolio Management, 15, no. 1 (1988): 25–28.

Campbell, John Y., Sanford J. Grossman, and Jiang Wang. ‘Trading Volume and Serial
Correlation in Stock Returns.’ The Quarterly Journal of Economics, 108, no. 4 (1993): 905–
39.

Cao, Charles, John M. Griffin, and Zhiwu Chen. Informational Content of Option Volume Prior
to Takeovers. The Journal of Business, 78, no. 3, (2005): 1073-1109

Carhart, Mark M. ‘On Persistence in Mutual Fund Performance.’ The Journal of Finance, 52, no.
1 (1997): 57–82.

Chan, Wesley S., Richard Frankel, and S.P. Kothari. ‘Testing Behavioral Finance Theories
Using Trends and Consistency in Financial Performance.’ Journal of Accounting and
Economics, 38 (2004): 3–50.

Chen, Yubo, and Jinhong Xie. ‘Online Consumer Review: Word-of-Mouth as a New Element of
Marketing Communication Mix.’ Management Science, 54, no. 3 (2008): 477–91.

Chen, Zhuo, and Andrea Lu. Slow Diffusion of Information and Price Momentum in Stocks:
Evidence from Options Markets. Journal of Banking and Finance, vol. 75(C), (2017): 98-108.
Christophe, Stephen E., Michael G. Ferri, and Jim Hsieh. ‘Informed Trading before Analyst
Downgrades: Evidence from Short Sellers.’ Journal of Financial Economics, 95, no. 1 (2010):
85–106.

Clarke, Roger G., and Meir Statman. ‘Bullish or Bearish?’ Financial Analysts Journal, 54, no. 3
(1998): 63–72.

Cochrane, John H. Asset Pricing. Princeton University Press, Princeton, NJ, 2009.

Daniel, Kent, David Hirshleifer, and Avanidhar Subrahmanyam. ‘Investor Psychology and
Security Market under- and Overreactions.’ The Journal of Finance, 53, no. 6 (1998): 1839–
85.

Diamond, Douglas W., and Robert E. Verrecchia. ‘Constraints on Short-Selling and Asset Price
Adjustment to Private Information.’ Journal of Financial Economics, 18, no. 2 (1987): 277–
311.

Easley, David, Robert F. Engle, Maureen O’Hara, and Liuren Wu. ‘Time-Varying Arrival Rates
of Informed and Uninformed Trades.’ Journal of Financial Econometrics, 6, no. 2 (2008):
171–207.

Easley, David, Soeren Hvidkjaer, and Maureen O'Hara. ‘Is information risk a determinant of
asset returns?’ The Journal of Finance, 57 no. 5 (2002): 2185-2221.

Easley, David, Maureen O’Hara, and P.S. Srinivas. ‘Option Volume and Stock Prices: Evidence
on Where Informed Traders Trade.’ The Journal of Finance, 53, no. 2 (1998): 431–65.

Easley, David, Nicholas M. Kiefer, Maureen O’Hara, and Joseph B. Paperman. ‘Liquidity,
Information, and Infrequently Traded Stocks’ The Journal of Finance, 51, no. 4 (1996): 1405–
1436.

Fama, Eugene F. ‘Efficient Capital Markets: A Review of Theory and Empirical Work.’ The
Journal of Finance 25, no. 2 (1970): 383-417.

Fama, Eugene F., Lawrence Fisher, Michael C. Jensen, and Richard Roll. ‘The Adjustment of
Stock Prices to New Information.’ International Economic Review, 10, no. 1 (1969): 1-21.

Fama, Eugene F., and Kenneth R. French. ‘Common Risk Factors in the Returns on Stocks and
Bonds.’ Journal of Financial Economics, 33, no. 1 (1993): 3–56.

Frazzini, Andrea. ‘The Disposition Effect and Underreaction to News.’ The Journal of Finance,
61, no. 4 (2006): 2017–46.

Han, Bin. ‘Limits of Arbitrage, Sentiment and Pricing Kernel: Evidences from Index Options’
Working Paper Series 2004-2, Ohio State University, Charles A. Dice Center for Research in
Financial Economics (2004).
Hong, Harrison, and Jeremy C. Stein. ‘A Unified Theory of Underreaction, Momentum Trading,
and Overreaction in Asset Markets.’ The Journal of Finance, 54, no. 6 (1999): 2143–84.

Hou, Kewei. ‘Industry Information Diffusion and the Lead-Lag Effect in Stock Returns.’ Review
of Financial Studies, 20, no. 4 (2007): 1113–38.

Hu, Jianfeng. ‘Does Option Trading Convey Stock Price Information?’ Journal of Financial
Economics, 111, no. 3 (2014): 625–45.

Jegadeesh, Narasimhan, and Sheridan Titman. ‘Returns to Buying Winners and Selling Losers:
Implications for Stock Market Efficiency.’ The Journal of Finance, 48, no. 1 (1993): 65–91.

Jegadeesh, Narasimhan, and Sheridan Titman. ‘Profitability of Momentum Strategies: An

Evaluation of Alternative Explanations.’ The Journal of Finance, 56, no. 2 (2001): 699–720.

