Charming!

Retail Option Volume and Delta

Hedging Effects on Stock Pricing.

Matthew Flynn 1

Increased liquidity in the options market creates price pressure in the underlying stock market
through the hedging activity of intermediaries. When public option demand is imbalanced,
liquidity providers have inventory imbalance and create predictable price action in the
underlying through dynamic hedging. Using a simple proxy for public demand for options, I
identify a strong and persistent relationship with synchronous and future stock returns in a
manner predicted by net short delta hedgers' trades, distinct from information frictions. This
phenomenon is most likely due to retail investors hoarding on specific options.

Keywords: Options, option to stock volume, equity pricing, retail investors, price predictability,
price pressure

JEL Classification: G12, G13, G41

1
Matthew is a Ph.D. candidate in the finance department at Auburn University, Auburn, AL, USA. I would like to
thank my dissertation committee, Jimmy Hilliard, Jitka Hilliard, Zhongling Qin, and Justin Benefield for their
helpful suggestions and guidance. Thank you to Chris Stivers, the FMA, and SFA for the discussion of earlier
versions of the paper. Contact Matthew at [email protected]
 Introduction

Options markets are an attractive destination for speculative trades due to the embedded leverage
available for a small capital outlay. Influential studies such as Roll, Schwartz, and
Subrahmanyam (2010), Johnson and So (2012), and Ge, Lin, and Pearson (2015) have shown
that option markets can predict stock prices. Although the most popular mechanism for this
predictive ability is attributed to informed trading in the option market, new studies such as Ni,
Pearson, Poteshman, and White (2021), Barbon and Burashi (2021), and Bryzgalova, Pavlova,
and Sikorskaya (2023) have highlighted the mechanical connection between the option and
underlying stock market. Specifically, how the option inventories of market makers and broker
dealers can impact stock prices through dynamic hedging techniques. In this study, I provide
evidence that retail option trade has exacerbated hedgers' inventory imbalances and created
considerable predictable price pressure in stocks.

The impetus of this study is the substantial increase in options market liquidity due to the
participation of retail investors. Under pressure from retail-focused zero commissions brokerage
Robinhood, Charles Schwab, the largest brokerage in the world2, instituted commission free
trades in October 2019. Almost all other brokerages followed this move to remain competitive.
As a result, traded equity options have increased by roughly 67% year over year. Further, this
increase is concentrated in options with a "gambling" tilt, short term, out of the money call
options. These calls reflect positive sentiment and have the largest embedded leverage available
due to their low price. These options are consistent with the behavioral motivations of retail
traders, who have been shown to value lottery characteristics such as low prices, high
idiosyncratic volatility, and high idiosyncratic skewness 3. From 2019 – 2022, the average
volume traded for 0-2 weeks calls in billions of dollars was roughly $500 billion greater than all
other expiry classifications combined. This increase in liquidity and the changing nature of
options trading calls for a reexamination of the interconnectedness of the option and stock
markets.

2
For context, Schwab had AUM of over $3.5 trillion and approximately 33 million active accounts in 2019.
3
Bauer, Cosemans, and Eichholtz (2009), Kumar (2009), Green and Hwang (2012), Han and Kumar (2013),
Hvidkjaer (2008), Barber, Odean, and Zhu (2009), Kaniel, Liu, Saar, and Titman (2012), and Kumar, Page, Spalt
(2013).
 I hypothesize that increased demand for long option positions from retail traders will be
primarily serviced by market makers to fulfill their liquidity provision to the market. Market
makers are noted delta hedgers (Cox and Rubinstein (1985), Hull (2000), among others); thus,
the hedging trades resulting from their net written option positions create price pressure in the
underlying stock. The delta hedging hypothesis has two empirically testable implications:

1. Contemporaneous stock returns vary with the net option delta position of market makers.
Short delta (short calls) = buying pressure, long delta (short puts) = selling pressure.
2. Future returns on stocks where market makers are net short options will vary with the
process of the option's delta.
The derivative of delta with respect to time is an option's charm. Because charm is strictly
positive or negative depending on the characteristics of the option, it allows for inference about
the future stock trades of delta hedgers. I verified this claim in aggregate data and signed trades
with identified accounts of origin, i.e., tagged retail and market maker trades. The effect of
charm is in addition to hedgers' net gamma position, which captures the rate of change of delta
and, thus, rebalancing of hedges. Due to heightened imbalances, mechanical decay of delta
induced by time is an essential consideration in addition to the first-order rebalancing effect
captured by gamma imbalance. That is to say, the rebalancing implied by charm can cause stock
pressure independent of information or demand driven price shifts in the stock.
To test the stock pricing implications of options volume, I developed two simple
measures of abnormal option trading: excess call/put option volume, ECOV, and EPOV. The
measures are constructed as daily option volume less a rolling average of the previous five days.
In aggregate option volume data, these measures capture abnormal trading and proxy for the
short inventory of liquidity providers. The idea is that liquidity providers are likelier to take short
positions when volume increases unexpectedly. Although not a perfect proxy, these measures
have merit in their simplicity and reproducibility. Thus, strategies using excess option volume
(EOV) are reproducible with no private information.
EOV relates to contemporaneous and future price pressures in the direction predicted by
net short option delta hedging trades of liquidity providers. By multiplying EOV times option
delta and charm, I can calculate a proxy for the dollar amount needed to enter and rebalance a
delta hedge over time. Portfolios sorted by dollar value required to enter a delta hedge on short
option positions explain daily contemporaneous stock price pressure by as much as 50 basis
points (bps) in either direction. Further, subsequent day returns can be pressured by roughly 40
bps based on the relative volumes in the options market of the stock. By disaggregating the short
inventory of liquidity providers EOV into subgroups, I show options traded by retail investors
have the highest impact on stock pricing. If option volumes become more imbalanced in specific
charm classifications, liquidity provider inventory becomes more imbalanced, and predictability
increases. Predictability also increases in the average gamma of the options traded as higher
gammas represent a larger dollar amount needed to rebalance hedges. These price effects are
unlikely to be driven by information as they are highly transitory, with predictability rarely
lasting past one day.
These tests broadly suggest that mechanical price pressure caused by option inventory
imbalances of delta hedging liquidity providers can have economically significant impacts on
stock pricing. Mechanical effects even outpace private information that may be nested in option
volumes. For example, a one standard deviation increase in high embedded leverage call options
corresponds with decreased stock returns of roughly 35 bps on the next trading day. While this
result is consistent with mechanical rebalancing induced by net charm position, it makes little
sense in an informational context, as one would assume informed trading in high leverage call
options should relate to increases in stock price.
A zero-investment portfolio created by grouping deciles of stocks based on gamma
weighted charms of their options earns excess returns of 7.755 bps per day with a highly
significant t-stat of 3.246, or 19.5% per annum unadjusted for transaction costs. This portfolio
has relatively low risk as returns rely on mechanical pressure from hedging flow. Returns also
survive risk adjustment from the Fama French 3 factor model plus momentum.
Several pieces of information suggest that retail traders are the driving force behind the
stock price pressure caused by hedging. Tests using CBOE data bucketed by user account
indicate that most predictability occurs when retail traders are long options written by delta
hedging liquidity providers. A one standard deviation increase in long holdings by retail accounts
is associated with a 40 basis point decrease in the stock's next day return and correlates highly
with hedger charm position. Future returns are also much lower for stocks with high excess call
option volumes simultaneously highly held by Robinhood investors, despite the generally good
timing of Robinhood investors documented in Welch (2022). The differing outcomes of retail
interest in stocks and options are puzzling if not for understanding the hedging flow of parties
servicing retail flow in the options market. Results linking retail option trade to stock pricing are
novel and support the conclusions of Bryzgalova, Pavlova, and Sikorskaya (2023) retail option
trade affects stock price through the hedging of intermediaries servicing retail order flow.
The paper proceeds as follows. Section I outlines the place and contribution of the paper.
Section II documents the abrupt increase in options volume following broad access to
commission free trading, section III models option charm and presents testable hypotheses,
section IV documents data and variable construction, section V shows stock price pressure in the
aggregate, section VI connects aggregate predictability to retail trade, and section VII concludes.

I. Place and Contribution

The predictive ability of the relative volume of options to stocks is well documented (Roll,
Schwartz, and Subrahmanyam (2010), Johnson and So (2012), Ge, Lin, and Pearson (2015), Pan
and Poteshman (2006)). Primarily, these studies suggest an informed trading explanation in
which negative information for stocks is nested in higher options volume, leading to a negative
relation between the O/S ratio and future stock returns. This is because of short-sale constraints
in the stock market and the embedded leverage provision of options. However, these
explanations leave many existing studies with somewhat puzzling findings. Work by Ge, Lin,
and Pearson (2015), among several others, finds the predictability of signed option volume is
more robust for out-of-the-money options and trades originating from small accounts. In
addition, existing studies find the negative relation between the O/S ratio holds even when
considering only call options, which should have a positive relationship with future returns if the
option trading is informed.

