-
Algorithmic Traders and Volatility Information
Trading
Anirban Banerjee∗ Ashok Banerjee†
October 1, 2019
Abstract
Are algorithmic traders informed about future realized
volatility? We construct
demand for volatility through trading volume in stock options
and relate this to
future realized volatility in the spot market. Using six months
(Jan - Jun 2015)
of trading data in both stock and stock options market for 160
stocks, we find
that non-algorithmic traders and not algorithmic traders are
informed about future
volatility. Both propitiatory and agency algorithmic traders
behave similarly in
this regard. We also find that the predictability for future
realized volatility in
the spot market does not last beyond two trading days. We use
both scheduled
earnings announcements and unscheduled corporate announcements
as exogenous
information events. The primary results are robust for various
measures of realized
volatility.
1 Introduction
Do algorithmic traders have information on future volatility?
Informational role of al-
gorithmic traders has been discussed extensively in the
literature. Most of the studies
∗Assistant Professor, Finance, Accounting and Control Group,
Indian Institute of ManagementKozhikode, Email:
[email protected]
†Professor, Finance and Control Group, Indian Institute of
Management Calcutta
1
-
suggest that algorithmic traders do not have directional
information, but react much
faster to publicly available information (Frino, Viljoen, Wang,
Westerholm, & Zheng,
2015). Unlike directional information, which is primarily
utilized in the spot (cash) or
futures market, the options market is uniquely suited for
traders with volatility related
information. In this paper, we examine whether algorithmic
trades in the Indian stock
options market have predictive ability for future realized
volatility in the spot market.
The benefit of leverage and lower margin requirements suggest
that derivative markets
are better suited for informed traders. The nature of
information that traders use could
be either directional or volatility related. In the case of
directional information, the trader
is supposed to know if the price of a particular security was to
go up or down. In case of
volatility information, the direction of future price movement
is not known to the trader.
However, the trader is better informed to predict if the price
level is supposed to move
from its current level (in either direction).
The last decade has witnessed a significant growth in
algorithmic trading activities,
not just in developed markets, but also in developing markets. A
significant proportion
of the order messages received by the exchanges is generated
automatically through com-
puters without any real time manual intervention. A subset of
these algorithmic traders
are known as high-frequency traders (HFT) who use the advantage
of speed to bring the
round-trip trade execution time down to microseconds. Academic
research shows that
these HFTs have taken on the role of ‘modern market makers’
(Menkveld, 2013). This
significant change in dynamics calls for a better understanding
of the role of algorithmic
traders, especially in derivative markets, where they are more
active.
We use the framework provided by Ni, Pan, and Poteshman (2008)
to estimate if any
particular trader group has volatility related information while
trading in the options
market. We use a unique dataset obtained from the National Stock
Exchange of India,
which provides identifiers for algorithmic trades. NSE is a
completely order-driven market
with no designated market maker. Due to their non-linear payoff
structures, stock options
are usually perceived riskier by the less sophisticated (retail)
traders. Considering that
2
-
NSE also has a liquid stock futures market, the stock options
market is usually more
attractive for algorithmic and other sophisticated traders.
We estimate the volatility demand of algorithmic and
non-algorithmic traders and
check if this demand has predictive ability for future realized
volatility in the spot market.
We use six months (Jan-Jun 2015) of intraday data for all 159
stocks which are permitted
to be traded in the derivatives market during this period. We
use data for both spot and
options market to estimate the volatility demand and realized
volatility measures. We
also further split algorithmic traders into proprietary and
agency algorithmic traders and
check if they behave differently with respect to trading on
volatility related information.
Our primary findings suggest that non-algorithmic traders are
informed regarding
future volatility while algorithmic traders are not. The options
market volatility demand
for non-algorithmic traders has predictive ability for future
realized volatility in the spot
market even after controlling for options implied volatility and
other relevant controls.
However, the predictive ability of options market volatility
demand rarely lasts more than
two days into the future. We also find that neither proprietary
(who trade in their own
account) nor agency (who execute trades on behalf of others)
algorithmic traders have
volatility related information. We consider both scheduled and
unscheduled corporate
announcements for periods with higher information asymmetry. Our
findings are robust
for both these announcement types. We also document the
variation in results with
respect to different estimates of realized spot market
volatility.