Johnson, Travis L., and Eric C. So. ‘The Option to Stock Volume Ratio and Future Returns.’
Journal of Financial Economics, 106, no. 2 (2012): 262–86.

Karpoff, Jonathan M., and Xiaoxia Lou. ‘Short Sellers and Financial Misconduct.’ The Journal
of Finance, 65, no. 5 (2010): 1879–1913.

Kerrigan, Thomas J. ‘The Short Interest Ratio and Its Component Parts.’ Financial Analysts
Journal 30, no. 6 (1974): 45–49.

Kirilenko, Andrei A., Albert S. Kyle, Mehrdad Samadi, and Tugkan Tuzun. The Flash Crash:
The Impact of High Frequency Trading in an Electronic Market. The Journal of Finance, 72,
no. 3 (2017): 967-998.

Kumar, Manmohan S., and Avinash Persaud. ‘Pure Contagion and Investors’ Shifting Risk
Appetite: Analytical Issues and Empirical Evidence.’ International Finance, 5, no. 3 (2002):
401–36.

Lamont, Owen A., and Jeremy C. Stein. Aggregate Short Interest and Market Valuations.
American Economic Review, 94, (2004): 29-32

Lashgari, Malek. ‘Information Content of U.S. Treasury Inflation-Indexed Bonds.’ International

Advances in Economic Research 6, no. 3 (2000): 520–30.

Lo, Andrew, and A. Craig MacKinlay. ‘When are Contrarian Profits due to Stock Market
Overreaction?’ Review of Financial Studies, 3, no. 2, 1990: 175–205.

Lynch, Peter, and John Rothchild. One Up On Wall Street: How To Use What You Already Know
To Make Money In The Market. Simon and Schuster, NY, NY, 2000.

Maheu, John M., and Thomas H. McCurdy. ‘News Arrival, Jump Dynamics, and Volatility
Components for Individual Stock Returns.’ The Journal of Finance, 59, no. 2 (2004): 755–93.
Mullainathan, S. and Andrei Shleifer. The market for news. American Economic Review, 95 no.
1 (2005): 1031-1053.

O’Hara, Maureen, and Mao Ye. ‘Is Market Fragmentation Harming Market Quality?’ Journal of
Financial Economics 100, no. 3 (2011): 459–74.

Pan, Jun, and Allen M. Poteshman. ‘The Information in Option Volume for Future Stock Prices.’
Review of Financial Studies, 19, no. 3 (2006): 871–908.

Sharpe, William F. ‘Adjusting for Risk in Portfolio Performance Measurement.’ The Journal of
Portfolio Management, 1, no. 2 (1975): 29–34.

Simon, David P., and Roy A. Wiggins. ‘S&P Futures Returns and Contrary Sentiment
Indicators.’ Journal of Futures Markets, 21, no. 5 (2001): 447–462.

Tsuji, Chikashi. ‘Are Investment Strategies Exploiting Option Investor Sentiment Profitable?
Evidence from Japan.’ International Journal of Business and Management, 4, no. 5 (2009):
92-105.

Whaley, Robert E. ‘The Investor Fear Gauge.’ The Journal of Portfolio Management, 26, no. 3
(2000): 12–17.

Vermeulen, Ivar E., and Daphne Seegers. ‘Tried and Tested: The Impact of Online Hotel
Reviews on Consumer Consideration.’ Tourism Management, 30, no. 1 (2009): 123–27.

Zhu, Feng, and Michael Zhang. ‘Impact of Online Consumer Reviews on Sales: The Moderating
Role of Product and Consumer Characteristics,’ Journal of Marketing, 74, No. 2, (2010): 133-
148.
Table I. Market Data Derived Indicators.

Indicator Definition
VIX is a measure of implied volatility of front month and second month expiration
Volatility Index (VIX) out-of-the-money put and call options. Calculated every minute. Known as a fear
factor gauge (Whaley, 2000). High volatility is indicative of uncertainty.
Put option volume divided by Call option volume. Billingsly & Chance (1988)
showed that values over 0.7 are bearish signals and values under 0.7 are bearish
Put-call ratio (PCR)
signals. Similar to the reciprocal of the ISE ratio used in this research, but the PCR
uses all option volume data, regardless of trader or trade size.
Weekly survey of 150 newsletters marked as bullish, bearish or neutral based on
Bull-Bear Spread (IIS) future expected returns. Clarke & Statman (1998) showed as author sentiment
shifts from bearish to bullish is a proxy for future significant returns.
Average yield on high-grade bonds divided by average yield of intermediate grade
Barron’s Confidence Index (BCI)
bonds. Lashargi (2000) leverages this ratio as a market confidence indicator.
Based on rank correlation across assets riskiness and excess returns. Kumar &
Risk Appetite Index (RAI) Persaud (2002) showed that investors risk appetite changes over time and is
measurable. > 0 positive, < 0 negative
Survey polling ~500 people monthly about their opinion of the near-term health of
the economy. This index could have been used to predict the recession of 1991
Consumer Confidence Survey (CCI)
(Batchelor & Dua, 1998). Benchmarked to 100%. > 100% optimism < 100%
pessimism
Arms 1989 showed an oversold market equates to volume in declining stocks that
far outweigh volume in advancing stocks. Short-term indicator = (advancing
Traders Index (TRIN)
issues/declining issues) / (advancing volume/declining volume). < 1 bullish, > 1
bearish
Total number of shares shorted, in term of percentage of total outstanding shares,
Short Interest (SIR) high % bearish, low % bullish. Kerrigan (1974) showed a correlation between SIR
and asset returns.