Others have considered stock clustering or pinning, the phenomena by which stocks stick
close to the strike price of options near expiry (Ni, Pearson, and Poteshman (2004)). Such studies
offer a joint explanation, citing hedge rebalancing and informational channels working
independently or in conjunction. The delta hedging method of predicting mechanical pressure
differs from existing studies because it does not rely on an informational channel. Excess
demand for call (put) options should predict positive (negative) future returns in an informational
channel. Instead, they will depend on the option's moneyness within the delta hedging channel.
In several instances, a delta hedging channel will predict opposite price movement to an
informational one. In this way, I can plausibly separate the effects of private information and
mechanical pressure when considering the interconnectedness of the option and stock markets
and provide insight into which channel dominates and in what situations.

While other authors have considered the mechanical effects of delta hedging (Hu (2012),
Ni, Pearson, Poteshman, White (2021)), these studies primarily focus on the volatility of the
underlying because of the hedging trades. I supplement this work by showing instances where
hedging creates directional pressure in addition to volatility changes. Stivers and Sun (2013)
show that stock prices are pressured on option expiration weeks in a manner consistent with net
long option delta hedgers. I provide evidence that this relation has flipped since their study, as
retail traders have forced market liquidity providers into a net short option position. This
research is broadly related to the effect of option derivatives on underlying pricing, such as
Barbon and Burashi (2021) and Ni, Pearson, Poteshman, and White (2021). This study is the first
to sort by option charm, a purely mechanical movement that does not rely on the independent
price discovery of the underlying.

This paper also relates to much of the behavioral literature on retail trading. Retail
investors have been shown empirically to value lottery-like payoffs in their direct investments.
Bauer, Cosemans, and Eichholtz (2009), Kumar (2009), Green and Hwang (2012), and Han and
Kumar (2013) document the tendency of retail investors to speculate in equity markets by
overpaying for stocks with features exhibiting a higher probability of significant magnitude
returns, such as low prices, high idiosyncratic volatility, and high idiosyncratic skewness. The
correlated trading decisions of return-seeking retail investors are also shown to move stock
prices in a manner inconsistent with market efficiency and stock fundamentals by Hvidkjaer
(2008), Barber, Odean, and Zhu (2009), Kaniel, Liu, Saar, and Titman (2012), Han and Kumar
(2013), and Kumar, Page, Spalt (2013), among others. In a more recent and pertinent example,
Umar, Gubavera, Yousaf, and Ali (2021) use the case of GameStop to provide evidence that this
type of sentiment-driven trading by retail investors can cause extensive and economically
significant asset mispricing. In the managed fund literature, Agarwal, Jiang, and Wen (2021)
provide evidence that mutual fund managers strongly prefer stocks with "lottery-like"
characteristics at the behest of their investors. Funds holding more lottery stocks attract higher
flows than their conservative counterparts. Retail investors tend to overweight small probability
events that have impacts of a large magnitude. This behavioral phenomenon is the same as
identified in much of the theoretical risk management and insurance literature, where individuals
are willing to pay a premium to avoid the chance of exceptionally bad or damaging outcomes, no
matter the actual probability (Mendez, Hanson 1970). The options market is an obvious
destination for investors seeking lottery-like returns due to the limited downside risk and large
amounts of embedded leverage. Choy (2014) shows that a higher retail trading proportion in an
option chain is related to significantly lower option returns. This evidence suggests that retail
investors speculate on options and pay a "lottery premium" on expected future volatility,
resulting in more expensive options with higher implied volatilities. Choy and Wei (2020) show
retail investors tend to buy more calls and puts on the daily winner and loser stocks, leading to an
overvaluation of those options. I add to this literature by linking the behavioral tendencies of
retail traders in the derivatives market to price pressures in the underlying.

II. Option Volume after Broad Access to Commission Free Trading

Robinhood was founded on April 18th, 2013, with the mission of "democratizing finance for all"
– Vlad Tenev, Robinhood CEO. Robinhood targets a user base of retail clientele by offering
commission-free trades and simplifying trading with a user-friendly app as its primary interface.
As an example of the tailored experience for novice investors, Robinhood offers a learning
experience in its host of brokerage services. The learn tabs on the Robinhood website include
articles such as "What is an investment?" and "What is the stock market?". Robinhood quickly
grew in popularity, accruing more than fifteen million active accounts as of 2023. The growth of
Robinhood coincides with a cultural shift in how young retail traders view investing. Investment-
related topics receive more public attention than ever in traditional media and social media, such
as the online Reddit community behind the GameStop short squeeze: WallStreetBets. Increased
interest in finance is also evident in the success of several recent films, such as Dumb Money and
The Big Short, with prevalent themes of risk and the gamification of trading. These cultural
headwinds culminated in Charles Schwab's decision to implement commission-free trading in
October 2019, reducing options trading costs from $4.95 to $0.65 per contract. Other major
brokerages quickly followed with similar reductions in the price of options trading to remain
competitive in the global marketplace.
 To visualize the increase in options volume, Figure 1 plots the time series of monthly
aggregate option volume from January 2010 to December 2022 with a line indicating the
institution of commission free trading by Schwab. As shown in Figure 1, the options volume
remained relatively flat from 2010 – 2019, averaging around 375 million monthly contracts.
However, since the institution of commission free trading, the average number of monthly
options contracts has increased precipitously, primarily in equity options. The average number of
contracts traded after the broad introduction of commission free trading has increased by roughly
67% year over year, with an average of 625 million contracts per month in 2021 and over 750 in
2022.

Figure 1 – Monthly Option Volume

Aggregate monthly option volume in millions from January 2010 to December 2022. The solid
black line denotes equity options, and the dashed line indicates index options. The solid vertical
line represents Charles Schwab moving to commission free trade.

Further, the type of options being traded has also shifted. Retail traders have been shown
to prefer lottery characteristics in their direct investments 4. In the options market, these

4
Bauer, Cosemans, and Eichholtz (2009), Kumar (2009), Green and Hwang (2012), and Han and Kumar (2013),
Umar, Gubavera, Yousaf, and Ali (2021).
preferences are fulfilled with short term options. Figure 2 plots the monthly dollar value of
aggregate option volume by expiry classification and call/put.

Figure 2 – Grouped Dollar Option Volume

U.S. dollar option volume is computed as the number of contracts traded times the strike price
and is represented in billions of dollars monthly from January 2016 – December 2022. Dollar
volume is grouped by expiry classification and call/put. The legend matches the color grouping
of the figure. The dashed vertical line denotes Schwab commission free trade.

Figure 2 indicates that the average volume of options with less than two weeks to expiry
is roughly $500 billion greater per month than all other expiry classifications combined after
Schwab zero commissions. The aggregate increase is also highly concentrated in calls, with call
options comprising roughly 68% of the total option market after the broad adoption of
commission free trading. Untabulated results indicate that highly traded retail "meme" stocks are
among the top ten stock option markets by volume from 2019 - 2022.

Because retail traders prefer long options to pursue lottery-like returns, liquidity
providers such as market makers must write these options. When the liquidity provider cannot
quickly offload the short option position, such as retail hoarding causing imbalanced demand,
they will offload their directional risk exposure through dynamic hedging. If these inventory
imbalances become significant and pervasive, there can be effects on underling asset pricing.
 For example, consider the well covered case of GameStop. In January 2021, a
coordinated effort by retail investors drove GameStop to a market capitalization of over $22
billion, roughly 30 times the valuation at the start of the year. Because approximately 140% of
GameStop's public float had been sold short, closing these short positions caused even further
price increases. Even before the short squeeze, Keith Gill, more popularly known by the alias
"Roaring Kitty", purchased roughly $53,000 in call options and saw the position rise to $48
million by January 27, 2021. While the gains of this trade are undoubtedly stunning, the
mechanism that drove the increase is not unique to GameStop. Retail traders frequently
coordinate online in their option positions, leading to imbalanced option markets and, thus,
imbalanced portfolios of liquidity providers. In addition to the short squeeze, GameStop was
further driven upward by the hedging trades of short gamma liquidity providers in the options
market.

III. Modeling Price Pressure and Hypothesis Development

A market maker's primary function is to service public demand in markets by providing liquidity.
This function is crucial in the options market, where unhedged short option positions have
damaging implications for retail traders. According to the Chicago Board Options Exchange
(CBOE): "for a naked written call or put position, initial margin includes all of the option
proceeds and 20% of the value for the underlying securities, whereas, for covered option
positions, it includes 50% of the value for the underlying securities." In terms of economic
magnitude, a single naked call option contract written on the SPDR S&P 500 ETF (the most
popular call option) will require at least $8,000 in cash or cash equivalents in the writer's account
based on a $400 strike price. Should the option be called, the writer must produce $40,000. This
calculation can be scaled upward by the number of contracts written. Due to the implicit risk,
market makers will likely take short option positions of highly demanded options in service of
their liquidity provision to the market. Further, the significant increases in options volume shown
in section II implicitly require liquidity provision, especially considering the concentration of the
option volume.

I formally test the assumption that retail traders primarily take long option positions using
the CBOE Open-Close Volume Summary. This data provides aggregated volume bucketed by the
account of origin, buy/sell, and open/close. Customer trades under 100 are shown to be a
realtively strong proxy for retail trade by Bryzgalova, Pavlova, and Sikorskaya (2023). It should
be noted, however, that small lot trades may also capture iceburg orders, which are used to
mitigate costs and price impact. The prevalence of iceburg orders should be relatively small,
however, given that CBOE identifies non professsional public customers and that iceburg
techniques are typically used for very large transactions. This data ranges from 2019 to 2021 and
covers roughly 30% of all U.S. option volume.