Note: This table highlights favorite market data derived sentiment indicators used to both gauge
market conditions and commonly used as trading signals.
Table II. Trader Composition.

May 2005 to December 2012

Open Buy Open Sell Close Buy Close Sell
Put Call Put Call Put Call Put Call
Average Volume 41.28 45.56 38.42 39.16 14.46 15.33 16.63 21.44
Proprietary 22.97% 21.46% 25.09% 22.50% 18.77% 15.84% 13.89% 12.52%
Customer 62.99% 66.69% 58.70% 64.21% 74.20% 78.64% 80.54% 83.09%
Broker Dealer 14.03% 11.85% 16.21% 13.29% 7.02% 5.52% 5.57% 4.39%
Note: This table shows each trader average put and call volume (option contracts) for all four
order types. Also, we break the data further out into percentage breakdowns to see what trader
is driving most of the volume. The customer trader dominates the options volume for both
transaction type and option type. The composition breakout is only using the data provided by
ISE and is only trade data for their exchange which constituted 29.6% market share at the
beginning of our period, May 2005, and 16.75% market share by the end December 2012.
Table III. Regression Test Results.
Return 4 Factors Risk Adjusted Return (Carhart 1997)
Volume Customer Broker Dealer Proprietary Professional Customer Broker Dealer Proprietary Professional
All Intercept -0.049 -8.586 *** -0.04 -7.3728 *** 0.0025 0.4267 -0.057 -10.49 *** 0.0031 0.634 0.0081 1.7646 * 0.0269 5.4805 *** -7E-05 -0.0145
R-5,-1 -4E-05 -1.435 -4E-05 -1.3728 -4E-05 -1.4866 -4E-05 -1.441 -3E-05 -1.2138 -3E-05 -1.169 -3E-05 -1.2431 -3E-05 -1.2194
Top 10% 0.1256 5.889 *** 0.3218 7.2969 *** 0.3453 14.83 *** 0.3223 8.1867 *** 0.0854 4.7501 *** 0.2622 7.0547 *** 0.2634 13.408 *** 0.2174 6.5513 ***
Bottom 10% -0.1787 -9.254 *** -0.353 -15.365 *** -0.6146 -38.435 *** -0.102 -3.842 *** -0.1059 -6.5111 *** -0.213 -11.02 *** -0.3302 -24.475 *** -0.0946 -4.2267 ***
Adjusted R2 0.0003 0.0047 0.0002 0.0022
>=50 Intercept -0.0457 -2.744 *** 0.2869 5.70394 *** 0.1147 5.6604 *** 0.1307 4.7766 *** 0.0089 0.6378 0.1455 3.4273 *** 0.0653 3.9215 *** 0.0319 1.3299
R-5,-1 -1E-05 -0.146 -0.008 -1.8868 * -0.0038 -5.5902 *** -0.077 -17.16 *** -1E-05 -0.238 -6E-05 -0.018 -0.0027 -4.9389 *** -0.062 -15.693 ***
Top 10% 0.098 1.522 -0.152 -0.9094 0.1956 2.7486 *** -0.012 -0.148 0.0121 0.2235 -0.08 -0.566 0.1417 2.4246 ** 0.0063 0.0886
Bottom 10% -0.2213 -3.745 *** -0.764 -8.7654 *** -0.7228 -16.619 *** -0.373 -5.461 *** -0.1293 -2.6085 *** -0.45 -6.115 *** -0.364 -10.189 *** -0.2259 -3.7694 ***
Adjusted R2 0.0004 0.0095 0.0001 0.004
>=100 Intercept -0.059 -(6.42) *** 0.106 (4.33) *** 0.025 (2.07) ** 0.029 (1.41) -0.005 -(0.66) 0.073 (3.50) *** 0.021 (2.09) ** -0.024 -(1.38)
R-5,-1 0.000 -(0.78) 0.000 (0.19) -0.004 -(7.22) *** -0.024 -(7.05) *** 0.000 -(0.66) 0.000 (0.06) -0.003 -(5.24) *** -0.021 -(7.23) ***
Top 10% 0.139 (5.17) *** 0.199 (3.09) *** 0.358 (11.40) *** 0.327 (6.40) *** 0.082 (3.65) *** 0.143 (2.63) *** 0.212 (8.18) *** 0.237 (5.35) ***
Bottom 10% -0.243 -(5.05) *** -0.503 -(10.97) *** -0.783 -(22.68) *** -0.312 -(4.51) *** -0.133 -(3.29) *** -0.342 -(8.86) *** -0.458 -(16.12) *** -0.155 -(2.58) ***
Adjusted R2 2.9E-04 6.2E-03 1.3E-04 3.1E-03

Note: This table shows the results of the daily cross-sectional regression for all traders with
different volume of contracts for the entire period, May 2005 to December 2012. Professionals’
data is since October 2009. The dependent variables are the post-event daily return and four
factors risk-adjusted returns according to Carhart (1997). The independent variables are dummy
variables for stocks at or above the top decile (Top 10%) and at or below the lowest decile
(Bottom 10%) of the ISE call-put ratio to ascertain if extreme call-put ratios explain the
variability of the post-event day return. Return of the five previous days (R-5,-1) is also an
independent variable to evaluate if there is short-term price reversal. The first number is the
coefficient estimate, the number in parenthesis is the t-value, and *, ** and *** represent
statistical significance at the 10%, 5%, and 1% levels, respectively.
Table IV. Average Annual Sharpe Ratios by Strategy and Volume.