Table I reports the percentage of volume per account group per year. Retail traders most
frequently buy calls to open and generally buy to open at a much higher rate than any other
group, consistent with a gambling preference. While market makers are shown to be balanced at
the aggregate, they service a large portion of the retail flow and, thus, frequently have
imbalanced inventory at the stock level5. If retail traders begin hoarding to specific options, the
market will not clear, and market makers will be forced to hedge directional risk in the
underlying stock market.

While market makers do not typically pass through all option volume to the underlying stock,
hoarding by retail traders creates special market conditions that make delta hedging by
intermediaries more likely. The following summarizes the standard risk mitigation practices of
market makers and how retail trade complicates these mechanisms:

1. Balanced markets eliminate the need for hedging. i.e., two-way order flow allows market
makers to offload option inventory quickly. Because retail traders hoard on options, total
demand becomes imbalanced.
2. The net delta impact is muted as net inventory grosses put and call option deltas. Because
retail traders overwhelmingly purchase calls, the aggregate portfolio of liquidity
providers is more dependent on the delta of these options.
3. Market makers can sometimes use proxy hedges such as ETFs. With such substantial
increases in options volume, proxy hedges can become relatively expensive due to fees.

5
Bryzgalova, Pavlova, and Sikorskaya (2023) indicate that market makers service a large percentage of total retail
flow as brokers receive payment for order flow when routing trades.
 Table I - Signed Trades as a Percentage of Total Group Volume Per Year
Values are reported as a percentage of a signed trade class's total yearly group volume. Groups are firms (firm), broker dealers (BD),
customer trades under 100 contracts (Ret), and market makers (MM). Signed trade classifications are buy to open (OB), sell to open
(OS), buy to close (CB), and sell to close (CS).
Panel A - Firm Firm Firm Firm BD BD BD BD Ret Ret Ret Ret MM MM
2019 OB CB OS CS OB CB OS CS OB CB OS CS Buy Sell
C 6.70 16.25 6.80 16.25 6.05 16.30 5.54 16.30 21.21 9.09 14.70 9.09 25.67 26.28
P 8.41 18.68 8.23 18.68 8.60 20.06 7.11 20.06 15.80 8.00 14.10 8.00 23.69 24.36
Panel B -
2020
C 8.56 13.60 8.58 13.60 4.97 15.54 4.76 15.54 23.73 9.05 14.39 9.05 27.15 26.76
P 10.70 17.23 10.49 17.23 8.89 22.75 4.79 22.75 15.54 7.71 12.83 7.71 23.14 22.95
Panel C -
2021
C 9.71 12.35 9.62 12.35 3.16 11.42 11.68 11.42 24.51 8.97 14.70 8.97 27.94 27.00
P 13.10 15.47 11.94 15.47 15.71 21.57 3.48 21.57 14.99 7.56 12.74 7.56 22.53 22.53
 As a risk-neutral party, market makers will seek to hedge their directional exposure to the
underlying stock. Unbalanced option portfolios are becoming more common for liquidity
providers as option trade increases in popularity with retail investors (Barbon and Baruchi
(2021)). Market makers can manage their risk exposure when inventory is imbalanced through
dynamic delta hedging. Option delta represents the change in option value relative to the change
in the underlying asset's value. Delta hedging involves trading in the underlying asset to achieve
a net-neutral delta position for the complete portfolio. Market makers can offload the directional
risk of an unbalanced option portfolio through this practice. As a testament, market makers are
noted delta hedgers (Cox and Rubinstein (1985), Hull (2000), among others). Option volume can
have pricing implications for the underlying stock through these hedging trades.

A. Modeling Market Maker Price Pressure

Suppose one assumes the elasticity of the underlying stock price concerning selling and buying
volume is non-zero (as in the Avellanda and Lipkin (2003) Model). Then, the rebalancing of
delta hedges will affect stock prices. To understand how delta hedging creates price pressure,
consider the Black-Scholes formula for the price of a call on a non-dividend paying stock:

S 1
ln� �+�r+ σ2 �t
C = SN(d1 ) − Ke−rt (d2 ), where d1 = K 2
, and d2 = d1 − σ√t
σ√t

Within the Black Scholes framework, S = stock price, K = strike price, r = risk-free rate, σ =
volatility of the underlying, t = time to expiry, and N (.) = the cumulative normal distribution
function. Delta (Δ) is the partial derivative of the call price with respect to the stock price and
can be expressed by N(d1 ). Application of the put-call parity allows the delta of a put option to
be solved as N(d1 ) − 1. This result implies the derivative of delta with respect to time (or charm)
is the same for call and puts, as is shown here:

S σ2
∂Δ ln� �−�r+ �t
K 2
= N′(d1 ) 2σt3/2
, where N′(.) is the pdf.
∂t

As time to expiry approaches zero, an explicit sign of charm may be inferred by the option's
σ2
moneyness. As t approaches zero, the term �r + � t becomes insignificantly different from
2

zero, resulting in the following inequalities for charm:

 S
∂Δ N′ (d1 ) ln� �
= 2σt3/2
K
> 0 if S > K
∂t

S
∂Δ N′ (d1 ) ln� �
= K
< 0 if S < K
∂t 2σt3/2

Figure 3 models the charm of a call option with a $100 strike price. As the option approaches
expiration, the absolute value of charm increases substantially. Charm becomes sharply negative
for OTM calls where S < K and vice versa. Charm's effects on option delta can result in large
trading volumes in the underlying asset from hedgers.

Figure 3 – Charm of Call Option

This figure models the charm of a call option with a strike price of $100 as a function of asset
price and time to expiration. The legend is color-coded to represent the rate of change of option
delta on the three-dimensional surface.

Charm implies that option delta is predictable over time. By this fact, price pressure from
delta hedging is also predictable, given the net charm position of hedgers. A long position in
stock is a positive delta position. Therefore, a short call is short delta; a short put is long delta,
and vice versa. As delta changes, market makers must readjust their stock holdings to retain a
delta-neutral position. Depending on the initial delta position of market makers, the charm of the
options will imply directional pressure in the underlying stock as market makers trade in the
direction of the movement of the delta. This process and resulting price pressure for hedgers'
short options are described in Figure 4.

Figure 4 – Price Pressure from Hedging

This figure represents the delta position and corresponding action for parties engaged in delta
hedging. These positions and actions assume delta hedgers are short options.

The effect of charm imbalance summarized in Figure 4 is necessarily independent of the

effect of hedger net gamma position proposed by Ni, Pearson, Poteshman, and White (2021). The
effect of charm is purely mechanical and implies price pressure even with static underlying
markets. However, this is not to say that the net gamma position does not influence the net charm
position of liquidity providers. In fact, the directional pressure predicted by the net charm
position is increased by the gamma of the traded options. In this way, charm is not a secondary
but independent effect.

If public demand is significant enough to create inventory imbalances in market maker's

option portfolio, we should note predictability, as summarized in Figure 4. This process implies
that large call (put) option volumes will be associated with buying (selling) pressure from market
makers contemporaneously. These returns will be large and economically significant as the
majority of the total dollar volume for the life of a delta hedge is the opening of the stock
position.

Over time, the delta of the option will evolve in the process described by the option's
charm and, thus, will predict the hedging pressure on the underlying stock. Demand for call
options with positive (negative) charm implies future buying (selling) pressure as the delta
increases (decreases). Conversely, demand in put options with positive (negative) charm means
future selling (buying) pressure as the delta increases (decreases). Notably, many of the
predictions of Figure 4 are juxtaposed to expectations if options trading were informed. For
example, demand for high embedded leverage call options (calls with negative charm) indicates
downward price pressure in the delta hedging framework.

In contrast, the opposite should be true if trading in the call options is informed. The
differing predictions between the hedging framework presented here and an informed trading
hypothesis offer the opportunity to test the relative power of the two channels. As several
previous studies have provided some form of an informed trading explanation6, heterogeneous
results would be interesting.

B. Hypotheses

I summarize the empirically testable implications of the delta hedging hypothesis as follows:

1. Contemporaneous stock returns will be pressured by option demand through intermediary

hedging.
2. Future stock returns will be predictable by proxying for the charm position of option
intermediaries.
3. Stock predictability via charm is an increasing function with respect to option gamma.

I validate these hypotheses in the aggregate and in more granular data, allowing for precise trade
direction and account of origin identification. While aggregate results speak to the economic
magnitude of the price pressure, a study with granular data provides a sharper identification of
retail interest and intermediary holdings. Option gamma is the second derivative of the option
price with respect to stock price and represents the rate of change of delta. Ceteris paribus, when
gamma is larger, rebalancing on delta hedges will be larger. As these price pressures do not
necessarily represent information, the impact on prices is transitory and should only have a larger
effect when imbalances are higher.

IV. Data and Important Variable Construction

6
See Roll, Schwartz, and Subrahmanyam (2010), Johnson and So (2012), Ge, Lin, and Pearson (2015), Pan and
Poteshman (2006).
Aggregate option data is collected daily from Option Metrics and matched with CRSP daily
stock files from 2016 – 2022. Account-matched data from CBOE is collected daily from 2019 to
2021. The relatively short timeframe of this study is due to the limited participation of retail
traders pre-2019. Robinhood user holding data is collected from May 5, 2018, to August 13,
2020, from the Robintrack7 website to document retail equity preferences. Unfortunately, the API
used by Robintrack to import user holdings was suspended in August 2020. Factor portfolios are
retrieved from Ken French's website for risk-adjusted returns.

A. Important Variable Construction

To proxy for public demand, I construct the following simple variable that captures abnormal
volume in options:

∑t−1
t−6 #conracts tradedi,d
E(C/P)OVi,t = # of contracts tradedi,t −
5

Excess option volume (EOV) refers to the number of option contracts traded for firm i on day t
less a rolling average of the number of contracts traded for firm i on the previous five days8.
EOV is calculated separately for calls and puts, ECOV, and EPOV at the individual option level.
By construction, EOV captures abnormal increases in volume, making it an intuitive measure for
capturing option demand and resultant short positions of hedgers. Although simplistic, this
measure is advantageous regarding public availability and ease. Further, any private information
nested in EOV is noisy. Since negative values of EOV have no intuitive meaning, they are set to
zero.

Option charm is calculated as:

S S σ2
∂Δ N′ (d1 ) ln� � log� �+�r+
K 2
�t
= K
, where d1 =
∂t 2σt3/2 σ√t

The sign of charm depends on the characteristics of the option. Option deltas and gammas are
collected from Option Metrics. EOV is calculated by charm group to proxy for charm position of
liquidity providing delta hedgers.

7
An excellent description and application of Robintrack data can be found in Welch (2022).
8
Results using EOV are robust to several choices for number of days used in the rolling average.
 Supposing EOV proxies for written options by liquidity providers, I can then construct
net charm and net gamma position by multiplying the short interest of liquidity providers times
100 (as options are written in groups of 100) and option charm and gamma, respectively. Without
two-way volume, we can expect inventory imbalances to be passed through to the underlying
stock mechanically, through charm rebalancing, and dynamically, through gamma rebalancing.
Because EOV captures abnormal volume by construction, I believe it is reasonable to assume the
volume is concentrated on a specific side of the trade, suggesting the ability of liquidity
providers to unload inventory quickly is limited. 9

For comparability to other studies focusing on aggregate option volume, I also include
the daily option to stock ratio as a control. I follow Roll, Schwartz, and Subrahmanyam (2010)
and construct the option-to-stock ratio as follows:

# of option tradesi,t
(O/S)i,t = # of stock tradesi,t

Because option volume causes stock volume through hedging in my model, the O/S ratio will not
appropriately capture stock pricing effects.

B. Summary Statistics

Summary statistics and correlations are presented in Table II. The differential information
captured by the O/S ratio and EOV is underscored by their respective standard deviation in Panel
A. The standard deviation of the O/S ratio is low as, by construction, option volumes are
normalized by stock volume. In contrast, EOV has a standard deviation of over 13,000 contracts
daily. The high variability of EOV is critical as I assume abnormal demand is likely to be written
by hedging liquidity providers. Hoarding events on an option will cause substantial spikes in
EOV that are likely to be written and subsequently hedged by option liquidity providers. Because
these spikes are unexpected and large in magnitude, the pass through to the stock market causes
transitory price pressure. The heightened variability of EOV after commission free trading can be
seen in Figure 5. The increase in the volatility of EOV is most likely due to retail participation in
the options market, which further suggests EOV captures public long interest in an option and
,thus, written short positions by liquidity providers.

9
Should a reader find this assertion unconvincing, I validate aggregate results using signed volume in section VI.
 Figure 5 – Variability of EOV

This figure plots the aggregate EOV in millions of monthly option trades from 2016 to 2022 by
calls and puts. The red line plots calls and puts the green line. The black vertical line denotes the
institution of commission free trading by Charles Schwab.

The hedging relationship can also be seen in Panel B of Table II. While the O/S ratio has
a relatively low correlation with stock volume, EOV and excess stock volume (ESV) correlate
25.12%. The relative correlations provide evidence that option volume causes stock volume
through hedging in a manner not captured by the O/S ratio.

Unablated results suggest the relationship between stock market volume and options
market volume has increased by 13% since the institution of broad commission free trading.
Further, a Chow test indicates a structural break in the time series of option to stock volume after
the Schwab decision in October 2019 with an F-statistic of 44.951. This relation can be
understood through intermediary hedging outlined in section III, whereby option inventory
imbalances due to the increased long volume are passed through to the underlying stock.
 Table II - Summary Statistics and Correlations
Summary statistics and correlations are calculated from a merged panel of Option Metrics and
CRSP daily data from 2016 - 2022. EOV and ESV are computed as daily option/stock volume
less a rolling average of the previous five days. The O/S ratio is calculated as daily option
volume divided by daily stock volume.
Panel A – Summary Statistics
Variable Mean Std. Dev. 1st Median 3rd N
Quartile Quartile
O/S 0.00092 0.005576 0.00005 0.00022 0.00077 4,263,105
ECOV 4.5 13,766 -157 -16 28 4,263,105
EPOV 0.5 13,608 -124.4 -13.7 22.3 4,263,105
ESV -2,163 3253739 -208,433 -27,659 96,937 4,263,105
Option Volume 4,381 40,805 18 115 761 4,263,105
Stock Volume 2275000 8143250 247,900 653,300 1,787,000 4,263,105
Panel B - Correlations
Correlation Stock Volume EOV ESV O/S
Stock Volume 1 0.0469 0.1386 0.0051
EOV 0.0469 1 0.2512 0.0128
ESV 0.1386 0.2512 1 -0.0031
O/S 0.0051 0.0143 -0.0031 1

V. Aggregate Stock Predictability

The first two implications of the delta hedging hypothesis deal with empirically testable patterns
in stock price as a function of option volume. These patterns should be consistent with hedging
pressure by net short option intermediaries.

A. Contemporaneous Returns

Suppose liquidity providers are net short options and delta hedge. In that case, higher ECOV
(EPOV) will predict higher (lower) synchronous returns as hedgers must buy (sell) the
underlying stock to enter the hedge. To test for this relation, I sort stocks by:
 $Enteri,t = � EOVo,t ∗ Stock Priceo,t ∗ Option Deltao,t ∗ 100

$Enter approximates the short interest of liquidity providing delta hedgers and estimates the
dollar amount needed to enter a delta hedge. For example, consider a market maker short ten call
options (proxied by EOV) with an underlying price of $60 and a delta of 0.8. Because options
are executed in orders of 100, the market maker must purchase 10*60*(0.8)*100 = $48,000
worth of the underlying stock to hedge directional exposure. Because calls/puts have a
positive/negative delta, $Enter will correlate positively to stock return. Table III sorts stocks into
deciles based on their respective values of $Enter.

Table III – Contemporaneous Stock Return by $Enter

$Enter is calculated by multiplying the EOV of an option times stock price, option delta, and
100. N denotes the number of observations, return is the average daily return, and trades
represent the average number of stock trades per day per firm. T-statistics are reported from a
one-sample t-test per group where the theoretical mean is the full sample daily return. The
sample period is from January 2016 – December 2022.
Decile N Return t-stat Trades
1 (Lowest) 380,037 -0.00523 -147 10,101
2 380,037 -0.00315 -97.6 4,735
3 380,037 0.00000005 -28.5 2,355
4 380,036 -0.000138 -28.7 2,462
5 380,036 -0.000647 -40.1 3,011
6 380,036 -0.000298 -26.9 3,146
7 380,036 0.00105 -1.68 4,302
8 380,036 0.00192 16.2 5,743
9 380,036 0.00318 41.5 8,039
10 (Highest) 380,036 0.00515 79.5 14,299
High-Low 0.01038 102.6 4,198

Per Table III, equal-weighted returns increase nearly monotonically by $Enter. Firms in the
lowest decile of $Enter have average returns of -0.523% per day, and firms in the highest decile
have average returns of 0.515% per day. Although impossible to recreate given that returns are
contemporaneous, a long-short strategy on $Enter would produce returns of 1.038% per day.

The relationship between $Enter and stock return in Table III is also broadly consistent
with informed option trading. Because option and stock markets are complementary, traders will
rationally allocate some of their investment to the stock market and some to the option market
when new information arises. In this way, higher volumes in the options market for calls (puts)
will indicate good (bad) news for the stock and drive the observed relation. However, the
disaggregation of option volumes to calls and puts paints a clearer picture. To illustrate this point,
Table IV sorts stocks into deciles by their contemporaneous level of excess call and put option
volume.