H 90% H 80% H 70% L 10% L 20% L 30% HL 90-10% HL 80-20% HL 70-30%

>0 0.2 0.15 0.09 -0.33 -0.32 -0.3 -0.18 -0.12 -0.1
>=50 0.1 0.06 0.01 -0.38 -0.34 -0.33 -0.25 -0.17 -0.15
>=100 0.24 0.21 0.16 -0.45 -0.43 -0.39 -0.07 0.01 0.01

(a) One-day long-short strategy: H and L: t+1

H 90% H 80% H 70% L 10% L 20% L 30% HL 90-10% HL 80-20% HL 70-30%

>0 0.2 0.15 0.09 0.15 0.14 0.14 0.15 0.14 0.11
>=50 0.1 0.06 0.01 0.15 0.16 0.17 0.12 0.11 0.08
>=100 0.24 0.21 0.16 0.12 0.14 0.15 0.18 0.19 0.16

(b) Two-days long-short strategy: H: t+1 and L: t+2

Note: Annual Sharpe ratios across strategies and volumes for the entire period, May 2005 to
December 2012 for the (a) one-day long-short strategy and (b) the two-days long-short strategy:
long and short position initiated one and two days after the event day, respectively. Event day is
when an asset’s ISE ratio qualifies for each respective trading strategy. Professionals’ data is
since October 2009. Sharpe ratios are annualized averages of monthly ratios calculated every 21
days using excess returns with respect to the S&P 500 composite index and adjusted for
transaction costs. t+n represents n days after the event day (t).
Table V. Average Annual Sharpe Ratios by Trader and Period: post-event 1-5 days.
Trader/ t+1 t+2 t+3 t+4 t+5 t+2 - t+5
Strategy H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10%
Customer -0.03 -0.21 *** -0.06 *** -0.01 0.05 0 -0.06 *** 0.04 -0.04 * -0.14 *** 0.19 *** -0.07 *** -0.19 *** 0.15 *** -0.11 *** -0.24 *** 0.07 -0.01
Broker Dealer 0.26 *** -0.29 *** -0.13 *** -0.08 0.15 *** 0.09 *** -0.26 *** 0.05 -0.04 -0.2 *** 0.07 * -0.01 -0.21 *** 0.12 *** 0.03 -0.35 *** 0.08 * 0.08 **
Proprietary 0.33 *** -0.76 *** -0.15 *** -0.08 *** 0.19 *** 0.03 -0.18 *** 0.14 *** -0.03 -0.22 *** 0.1 *** -0.08 *** -0.17 *** 0.18 *** -0.01 -0.37 *** 0.18 *** 0.15 ***
Professional 0.39 *** -0.55 *** 0.08 -0.11 * 0.11 -0.04 -0.17 ** 0.11 -0.08 -0.24 *** 0.15 * -0.11 ** -0.13 ** 0.15 ** -0.04 -0.32 *** 0.12 0.02

(a). One-day long-short strategy: H and L: t+1

Trader/ H: t+1, L: t+2 H: t+2, L: t+3 H: t+3, L: t+4 H: t+4, L: t+5 H: t+5, L: t+6
Strategy H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10%
Customer -0.03 0.05 -0.01 -0.01 0.04 0 -0.06 *** 0.19 *** -0.01 -0.14 *** 0.15 *** -0.08 *** -0.19 *** 0.09 ** -0.13 ***
Broker Dealer 0.26 *** 0.15 *** 0.19 *** -0.08 0.05 0.01 -0.26 *** 0.07 * -0.03 -0.2 *** 0.12 *** 0.04 -0.21 *** 0.1 *** 0.01
Proprietary 0.33 *** 0.19 *** 0.26 *** -0.08 *** 0.14 *** 0.02 -0.18 *** 0.1 *** -0.05 ** -0.22 *** 0.18 *** -0.04 * -0.17 *** 0.08 ** -0.05 **
Professional 0.39 *** 0.11 0.29 *** -0.11 * 0.11 -0.04 -0.17 ** 0.15 * -0.06 -0.24 *** 0.15 ** -0.11 ** -0.13 ** 0.03 -0.07

(b). Two-days long-short strategy: H: t+1 and L: t+2

Note: This table shows the annual Sharpe ratios across all periods, and traders with a minimum
volume of 100 contracts for (a) the one-day long-short strategy and (b) the two-days long-short
strategy: long and short position initiated one and two days after the event day, respectively.
Event day is when an asset’s ISE ratio qualifies for each respective trading strategy. Sharpe
ratios are annualized averages of monthly ratios calculated every 21 days using excess returns
with respect to the S&P 500 composite index and adjusted for transaction costs. Professionals’
data is since October 2009. t+n represents n days after the event day (t). t+2 – t+5 refers to the
Sharpe ratio using accumulated excess return from the second to the fifth day after an event day.
*, **, and *** represent statistical significance of a t-test of 0 at 10%, 5% and 1% respectively.
Table VI. Average Annual Sharpe Ratios by Trader, Period, and Volume.