Table IV – Synchronous Returns by ECOV and EPOV Decile

This table presents the daily summary statistics of firms sorted by ECOV (Panel A) and EPOV
(Panel B) levels. EC(P)OV is the daily call (put) volume on an underlying stock minus a
rolling average of the previous five trading days. N denotes the number of firm day
observations within each decile sort, return is the average daily return of the associated decile,
and trades indicate the average number of stock trades per day per firm in the associated
decile. T-statistics are reported from a one-sample t-test per group where the theoretical mean
is the full sample daily return. The sample period is from January 2016 – December 2022.
Panel A – ECOV
Decile N Return t-stat Trades
1 (Lowest) 428,927 -0.00232 -80.1 14,569
2 428,373 -0.00176 -71.7 11,478
3 427,863 -0.00143 -64.2 10,039
4 427,365 -0.00109 -55.4 9,179
5 426,814 -0.00069 -45.2 8,717
6 425791 -0.000158 -31.2 8,285
7 425,244 0.000955 -3.92 8,853
8 424,732 0.0025 26.7 10,337
9 424,237 0.00476 60.4 12,784
10 (Highest) 423,759 0.012 105 22,104
 High-Low 0.01432 152.94 7,535
Panel B – EPOV
1 (Lowest) 371,328 0.00115 0.318 16,378
2 370,763 0.0011 -0.617 13,249
3 370,240 0.00118 1.07 11,570
4 369,726 0.000987 -3.15 10,670
5 369,144 0.000988 -3.13 10,036
6 368,206 0.000726 -8.81 9,695
7 367,709 0.000633 -9.82 10,340
8 367,182 0.000223 -16.1 12,096
9 366,706 -0.000291 -21.6 14,931
10 (Highest) 366,189 -0.00112 -20.3 24,964
High-Low -0.00227 -78.3 8,586

Interestingly, the relationship between option volume and stock return grows stronger
when focusing exclusively on call options. By simply sorting stocks into deciles based on call
option volume, a contemporaneous long-short strategy produces returns of 1.432% per day from
2016 – 2022, roughly six times greater than the returns generated by recreating the same strategy
with puts. In an informational context, this is puzzling, given that puts should have a higher
correlation with the price discovery of stocks due to short sale constraints in the stock market.
The stronger correlation of call option volume with contemporaneous return is therefore likely
due to vastly larger volumes in call vs. put options and resultant hedging flow. Further, the call
long-short strategy annual returns increase by 21% after the broad adoption of commission free
trading in 2019. Back of the envelope calculations suggest that hedging flow proxied by $Enter
represents roughly 25% of total stock volume from 2019 – 2022.

B. Predictable Future Returns

After liquidity providers make contemporaneous trades to open a delta hedge, future rebalancing
trades should be predictable by the charm of short option inventory. To test future predictability
and the ability of EOV to proxy for short positions, I generate EOV in four charm subgroups per
underlying stock. I summarize the groups, their delta implication, and their price pressure
predictions below:

1. Negative Charm Call Options = Decreasing Delta = Selling Pressure

2. Positive Charm Call Options = Increasing Delta = Buying Pressure
3. Negative Charm Put Options = Increasing Delta = Buying Pressure
4. Positive Charm Put Options = Decreasing Delta = Selling Pressure
If EOV captures delta hedging and impacts pricing, more extreme values of charm will indicate
more significant changes in delta and, thus, more price pressure. In addition, the predictive
ability of charm should be an increasing function of gamma since gamma represents the rate of
change of delta. To account for this effect, I also compute the average option gamma of each
charm group.

B.1. Net Pressure

Hedging pressure generated from option inventory imbalance in different charm categories will
sometimes compete with one another at the stock level. For example, excess short inventory in
both calls and puts with negative charm imply opposite directional pressure and thus will net out.
I, therefore, proxy for the complete inventory of liquidity providers by constructing a stock-level
net hedge. Net hedge sums together EOV for all charm subgroups that indicate future buying
pressure less EOV for all subgroups that indicate future selling pressure:

Net Hedgei,t = ECOV Positive Charmi,t + EPOV Negative Charmi,t − ECOV Negative Charmi,t
− EPOV Positive Charmi,t

By construction, net hedge should positively correlate with future returns as hedgers rebalance
their positions according to their option inventory charm. Additionally, I weight option volumes
by the gamma of the group. This weighting should increase the magnitude of predictability as
higher gamma implies larger rebalancing trades. To test this relation, I regress the following
model:

Ret i,t+1 = Net Hedgei,t + Gamma Weighted Net Hedgei,t + Controls + ϵi,t

This model represents the average change in subsequent day returns for the net change in EOV
for charm subgroups. Table V presents the results.
 Table V – Aggregate Return Predictability

This table analyzes next-day return predictability as a function of option volume in charm
subgroups. Columns 1 and 2 estimate separately for net hedge and gamma weighted net hedge:
𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡+1 = 𝑁𝑁𝑁𝑁𝑁𝑁 𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝑖𝑖,𝑡𝑡 + 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊ℎ𝑡𝑡𝑡𝑡𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁 𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝑖𝑖,𝑡𝑡 + 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝜖𝜖𝑖𝑖,𝑡𝑡

Where Net Hedge is constructed as ECOV Positive Charmi,t + EPOV Negative Charmi,t −
ECOV Negative Charmi,t − EPOV Positive Charmi,t , and gamma weighted net hedge is
weighted by the average gamma of the subgroups. Daily observations span from January 1,
2016 – December 31, 2022. Controls include lags of return, market return, stock volume, and
stock size. T-stats are reported in parentheses. Coefficients represent a one standard deviation
change.
Dependent Variable:

Next-Day Stock Return (bps)

(1) (2)
Net Hedge 2.97
(15.587)

Gamma Weighted Net Hedge 41.1

(11.874)

Controls YES YES

Observations 587,926 587,926
R2 0.141 0.141

The coefficients on net hedge and gamma weighted net hedge in Table V are positive and
highly significant in both specifications. One standard deviation increase in an equally weighted
net hedge is associated with a 2.97 bps increase in the stock's subsequent day returns. Gamma
weighting the EOV in charm subgroups dramatically increases the magnitude of the coefficient
to 41.1 bps. The large increase in magnitude when gamma weighting is consistent with hedging
pressure from option liquidity providers hedging net short exposure in the options market, as
higher gammas indicate larger absolute changes in delta. Further, the significance of the hedge
coefficients dissipates past a one-day lead of returns. Transitory pressure is consistent with
mechanical effects from hedging, but it is difficult to explain if option trading was informed
somehow.

B.2. Decomposed Pressure

Differential price impacts from EOV in different charm groups will be informative for two main
reasons:

1. Because retail option trading is highly condensed in specific options, liquidity providers
likely have higher imbalances. This relationship lends itself to higher directional pressure
consistent with long option retail traders.
2. Decomposing net pressure to charm subgroups allows for a direct comparison of
informational explanation of price predictability to mechanical hedging.

To test these propositions, I regress the following model:

Ret i,t+1 = � Charm Groupsi,t + � Average Group Gammai,t + �� Charm Groupsi,t �

∗ �� Average Group Gammai,t � + Controls + ϵi,t

Suppose EOV captures the short inventory of hedgers. In that case, the coefficients on charm
subgroups should be significant and sign in the direction predicted by delta hedge rebalancing
with increased magnitude when interacted with average group gamma. Table VI reports the
results.

Table VI – Decomposed Price Pressure

This table analyzes next-day return predictability as a function of option volume in

disaggregated charm subgroups. Columns 1 and 2 estimate without and with gamma interaction
coefficients:

𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡+1 = � 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + � 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + �� 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 �

∗ �� 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 � + 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝜖𝜖𝑖𝑖,𝑡𝑡

Where charm groups refer to the EOV within respective charm subgroups, and average option
gamma is calculated for each group. Controls include lags of return, market return, stock
volume, and stock size. Daily observations span from January 1, 2016 – December 31, 2022. T-
stats are reported in parentheses. Coefficients represent a one standard deviation change.

Dependent Variable:

Next-Day Stock Return (bps)

(1) (2)

ECOV Negative Charm -13.766 -10.3245

(-12.045) (-8.997)

ECOV Positive Charm -2.0649 -1.3766

(-1.421) (-0.0675)

EPOV Negative Charm 0.9788 13.608

(0.6585) (0.8525)

EPOV Positive Charm 2.0412 12.247

(1.9225) (2.3695)

ECOV Negative*Avg Gamma -27.532

(-4.349)

ECOV Positive*Avg Gamma 55.064

(2.251)

EPOV Negative*Avg Gamma -34.02

(-2.112)

EPOV Positive*Avg Gamma -115.67

(-5.213)

Controls YES YES

Observations 587,926 587,926
R2 0.142 0.143

All but one of the marginal effects of charm subgroups in Table VI enter as significant at
conventional levels, with the sign hypothesized by the delta hedging hypothesis. Results are
uniformly more robust for higher absolute values of charm and gamma, indicating increased
hedging flow to the stock. To illustrate the economic impact of these coefficients, let us consider
the marginal effect of EOV in negative charm call options. In line with the predictions of the
delta hedging hypothesis, demand in this group of options should indicate future selling pressure.
Indeed, the marginal effect on next-day bps returns of an increase in EOV for this group is -
10.3245 – 27.532 *(Average Gamma). Therefore, a one standard deviation increase in EOV for
negative charm call options equates to a 10.3 bps decrease in subsequent day returns for options
with zero average gammas or a 37.86 bps decrease as gamma approaches one. Conservatively,
this represents a 21.55 basis point drop from the total sample average or a 193% decrease. In
addition, the effect of charm is independently significant when controlling for stock return and
option gamma, illustrating the mechanical price pressure without underlying price movement.

The coefficients are consistent with retail option demand driving stock predictability. The
most significant and largest magnitude coefficients are on the highest retail demanded options,
negative charm calls, and positive charm puts. Section II shows these options are traded much
more than any other subgroup following broad access to commission free trading. As retail
traders hoard on these options, liquidity providers short the options and hedge exposure
consistent with the charm and gamma of the options.