Volume Trader/ May 2015 - December 2012 May 2005 - May 2008 May 2008-Sept 2010 Sept 2010 - December 2012
Strategy H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10%
All Customer -0.03 -0.11 *** -0.07 *** -0.15 *** 0.04 -0.06 ** -0.05 -0.06 * -0.06 ** 0.12 ** -0.28 *** -0.13 ***
Broker Dealer 0.27 *** -0.29 *** -0.17 *** 0.13 -0.13 *** -0.09 ** 0.31 *** -0.33 *** -0.18 *** 0.32 *** -0.44 *** -0.28 ***
Proprietary 0.27 *** -0.65 *** -0.37 *** 0.37 *** -0.43 *** -0.19 *** 0.3 *** -0.75 *** -0.42 *** 0.16 *** -0.72 *** -0.46 ***
Professional 0.29 *** -0.27 *** -0.1 *** 0.41 *** -0.28 *** -0.05 0.2 *** -0.25 *** -0.12 ***
>= 50 Customer -0.12 -0.14 * -0.13 ** -0.33 ** -0.17 -0.26 *** -0.04 -0.06 -0.06 0.05 -0.18 -0.09
Broker Dealer 0.23 -0.38 *** -0.28 *** -0.13 0.06 0.04 0.41 -0.54 *** -0.36 ** 0.06 -0.75 *** -0.59 ***
Proprietary 0.18 ** -0.61 *** -0.41 *** 0.12 -0.29 *** -0.22 *** 0.42 ** -0.73 *** -0.47 *** 0.07 -0.7 *** -0.45 ***
Professional 0.11 -0.39 *** -0.19 *** 0.26 -0.26 * -0.03 0.08 -0.46 *** -0.24 ***
>= 100 Customer -0.03 -0.21 *** -0.06 *** -0.13 *** -0.06 -0.12 *** -0.02 -0.28 *** -0.08 * 0.15 *** -0.26 ** 0.06
Broker Dealer 0.26 *** -0.29 *** -0.13 *** 0.14 -0.18 *** -0.1 ** 0.32 *** -0.41 *** -0.19 *** 0.37 *** -0.26 *** -0.09
Proprietary 0.33 *** -0.76 *** -0.15 *** 0.44 *** -0.59 *** -0.08 * 0.28 *** -0.89 *** -0.21 *** 0.2 *** -0.81 *** -0.19 ***
Professional 0.39 *** -0.55 *** 0.08 0.62 *** -0.65 *** 0.22 ** 0.21 *** -0.46 *** -0.01
(a). One-day long-short strategy: H and L: t+1
Volume Trader/ May 2015 - December 2012 May 2005 - May 2008 May 2008-Sept 2010 Sept 2010 - December 2012
Strategy H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10%
All Customer -0.03 0.18 *** 0.08 *** -0.15 *** 0.24 *** 0.06 ** -0.05 0.15 *** 0.06 ** 0.12 ** 0.11 *** 0.11 ***
Broker Dealer 0.27 *** 0.12 *** 0.16 *** 0.13 0.22 *** 0.21 *** 0.31 *** 0.12 *** 0.16 *** 0.32 *** 0.07 0.13 ***
Proprietary 0.27 *** 0.18 *** 0.21 *** 0.37 *** 0.22 *** 0.27 *** 0.3 *** 0.12 *** 0.18 *** 0.16 *** 0.22 *** 0.2 ***
Professional 0.29 *** 0.1 *** 0.16 *** 0.41 *** 0.06 0.18 *** 0.2 *** 0.12 *** 0.14 ***
>= 50 Customer -0.12 0.08 0 -0.33 ** 0.24 ** -0.02 -0.04 0.05 0.03 0.05 0.01 0.01
Broker Dealer 0.23 0.17 ** 0.15 ** -0.13 0.33 ** 0.24 ** 0.41 -0.02 0.08 0.06 0.12 0.12
Proprietary 0.18 ** 0.17 *** 0.18 *** 0.12 0.3 *** 0.24 *** 0.42 ** 0.06 0.13 * 0.07 0.25 *** 0.21 ***
Professional 0.11 0.18 *** 0.15 ** 0.26 0.1 0.14 0.08 0.29 *** 0.19 **
>= 100 Customer -0.03 0.05 -0.01 -0.13 *** 0.14 * -0.06 -0.02 -0.03 -0.02 0.15 *** 0.12 0.14 ***
Broker Dealer 0.26 *** 0.15 *** 0.19 *** 0.14 0.19 *** 0.18 *** 0.32 *** 0.18 *** 0.23 *** 0.37 *** 0.02 0.14 **
Proprietary 0.33 *** 0.19 *** 0.26 *** 0.44 *** 0.21 *** 0.33 *** 0.28 *** 0.18 *** 0.23 *** 0.2 *** 0.19 *** 0.2 ***
Professional 0.39 *** 0.11 0.29 *** 0.62 *** -0.04 0.42 *** 0.21 *** 0.23 ** 0.21 ***
(b). Two-days long-short strategy: H: t+1 and L: t+2
Note: This table shows annual Sharpe ratios across all periods, traders, and volumes for (a) the
one-day long-short strategy and (b) the two-days long-short strategy: long and short position
initiated one and two days after the event day, respectively. Event day is when an asset’s ISE
ratio qualifies for each respective trading strategy. Sharpe ratios are annualized averages of
monthly ratios calculated every 21 days using excess returns with respect to the S&P 500
composite index and adjusted for transaction costs. Professionals’ data is since October 2009. *,
**, and *** represent statistical significance of a t-test of 0 at 10%, 5% and 1% respectively.
Table VII. Alpha of 3 and 4 Factors Models for Two-days Long-short Strategy.

Factors Trader/ May 2015 - December 2012 May 2005 - May 2008 May 2008-Sept 2010 Sept 2010 - December 2012
Strategy H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10%
3 Customer 0.07 *** -0.02 0.10 *** -0.23 *** -0.19 -0.22 ** 0.14 ** -0.12 0.14 * 0.11 *** 0.18 0.16 ***
Broker Dealer 0.26 *** 0.26 *** 0.33 *** -0.34 0.29 0.10 0.49 *** 0.59 *** 0.65 *** 0.18 ** 0.10 0.15 *
Proprietary 0.22 *** 0.34 *** 0.45 *** 0.54 *** 0.72 *** 0.69 *** 0.30 *** 0.55 *** 0.59 *** 0.08 ** 0.10 0.21 ***
Professional 0.20 *** 0.54 *** 0.39 *** 0.39 *** 0.78 ** 0.56 *** 0.11 * 0.38 * 0.29 ***
4 Customer 0.08 *** -0.02 0.11 *** -0.24 *** -0.23 -0.23 ** 0.14 ** -0.12 0.13 * 0.11 *** 0.19 0.16 ***
Broker Dealer 0.25 *** 0.26 *** 0.33 *** -0.33 0.28 0.11 0.45 *** 0.59 *** 0.64 *** 0.18 ** 0.10 0.15 *
Proprietary 0.22 *** 0.35 *** 0.45 *** 0.55 *** 0.68 *** 0.69 *** 0.29 *** 0.56 *** 0.58 *** 0.09 ** 0.13 0.23 ***
Professional 0.20 *** 0.54 *** 0.39 *** 0.39 *** 0.78 ** 0.56 *** 0.11 * 0.38 * 0.29 ***
Days 1124 450 339 335
H 80% L 20% HL 80-20% H 80% L 20% HL 80-20% H 80% L 20% HL 80-20% H 80% L 20% HL 80-20%
3 Customer 0.03 0.24 *** 0.11 *** -0.19 *** -0.26 -0.25 *** 0.07 0.28 * 0.12 ** 0.05 * 0.29 ** 0.12 ***
Broker Dealer 0.17 *** 0.26 *** 0.30 *** -0.13 0.29 0.10 0.33 *** 0.59 *** 0.55 *** 0.10 * 0.10 0.14 **
Proprietary 0.26 *** 0.34 *** 0.42 *** 0.41 *** 0.72 *** 0.61 *** 0.35 *** 0.55 *** 0.54 *** 0.16 *** 0.10 0.24 ***
Professional 0.22 *** 0.54 *** 0.35 *** 0.35 *** 0.79 ** 0.47 *** 0.14 *** 0.38 * 0.27 ***
4 Customer 0.03 0.25 *** 0.11 *** -0.20 *** -0.21 -0.24 *** 0.05 0.28 * 0.11 ** 0.05 * 0.31 ** 0.13 ***
Broker Dealer 0.17 *** 0.26 *** 0.30 *** -0.13 0.28 0.11 0.31 *** 0.59 *** 0.54 *** 0.10 * 0.10 0.15 **
Proprietary 0.25 *** 0.35 *** 0.43 *** 0.42 *** 0.68 *** 0.61 *** 0.34 *** 0.56 *** 0.54 *** 0.17 *** 0.13 0.25 ***
Professional 0.22 *** 0.54 *** 0.34 *** 0.35 *** 0.78 ** 0.47 *** 0.14 *** 0.38 * 0.27 ***
H 70% L 30% HL 70-30% H 70% L 30% HL 70-30% H 70% L 30% HL 70-30% H 70% L 30% HL 70-30%
3 Customer 0.01 0.11 ** 0.08 *** -0.12 ** -0.19 -0.18 ** 0.03 0.16 0.08 * 0.05 ** 0.16 * 0.11 ***
Broker Dealer 0.08 * 0.26 *** 0.22 *** -0.07 0.29 0.08 0.18 ** 0.59 *** 0.42 *** 0.03 0.10 0.09
Proprietary 0.23 *** 0.51 *** 0.41 *** 0.28 *** 0.74 *** 0.49 *** 0.33 *** 0.73 *** 0.52 *** 0.13 *** 0.33 *** 0.26 ***
Professional 0.17 *** 0.52 *** 0.28 *** 0.29 *** 0.69 ** 0.37 *** 0.10 *** 0.34 * 0.20 ***
4 Customer 0.01 0.11 ** 0.08 *** -0.13 ** -0.16 -0.17 ** 0.02 0.16 0.08 * 0.05 ** 0.17 ** 0.12 ***
Broker Dealer 0.08 * 0.26 *** 0.22 *** -0.07 0.28 0.09 0.16 * 0.59 *** 0.41 *** 0.03 0.10 0.10
Proprietary 0.22 *** 0.53 *** 0.41 *** 0.29 *** 0.72 *** 0.50 *** 0.31 *** 0.73 *** 0.51 *** 0.13 *** 0.38 *** 0.27 ***
Professional 0.17 *** 0.51 *** 0.27 *** 0.28 *** 0.69 ** 0.36 *** 0.10 *** 0.34 * 0.20 ***