Furthermore, the coefficients for ECOV negative charm are opposite to what would be
expected with an informational explanation. It is difficult to understand a negative coefficient on
excess demand in call options if trading was informed. This result suggests that hedging pressure
from liquidity providers can, at the least, mute price movement from informed option trade. This
effect reduces the cleanliness of using aggregate option volumes to infer information.

B.3. Hedging Portfolios

I now construct a trading strategy based on relative option volumes with no private information.
Because most of the increase in options volume is concentrated on short term calls and puts, I
consider EOV in those option groups. I sort stocks daily into two groups based on median,
quartiles, and deciles, equally weighted by EOV and gamma weighted. I then construct a zero-
investment portfolio by going long the highest group and short the lowest group. EOV in short
term calls (puts) implies forward selling (buying) pressure per delta hedging. Table VII reports
excess returns from this strategy in bps.

Table VII - Zero Investment Portfolio on Charm Signal (Daily Rebalancing)

This table presents returns of a zero-investment portfolio for median, quartile, and decile sorts
based on 0-2 week EOV in calls and puts. The top panel is equally weighted; the bottom panel
is weighted by the gamma of the options. Portfolios are rebalanced daily from January 1, 2016,
to December 31, 2022. T-statistics are reported from a one sample t-test per group where the
theoretical mean is the total sample daily return. Excess return is defined as portfolio return
less the daily risk-free rate.
Median Quartiles Deciles
Calls Puts Calls Puts Calls Puts
Average Excess Return (bps) -1.755 0.357 -3.18 0.476 -6.353 1.2
t-stat -1.829 0.39 -2.239 0.33 -2.808 0.443
Gamma Weighted Charm Signal
Average Excess Return (bps) -3.135 0.947 -5.033 0.325 -7.755 0.617
t-stat -3.058 0.99 -3.298 0.209 -3.246 0.217

The returns from the portfolios in Table VII are highly significant for calls and increases
in magnitude when weighting observations by gamma and choosing more extreme sorts. A
strategy that longs stocks in the lowest decile of gamma weighted charm and shorts stocks in the
highest decile generates an excess return of 7.755 bps per day or 19.5% per annum, unadjusted
for fees or trading costs. Since retail traders overwhelmingly trade call options with negative
charms, these results are consistent with downward stock price pressure caused by intermediary
hedging.

The returns of the portfolios also survive risk adjustment. Specifically, I regress the
returns of the gamma weighted ECOV portfolios against the Fama French 3-factor model plus
momentum. Figure 6 presents the alphas of each portfolio.
 Figure 6 – Alphas of Gamma Weighted ECOV Portfolios

Stocks are sorted into decile portfolios based on their level of ECOV in negative charm calls.
The alphas are generated from:

𝑅𝑅𝑅𝑅𝑅𝑅𝑝𝑝,𝑡𝑡+1 = 𝛼𝛼 + 𝑀𝑀𝑀𝑀𝑀𝑀𝑡𝑡 + 𝑆𝑆𝑆𝑆𝑆𝑆𝑡𝑡 + 𝐻𝐻𝐻𝐻𝐻𝐻𝑡𝑡 + 𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡 + 𝑀𝑀𝑀𝑀𝑀𝑀𝑡𝑡 + 𝜖𝜖𝑖𝑖,𝑡𝑡

Where return is the return of each portfolio daily, the factors are from Fama French plus
momentum. Observations are daily and span from 2016 – 2022.

Alphas are highly significant for 7/10 of the decile portfolios and decrease as ECOV in negative
charm options increases. Again, the decrease is consistent with net short liquidity providers
selling stock to rebalance their hedge.

VI. Connection to Retail Option Trade

Several pieces of evidence point to retail trade as the driving force in liquidity provider inventory
imbalance. Per Table I, the option trading patterns of institutional and professional investors have
not materially increased after the broad adoption of zero commission trading. Further, the
concentration of options volume in short term out of the money calls in high attention stocks is
broadly consistent with the documented behavioral patterns of retail traders. As further evidence
for this proposition, I directly link option volume and price pressure to retail activity.

A. Robinhood User Holdings

Using the Robintrack database, I observed Robinhood user equity holdings from May 5, 2018, to
August 13, 2020. Although Robintrack does contain option data, it is likely that equity holdings
correlate highly with option holdings for retail traders. While there is no concrete way of testing
the relationship, I believe it to be a reasonable assumption. To test the relationship between
option volumes and Robinhood user holdings, I construct the following model:

Ret i,t+1 = Dec(ECOV)i,t + Hold_FDi,t + Dec(ECOV)i,t ∗ Hold_FDi,t + Controls + ϵi,t

Where Ret i,t+1 refers to a one-day lead of returns for firm i on day t + 1. Dec(ECOV)i,t measures
a stock's decile of excess call option volume per day. I focus on call options in this test as they
represent a bullish position and should correlate more highly with equity holdings than put
options. Hold_FDi,t is calculated as Daily RH Holdersi − lag(Daily RH Holders)i for each stock
i on day t, creating a first difference of the holdings of Robinhood users. Hold_FD represents
aggregate changes in Robinhood user holdings and proxies for positive retail sentiment on a
stock. Because retail investors concentrate their options trades in short term out of the money call
options and liquidity providers delta hedge their short call option exposure, I expect a negative
coefficient on the interaction term Dec(ECOV)i,t ∗ Hold_FDi,t , as these options have negative
charm and imply selling pressure from rebalancing. This coefficient is key as it highlights the
liquidity provision of hedgers to long demand from retail traders. Table VIII presents the results.

Table VIII – ECOV and Robinhood Holdings

The full model in column 5 with all controls is:

𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡+1 = 𝐷𝐷𝐷𝐷𝐷𝐷(𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸)𝑖𝑖,𝑡𝑡 + 𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻_𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝑀𝑀𝑀𝑀𝑀𝑀𝑡𝑡 + 𝑆𝑆𝑆𝑆𝑆𝑆𝑡𝑡 + 𝐻𝐻𝐻𝐻𝐻𝐻𝑡𝑡 + 𝑀𝑀𝑀𝑀𝑀𝑀𝑡𝑡 + 𝐸𝐸𝐸𝐸𝐸𝐸𝑖𝑖,𝑡𝑡 +

𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝐷𝐷𝐷𝐷𝐷𝐷(𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸)𝑖𝑖,𝑡𝑡 ∗ 𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻_𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝜖𝜖𝑖𝑖,𝑡𝑡 .

Where Ret i,t+1 is a one-day lead of returns for firm i and ECOVi,t and ESVi,t are measures for a
stock's excess option /stock volume. Ken French's website gathers Fama French 3 factor
|R |
portfolio returns plus momentum. The Amihud (2002) measure is constructed as Illiqi,t = Voli,t .
i,t

Hold_FD is constructed as Weekly RH Holdersi,t − lag(Weekly RH Holders)i,t. Huber-White

robust standard errors are used with four lags. T-statistics are reported in parentheses.
Coefficients represent a one standard deviation increase.
 Dependent variable:

Next-Day Stock Return (bps)

(1) (2) (3) (4) (5)

Dec (ECOV) -4.01 -4.01 -4.01 -4.01 -4.01

(-49.26) (-48.26) (-49.06) (-49.87) (-48.06)

Hold_FD 0.641 0.641 0.641 0.641

(45.57) (45.28) (45.26) (45.67)

Dec (ECOV)*Hold_FD -0.101 -0.101 -0.101 -0.101

(-32.051) (-32.16) (-32.615) (-32.674)

Controls YES YES YES YES YES

Observations 374,970 370,834 370,834 370,834 370,834
R2 0.017 0.017 0.017 0.018 0.018

The decile of ECOV enters every regression as negative and highly significant. Firms
moving one decile higher in ECOV experience a four-basis point decrease in the next week's
return. This result implies that firms in the highest decile of ECOV have weekly returns 40 basis
points lower than firms in the lowest decile of ECOV, even when including all controls. The
interaction term of Hold_FD and ECOV enters as negative and highly significant despite the
standalone term of Hold_FD having a significant positive relation with returns. 10 The marginal
effect of ECOV on return in bps is -4.1 – 0.101*(Hold_FD). This effect implies returns will
decrease 14 bps for a modest increase in Robinhood holdings of 100 shares if a stock's EOV
decile increases by one. In an informational context, it is puzzling why Hold_FD and the
interaction term of Hold_FD and ECOV have opposite signs. If private information or picking
ability drives the positive relation between retail holdings and stock performance, option trades
should also display this relation, which is not observed empirically. Instead, these results are

10
Welch (2022) also notes stock picking ability of retail traders in their direct equity investments.
consistent with option liquidity providers servicing retail demand, creating downward price
pressure in the underlying stock through hedging.

B. Sharp Identification of Hedgers and Retail Trades

Although the return relations in the aggregate are consistent with liquidity provider hedging,
further rigor is needed to identify the source of return predictability. In this section, I document
two stylized facts which support the conclusions drawn from the aggregate relationships.

1. Options spreads are sensitive to market maker inventories and charm imbalance, and
retail traders are predominantly price takers and thus insensitive to options liquidity.
2. Market makers' rebalancing trades pressure stock returns, and the direction of these
trades is predictable, given the net charm position of delta hedgers.

These propositions are tested with the Chicago Board of Options Exchange (CBOE) Open-Close
Volume Summary. This data provides aggregated volume bucketed by account of origin
(customer, professional, broker-dealer, and market maker), buy/sell, and open/close.