Note: This table shows alpha for the two-days long-short strategy for different periods and
traders with a minimum volume of 100 contracts for the three factors of the Fama and French
(1993) model: market impact (Market), size premium (SMB) and book-to-market or value
premium (HML). The table also includes the return adjusted by the Carhart (1997) four factors
model that takes the Fama-French three factors framework and adds a fourth factor, the
momentum of a stock. Two-days long-short strategy requires a long and short position initiated
one and two days after the event day, respectively. Event day is when an asset’s ISE ratio
qualifies for each respective trading strategy. *, **, and *** represent statistical significance at
10%, 5% and 1% respectively. Professionals’ data is since October 2009.
Table VIII. Alpha of 3 Factor Fama-French and 4 Factor Carhart Models including Professional
Traders for Two-days Long-Short Strategy since October 2019.

Factors Trader/ Oct 2009 - Sept 2010 Oct 2009 - December 2012
Strategy H 90% L 10% HL 90-10% H 90% L 10% HL 90-10%
3 Customer 0.06 -0.06 0.06 0.10 *** 0.20 0.15 ***
Broker Dealer 0.18 0.40 0.33 * 0.17 ** 0.18 * 0.20 **
Proprietary 0.23 *** 0.36 ** 0.32 *** 0.12 *** 0.21 ** 0.26 ***
Professional 0.39 *** 0.78 ** 0.56 *** 0.20 *** 0.54 *** 0.39 ***
4 Customer 0.05 -0.13 0.04 0.10 *** 0.19 0.15 ***
Broker Dealer 0.19 0.40 0.33 * 0.17 ** 0.18 0.20 **
Proprietary 0.22 *** 0.35 * 0.31 *** 0.12 *** 0.20 ** 0.26 ***
Professional 0.39 *** 0.78 ** 0.56 *** 0.20 *** 0.54 *** 0.39 ***
Days 139 474
H 80% L 20% HL 80-20% H 80% L 20% HL 80-20%
3 Customer -0.01 0.24 0.05 0.04 0.31 *** 0.11 ***
Broker Dealer 0.22 0.40 0.33 ** 0.12 ** 0.18 * 0.20 ***
Proprietary 0.14 ** 0.36 ** 0.24 *** 0.17 *** 0.21 ** 0.26 ***
Professional 0.35 *** 0.79 ** 0.47 *** 0.22 *** 0.54 *** 0.35 ***
4 Customer -0.02 0.24 0.05 0.04 0.32 *** 0.12 ***
Broker Dealer 0.24 * 0.40 0.33 ** 0.13 ** 0.18 0.20 ***
Proprietary 0.13 ** 0.35 * 0.23 *** 0.17 *** 0.20 ** 0.26 ***
Professional 0.35 *** 0.78 ** 0.47 *** 0.22 *** 0.54 *** 0.34 ***
H 70% L 30% HL 70-30% H 70% L 30% HL 70-30%
3 Customer -0.01 0.01 -0.01 0.03 0.16 ** 0.09 ***
Broker Dealer 0.12 0.40 0.25 * 0.05 0.19 * 0.14 **
Proprietary 0.14 *** 0.36 ** 0.21 *** 0.14 *** 0.36 *** 0.27 ***
Professional 0.29 *** 0.69 ** 0.37 *** 0.17 *** 0.52 *** 0.28 ***
4 Customer -0.01 -0.01 -0.02 0.04 * 0.15 ** 0.09 ***
Broker Dealer 0.12 0.40 0.25 * 0.05 0.18 0.14 **
Proprietary 0.13 *** 0.35 ** 0.20 *** 0.14 *** 0.38 *** 0.27 ***
Professional 0.28 *** 0.69 ** 0.36 *** 0.17 *** 0.51 *** 0.27 ***

Note: This table shows alpha for the two-days long-short strategy for the different periods,
traders (including professionals) and strategies with a minimum volume of 100 contracts since
October 2009 for the three factors of the Fama and French (1993) model: market impact
(Market), size premium (SMB) and book-to-market or value premium (HML). The table also
includes the return adjusted by the Carhart (1997) four factors model that takes the Fama-French
three factors framework and adds a fourth factor, the momentum of a stock. Two-days long-short
strategy requires a long and short position initiated one and two days after the event day,
respectively. Event day is when an asset’s ISE ratio qualifies for each respective trading strategy.
*, **, and *** represent statistical significance at 10%, 5% and 1% respectively. Professionals’
data is since October 2009.
Table IX. Sharpe Ratios by Trader and Period including Professionals since October 2009.