B.1. Hedger Imbalance and Option Spreads

In addition to the dynamic hedging techniques (delta hedging) discussed in section III, market
makers often adjust option spreads when their inventory becomes imbalanced. As imbalanced
inventory exposes market makers to higher risk, spreads can be widened to discourage further
trading in the direction that exacerbates the inventory imbalance. Higher spreads on options also
compensate market makers for carrying more inventory risk as they represent higher premiums
on facilitated trades.

A natural question to the documented aggregate price pressures in section V is why these
inventory imbalances are not absorbed at the stock level if they are predictable. The
predictability of returns at the stock level relies on two key assumptions. First, retail spikes in
trading in the option market are largely unexpected¸ and second, retail option traders are
insensitive to transaction costs (option spread). Because retail spikes are not driven by one large
trade, aggregate increases in costs are not absorbed by any singular party and are instead broken
up among many traders. This insensitivity can further create inventory imbalance. To test these
proposed relations empirically, I present the following model:
SD of Option Spreadsi,t = α + |mm net charm|i,t + |bd net charm|i,t + retail long i,t +
pro long i,t + ∑ Controls + εi,t

MM / BD net charm is constructed as the short inventory of each party in a respective option
multiplied by the option charm, underlying stock price, and 10011. This measure is then
aggregated to the stock level (i) to represent the dollar amount needed to rebalance a delta hedge
on day (t). Therefore, a negative value would imply the need to short stock to rebalance a
position. Retail/pro bull is the number of open long positions the respective party holds. Controls
include several MM / BD charm lags, SD of option spreads, and group and time fixed effects.
Table IX reports the results.

Table IX – Hedger Charm Relation to Option Spreads

This table reports the results of the regression: SD of Option Spreadsi,t = α +
|mm net charm|i,t + |bd net charm|i,t + retail long i,t + pro longi,t + ∑ Controls + εi,t .
The dependent variable measures the volatility of option spreads on a given underlying stock
throughout a trading day. MM / BD net charm is constructed as the short inventory of each
party in a respective option multiplied by the option charm, underlying stock price, and 100.
This measure is then aggregated to the stock level (i) to represent the dollar amount needed to
rebalance a delta hedge on day (t). Retail/pro long is the number of open long positions the
respective party holds. Controls include five lags of MM net charm, five lags of SD of options
spread, and unit and time fixed effects in certain specifications. Coefficients represent a one
standard deviation increase.
Dependent Variable:
SD of Option Spreads
(1) (2) (3) (4)
| MM Net Charm | 0.10931 0.10929 0.06300 0.02114
(126.52780) (126.529) (62.5447) (30.41469)

| BD Net Charm | -0.05917 0.00328

(-9.76640) (2.32625)

11
This measure is multiplied by 100 as option contracts are traded in orders of 100.
 Retail Long -0.00651 -0.00239 0.00247
(-12.1858) (-10.9243) (5.76663)

Pro Long -0.00029 -0.00068

(-0.51455) (-1.4097)
Controls YES YES YES YES
Unit Fixed Effects NO NO NO YES
Time Fixed Effects NO NO NO YES
Observations 708,514 708,514 708,514 708,514
R2 0.0497 0.0500 0.1331 0.0141

Results from Table IX support the proposition that option spreads are sensitive to market
maker charm position and end-user demand. The coefficient on the absolute value of MM net
charm position is positive and highly significant in every specification. These coefficients
indicate between a two and ten cent increase in the standard deviation of option spreads for a one
standard deviation increase in MM net charm. The ability of the MM to adjust spreads can be
noted in the significant increase in magnitude and statistical significance over the coefficients on
broker dealer net charm.

Spreads are also sensitive to the type of option trader. Coefficients on net retail long
positions are statistically significant in every specification, while the coefficient on professional
trader long positions is never significant. Further, while retail long positions are negatively
related to spread in the pooled models, the relation flips to positive with the inclusion of unit and
time fixed effects. While one would typically expect end user demand to lead to tighter spreads
naturally, this positive coefficient on retail long indicates that after controlling for the type of
option, retail traders are insensitive to transaction costs and create more volatile spreads.

The coefficients on MM net charm and retail long positions provide evidence for a
passthrough effect in risk management from the option to the stock market. MM cannot fully
balance inventories within the options market and must adjust option spreads, allowing for
greater flexibility in dynamic hedging in the underlying stock market. Because risk-seeking retail
traders are primarily insensitive to transaction costs, MM inventories can become increasingly
imbalanced.

B.2. Hedger Imbalance and Stock Return Predictability

If inventory imbalances in the options market are passed through to the underlying stock market,
returns should be predictable with respect to hedge rebalancing. Following Barbon and Burashi
(2020), I consider market makers and broker dealers likely to be engaged in delta hedging. To
test for stock return predictability as a function of delta hedging, I regress the following model:

Ret i,t = hedger net charmi,t + retail/pro long i,t + high imbalance hedgeri,t +
hedger net charmi,t ∗ retail/pro long i,t ∗ high imbalance hedgeri,t + ∑ Controls + εi,t

Hedger net charm is constructed as MM net charm plus BD net charm and is aggregated to the
stock level (i) to represent the dollar amount needed to rebalance a delta hedge on day (t). For
example, if the hedger net charm was -$100, hedgers would need to short $100 worth of the
underlying stock to rebalance their delta position. High imbalance hedger is an indicator variable
equal to one if the hedger's short option interest on a day is in the top decile for that calendar
year. This model is split into two specifications with retail and pro traders to show heterogeneous
results based on the account of origin. The triple interaction term of hedger net charm, retail
long, and high imbalance is of particular interest as it indicates the effect of hedge trades when
inventories are highly imbalanced in conjunction with high retail demand. The delta hedging
hypothesis predicts this coefficient should be positive and significant as hedgers rebalance their
trades due to imbalanced inventories driven by retail demand. Table X presents the results from
this model.

Table X – Hedger Net Charm and Stock Returns

This table reports the results of the regression: Ret i,t = hedger net charmi,t +
retail/pro long i,t + high imbalance hedgeri,t + hedger net charmi,t ∗ retail/pro longi,t ∗
high imbalance hedgeri,t + ∑ Controls + εi,t . Hedger net charm is constructed as MM net
charm plus BD net charm and is aggregated to the stock level (i) to represent the dollar amount
needed to rebalance a delta hedge on day (t). High imbalance hedger is an indicator variable
equal to one if the hedger's short option interest on a day is in the top decile for that calendar
year. This model is split into two specifications with retail and pro traders to show
 heterogeneous results based on the account. Controls are 3-day lags of stock return, Amihud
illiquidity, O/S ratio, option volume, stock volume, and market return. Coefficients represent
basis point increases for a one standard deviation increase.
Dependent Variable:
Stock Return (bps)
(1) (2)
Hedger Net Charm -1.11180 -0.99045
(-1.36632) (-0.93510)

Retail Long -37.91351

(-49.41058)

Pro Long -1.85304

(-3.26531)

High Imbalance Hedger -10.39614 -9.65696

(-4.87163) (-4.52309)

Hedger Net Charm*Retail Long*High Imbalance 2.87404

(15.30288)

Hedger Net Charm*Pro Long*High Imbalance 0.33895

(0.62224)
Controls YES YES
Observations 1,259,373 1,259,373
R2 0.10780 0.10267

The results from Table X help to illustrate the differential effects of retail and
professional trade in the options market. The magnitude of the effect of retail long options
positions on stock returns is roughly twenty times greater than that of professional trade. A one
standard deviation increase in retail long option positions is equivalent to a 37.9 basis point
decrease in the next day's stock return, consistent with the aggregate level estimates in section V
and highly statistically significant. Unconditional on option trading, retail traders have been
shown to make good stock predictions.12

Additional evidence indicates that hedging pressure from parties servicing retail order
flow significantly contributes to the negative relation of retail long option positions and stock
returns. The coefficient on high imbalance hedger is significant at the 1% level and indicates a
~10 basis point decrease in stock return when hedgers are in the top decile of inventory
imbalance for a calendar year, consistent with negative charm inventory imbalance. Further, the
coefficient on the triple interaction term of hedger net charm, retail long, and high imbalance is
positive and highly significant, compared to an insignificant coefficient when retail long is
substituted for professional traders. Because retail traders overwhelmingly purchase calls with
negative charm, rebalancing delta hedges by parties short this demand creates short lived
downward price pressure in the stock, and the observed relations in Table X. In all, these results
indicate that mechanical pressure from net short intermediaries servicing retail option demand
results in stock return predictability.

VII. Conclusion

I provide evidence of a significant relationship between the net charm position of option liquidity
providers and underlying stock returns. This relation is consistent with the mechanical
rebalancing of hedges of intermediaries in the options market servicing long retail demand. The
charm effect is distinct from gamma because it is purely mechanical and exists even with static
underlying stock markets.

Since the broad institution of commission free trading in late 2019, retail option trading
has exploded in popularity. Total option volume has increased by roughly 67% year over year.
Further, the increase in options volume is concentrated primarily in short-term call options, as
these options serve as a gambling preference for retail traders. The total dollar volume of 0 – 2
weeks to expiration options is roughly $500 million greater per month than all other expiry
classifications combined.