Trader/ Oct 2009 - December 2012 Oct 2009-Sept 2010 Sept 2010 - December 2012
Strategy H 90% L 10% HL 90-10% H 90% L 10% HL 90-10% H 90% L 10% HL 90-10%
Customer 0.12 *** 0.09 0.12 *** 0.08 0.01 0.06 0.15 *** 0.12 0.14 ***
Broker Dealer 0.35 *** 0.06 0.13 *** 0.39 ** 0.1 0.19 ** 0.37 *** 0.02 0.14 **
Proprietary 0.24 *** 0.18 *** 0.22 *** 0.34 *** 0.18 ** 0.27 *** 0.2 *** 0.19 ** 0.2 ***
Professional 0.39 *** 0.11 0.29 *** 0.62 *** -0.04 0.42 *** 0.21 *** 0.23 ** 0.21 ***

Note: This table shows annual Sharpe ratios across all periods and traders for the two-days long-
short strategy: long and short position initiated one and two days after the event day,
respectively. The minimum volume is 100 contracts and the periods start in October 2009 when
data for professional traders become available. Event day is when an asset’s ISE ratio qualifies
for each respective trading strategy. Sharpe ratios are annualized averages of monthly ratios
calculated every 21 days using excess returns with respect to the S&P 500 composite index and
adjusted for transaction costs. Professionals’ data is since October 2009. *, **, and *** represent
statistical significance of a t-test of 0 at 10%, 5% and 1% respectively.
Table X. Trader Composition.

October 2009 to December 2012

Open Buy Open Sell Close Buy Close Sell
Put Call Put Call Put Call Put Call
Average volume 33.46 39.70 33.53 35.07 13.93 15.02 13.61 18.68
Proprietary 21.03% 19.94% 22.66% 22.06% 19.79% 15.86% 15.64% 13.61%
Customer 58.49% 61.26% 55.57% 58.47% 64.44% 68.52% 69.30% 72.73%
Broker Dealer 14.05% 11.79% 16.02% 12.86% 6.68% 5.25% 6.36% 4.60%
Professional 6.43% 7.00% 5.75% 6.61% 9.09% 10.37% 8.71% 9.07%

Note: This table shows each traders’ average put and call volume (option contracts) for all four
order types. The customer trader still dominates option volume. However, the percentage has
dropped compared to the trader composition seen in Table II, from 63% to 58.5% and overall
average volume fell for every opening order type, though remained consistent for closing order
types.
Table XI. Top 10 traded Stocks.

H 90% L 10% HL 90-10%

Ticker Freq Ticker Freq Ticker Freq
MU 289 MTG 55 MU 292
EMC 262 MAS 51 EMC 264
S 224 TSN 48 S 229
BMY 190 DHR 47 BMY 196
PFE 186 MBI 47 PFE 186
APC 172 SPG 46 APC 175
GLW 169 OMC 45 TWX 175
AMD 163 HRB 44 NBR 174
AMAT 161 EL 41 GLW 169
NBR 161 PHM 40 SYMC 169

Note: This table shows the ticker of the stocks that qualified for the various strategies for the
customer trader, H 90%, L 10% and HL 90-10% and the occurrence (Freq) of these trades. The
total number of trading days in the May 2005 to December 2012 used in the simulations is 1,124.
Figure I. Breakdown of ISE Option Data. This figure shows the ISE option data filtered down
to specific trader type call-put ratios, starting with particular asset order type, which for this
research was opening buy. Option volume data is segmented by specific asset, trader,
Customer/Broker Dealer/Proprietary/Professional, option type (put or call), and finally the daily
ISE call-put ratios, which is calculated by dividing the volume of opening buy call contracts by
the volume of opening buy put contracts. ISE Holdings is the International Securities Exchange
Holding company that operates two U.S. options exchanges: International Securities Exchange,
LLC and Topaz Exchange, LLC.

Transaction Types:
Opening Buy: Buying a call or put option, new position
Closing Buy: Buying a long position to close out
Opening Sell: Selling (writing) a call or put option, new position
Closing Sell: Selling a long position to close out
Trader Types:
Customer: Option trade volume for traders acting on behalf of discount and full-service
customers.
Broker-Dealer: Option trade volume for traders acting on behalf of institutional clients.
Proprietary: Option trade volume for proprietary traders acting on behalf of their firm.
Professional: Option trade volume for Non-Registered Broker Dealer traders whose daily
average is at least 390 trades (high-frequency traders). Including this trader, a separate analysis
was run between October 2009 and December 2012. A summary can be found in the
‘Professional Trader’ section of this paper.
Figure II. Daily returns of the SPY ETF during the highest volatile month (October 2008) and a
typical volatile month (October 2005) for the data series analyzed.

15.00%

10.00%

5.00%

0.00%

-5.00%

-10.00%

-15.00%

High Volatility Typical Volatility

 Figure III. Average Number of Stocks Traded by Trading Strategy: May 2005 to December
2012.
80

Average number of stocks per day

70

60

50

40

30

20

10

0
H 90% H 80% H 70% L 10% L 20% L 30%
Trading strategy

Customer Broker Dealer Propietary Professional

Note: Average total number of daily stocks selected during the complete period is 344.

View publication stats