12
Welch (2022) shows from mid-2018 to mid-2020, an aggregated crowd consensus portfolio (a proxy for the
household-equal-weighted portfolio) had both good timing and good alpha.
 Due to these vast increases in long demand, liquidity providers in the options market
have been forced into net short option positions. Without two-way order flow, these
intermediaries have been forced to hedge risk exposure in the underlying stock market because
of inventory imbalance. The price pressure caused by hedging flow is observable using aggregate
proxies and signed volume data that identifies the account of origin.

In the aggregate, the contemporaneous stock price can be pressured by as much as 50 bps
based on a proxy for net short option liquidity providers entering a delta hedge. This pressure is
much more significant when focusing exclusively on call options due to vastly higher call
relative to put option volumes post commission free trading. Estimates suggest that hedging flow
comprises roughly 25% of total stock volume from 2019 – 2022. Future stock returns are also
pressured by the rebalancing of hedges. Subsequent day returns are pressured by 40 bps based on
the net charm position of option liquidity providers. Again, this predictive power is highly
concentrated in short term call options.

Evidence suggests that long retail demand in the options market ultimately drives
aggregate stock pressure. Robinhood user holdings in stock are strongly predictive of lower
future returns when considering abnormal call option volume in the stock, despite Robinhood
users' generally good stock picking ability. Retail option trade is also shown to be directly related
to lower future stock returns by using signed account data from the CBOE. A one standard
deviation increase in long option holdings by retail investors correlates with a 40 bps decrease in
subsequent day returns, consistent with aggregate proxies for the net charm position of liquidity
providers. This evidence is consistent with net short option liquidity providers hedging exposure
in the underlying stock market and creating predictable price action.

References

Agarwal, V., Jiang, L., Wen, Q. (2021). Why do Mutual Funds Hold Lottery Stocks? Journal of
Financial and Quantitative Analysis. Volume 57, Issue 3, pp. 825 – 856.

Amihud, Y. (2002). Illiquidity and stock returns: cross-section and time-series effects. Journal of
Financial Markets, Volume 5, Issue 1, pp. 31-56.

Anderegg, B., Ulmann, F., Sornette, D. (2022). The impact of option hedging on the spot market
volatility. Journal of International Money and Finance, Volume 124.
Avellanda, M., Lipkin, M. (2003). A Market-Induced Mechanism for stock Pinning. New York
University Courant Institute Working Paper.

Barber, B., Odean, T., Zhu N. (2009). Do Retail Trades Move Markets? Review of Financial
Studies. Volume 22, Issue 1, pp. 151-186.

Barbon A., Burashi, A. (2021). Gamma Fragility. University of St.Gallen School of Finance
Research Paper Series.

Bauer, R., Mathijs Cosemans, Piet Eichholtz (2009). Option trading and individual investor
performance. Journal of Banking & Finance, Volume 33, Issue 4, pp. 731-746.

Black, F., Scholes, M. (1973). The Pricing of Options and Corporate Liabilities. The Journal of
Political Economy, Volume 81, Issue 3, pp. 637-654.

Boehmer, E., Jones, C., Zhang, X., Zhang, X. (2021). Tracking Retail Investor Activity. Journal
of Finance, Volume 76, Issue 5, pp. 2249 – 2305.

Bryzglova, S., Pavlova, A., Sikorskaya, T. (2023). Retail Trading in Options and the Rise of the
Big Three Wholesalers. Journal of Finance, forthcoming.

Cao, C., Chen, Z., Griffin, J. (2005). Informational Content of Option Volume Prior to
Takeovers. The Journal of Business, Volume 78, Issue 3, pp. 1073 – 1109.

Choy, S. (2015). Retail clientele and option returns. Journal of Banking and Finance, Volume
51, pp. 26 – 42.

Choy, S., Wei, J. (2020). Liquidity risk and expected option returns. Journal of Banking and
Finance, Volume 111.

Clogg, C. C., Petkova, E., & Haritou, A. (1995). Statistical methods for comparing regression
coefficients between models. American Journal of Sociology, Volume 100, Issue 5, pp. 1261–
1293.

Cox, J., Rubinstein, M. Options Markets. Englewood Cliffs, Prentice-Hall, Inc., 1985.

Deshpande, M., Sen, A., Pu, B., Gao, S., Krauklis, E. (2020). Impact of Retail Options Trading.
Barclays Equity Research, U.S. Equity Derivatives Strategy.
Ekkehart, B., Jones, C., Zhang, Xiaoyan, Zhang, Xinran. (2021). Tracking Retail Investor
Activity. Journal of Finance, Volume 76, Issue 5, pp. 2249 – 2305.

Fama, E., French, K. (1992). Common risk factors in the returns of stocks and bonds. Journal of
Financial Economics, Volume 33, pp. 3 – 56.

Farrell, M., Green, T.C., Jame, R., Markov, S. (2022). The democratization of investment
research and the informativeness of retail investor trading. Journal of Financial Economics,
Volume 145, Issue 2, Part B, pp. 616 – 641.

Garleanu, N., Pedersen, L., Poteshman, A. (2009). Demand-Based Option Pricing. Review of
Financial Studies, 22(10), 4259–4299.

Ge. L., Lin, T., Pearson, N. (2016). Why does the option to stock volume ratio predict stock
returns? Journal of Financial Economics, Volume 120, Issue 3, pp. 601 – 622.

Goncalves-Pinto, L., Grundy, B., Hameed, A., Heijden, T., Zhu, Y. (2020). Why Do Option
Prices Predict Stock Returns? The Role of Price Pressure in the Stock Market. Management
Science, Volume 66, Issue 9, pp. 3799 – 4358.

Green, T.C., Hwang, B. (2012). Initial public offerings as lotteries: expected skewness and first
day returns. Management Science, Volume 58, Issue 2, pp. 432 – 444.

Han, B., Kumar, A. (2013). Speculative retail trading and asset prices. Journal of Financial and
Quantitative Analysis, Volume 48, Issue 2, pp. 377 – 404.

Hu, J. (2014). Does Option Trading Convey Stock Price Information? Journal of Financial
Economics. 111, (3), 625-645.

Hull, J. Options, Futures, and other derivatives. Prentice-Hall, Inc., 2000.

Hvidkjaer, S. (2008) Small Trades and the Cross-Section of Stock Returns. Review of Financial
Studies, Issue 21, pp. 1123-1151.

Johnson, T., So, E. (2012). The option to stock volume ratio and future returns. Journal of
Financial Economics, Volume 106, Issue 2, pp. 262 – 286.
Jones, C. M., Reed, A. V., Waller, W. (2021). When Brokerages Restrict Retail Investors, Does
the Game Stop? Columbia Business School Research Paper Forthcoming, Available at SSRN:
https://ssrn.com/abstract=3804446 or http://dx.doi.org/10.2139/ssrn.3804446

Kaniel, R., Liu, S., Saar, G., Titman, S. (2012). Individual Investor Trading and Return Patterns
around Earnings Announcements. Journal of Finance, Volume 67, Issue 2, pp. 639 – 680.

Kumar, A. (2009). Who gambles in the stock market? Journal of Finance, Volume 64, Issue 4,
pp. 1889 – 1933.

Kumar, A., Lee, C. (2006). Retail Investor Sentiment and Return Comovements. Volume 61,
Issue 5, pp. 2451 – 2486.

Kumar, A., Page, J., Spalt, O. (2016). Gambling and Comovement. Journal of Financial and
Quantitative Analysis, Volume 51, Issue 1, pp. 85 – 111.

Menezes, C.F., and Hanson, D.L. (1970). On the Theory of Risk Aversion. International
Economic Review, Volume 11, Issue 3, pp. 481 – 487.

Ni, S., Pearson, N., Poteshman, A. (2005). Stock price clustering on option expiration dates.
Journal of Financial Economics, 78 (1), 49 - 87.

Ni, S., Pearson, N., Poteshman, A., White, J. (2021). Does Option Trading Have a Pervasive
Impact on Underlying Stock Prices? Review of Financial Studies, 34(4), 1952-1986

Pan, P., Poteshman, A. (2006). The Information in Option Volume for Future Stock Prices.
Review of Financial Studies, Volume 19, Issue 3, pp. 871 – 908.

Roll, R., Schwartz, E., Subrahmanyam, A. (2010). O/S. The Relative Trading Activity in Options
and Stock. Journal of Financial Economics, Vol 96, Issue 1, pp. 1 – 17.

Johnson, T., So, E. (2012). Journal of Financial Economics, Vol 106, Issue 2, pp. 262 – 286.

Ozik, G., Sadka, R., Shen, S. (2021). Flattening the Illiquidity Curve: Retail Trading During the
COVID-19 Lockdown. Journal of Financial and Quantitative Analysis, Volume 56, Issue 7, pp.
2356 – 2388.
Umar, Z., Gubareva, M., Yousaf, I., Ali, S. (2021). A tale of company fundamentals vs sentiment
driven pricing: The case of GameStop. Journal of Behavioral and Experimental Finance,
Volume 30.

Welch, I. (2022). The Wisdom of the Robinhood Crowd. The Journal of Finance, Volume 77,
Issue 3, pp. 1489 – 1527.

Zhing, Y., Zhang, X., Zhao, R. (2010). What Does the Individual Option Volatility Smirk Tell
Us About Future Equity Returns? Journal of Financial and Quantitative Analysis, 45(3), 641-
662.