2 Relevant Literature
The traditional financial theory had initially conceptualized
derivative products as a
medium for risk sharing (Arrow, 1964; Ross, 1976). But later on,
these securities turned
out to be important vehicles for informed investors (Black,
1975; Grossman, 1977). The
body of literature inspecting whether informed traders use
directional information market
in the options market is quite extensive (Stephan & Whaley,
1990; Amin & Lee, 1997;
3
-
Easley, Hara, & Srinivas, 1998; Chan, Chung, & Fong,
2002; Chakravarty, Gulen, &
Mayhew, 2004; Cao, Chen, & Griffin, 2005; Pan &
Poteshman, 2006). Evidence clearly
suggests that informed traders choose the options market as
their preferred choice of
venue. Comparatively the literature on whether options market is
preferred for volatil-
ity information trading (Ni et al., 2008) is comparatively
scarce. Ni et al. (2008) show
that Vega-adjusted net trading volume can be used to measure
volatility demand for a
particular trader group. They also show that non-market maker’s
demand for volatility
is positively related to future realized volatility in the spot
market. Considering im-
plied volatility has strong predictive ability regarding future
realized volatility (Latane
& Rendleman, 1976; Chiras & Manaster, 1978; Beckers,
1981; Canina & Figlewski, 1993;
Lamoureux & Lastrapes, 1993; Jorion, 1995; Ederington &
Lee, 1996; Christensen &
Prabhala, 1998), the Ni et al. (2008) model controls for it.
The literature on algorithmic trading is comparatively new.
Research seems to sug-
gest that an increase in algorithmic trading activity is related
to a decrease in arbitrage
opportunity and an increase in informational efficiency,
primarily by speeding up price
discovery (J. A. Brogaard, 2010; Chaboud, Chiquoine,
Hjalmarsson, & Vega, 2014). Al-
gorithmic or machine trading also increases the adverse
selection cost for slower traders.
The direction of trading of the HFTs is correlated with public
information (J. Brogaard,
Hendershott, & Riordan, 2014). Algorithmic traders react
faster to events (Hendershott
& Riordan, 2013). Return volatilities have increased since
the introduction of algorith-
mic trading (Kelejian & Mukerji, 2016), raising concerns
whether algorithmic and more
specifically HFT increases systemic risk (Jain, Jain, &
McInish, 2016).
3 Volatility Information Trading
Investors with access to private information regarding future
volatility are likely to take
positions in options contract that are positively related to
future realized volatility. Ni et
al. (2008) extends the literature on the relation of option
volume and future volatility and
4
-
show that non-market maker’s demand for volatility is positively
related to future real-
ized volatility, indicating that non-market makers trade on
private information related to
future volatility. Order-driven markets do not have any
designated market maker. Limit
orders from various market participants are matched to each
other by the exchange match-
ing engine. However, in recent times algorithmic traders, and
more specifically HFTs have
assumed the role of modern market makers. Unlike the traditional
market makers, they
are not obliged to provide quotes at all times. As such, it
might be expected that the
behavior of algorithmic traders should resemble that of
traditional market makers, while
non-algorithmic traders behave like non-market makers. Our
testable hypothesis with
respect to the information content of non-algorithmic traders’
demand for volatility can
be framed as -
Hypothesis 1 In an order-driven market, non-algorithmic traders’
demand for volatility
in the stock options market is positively related to future
realized volatility in the spot
market.
Corporate announcements create increase information asymmetry in
the market, with
market participants with access to private information able to
leverage that information
earlier compared to others. The situations result in volatility
spikes. Ni et al. (2008)
use earnings announcement as exogenous shocks to exploit the
time-varying nature of
information asymmetry. In periods leading to the corporate
announcements, informed
investors are likely to use volatility information in the
options market. We argue that
similar to pre-scheduled earnings announcements, un-scheduled
corporate announcements
create similar situations of information asymmetry. As such
trading volume of informed
investors prior to any corporate announcement should convey
additional information.
Hypothesis 2 Investors trading on volatility related information
in the stock options
market behave similarly in periods leading up to both scheduled
and unscheduled corporate
announcements.
5
-
Algorithmic traders are not expected to homogeneous in their
behavior. The motiva-
tion for proprietary and agency algorithmic traders are very
different. The proprietary
algorithmic traders, who primarily engage in high-frequency
trading, try to use their
advantage of speed to exploit any arbitrage opportunity existing
in the market. They
are day-traders, who rarely carry over inventory. On the other
hand, agency algorithmic
traders execute trades on someone else’s behalf. Their primary
role is to split orders in
such a way that the price impact is minimum. They also prevent
investors trading on
information from the risk of being front-run. As such, the
information content of insti-
tutional trades may not be present when the trade is executed
through algorithms. As
such we frame our final testable hypothesis as-
Hypothesis 3 Trades executed by both proprietary and agency
algorithmic traders in the
stock options market do not convey private information regarding
future realized volatility
in the spot market.
The demand for volatility of a particular trader-group (Ni et
al., 2008) can be esti-
mated through the net trading volume of that trader group in
call and put options con-
tracts. Unlike other financial contracts, both and put options
prices are positively affected
by increasing volatility. As such, investors with information of
increasing (decreasing)
volatility are likely to buy (sell) call and put options
contracts. Options contracts are
available in different expiry dates and strike prices. As such,
in order to construct the
aggregate measure of volatility demand, the net trading volume
in individual contracts
need to weighted by the contract Vegas 1. The volatility demand
D TGσi,t of a particular
trader group TG for i-th stock on t-th day can be expressed
as-
D TGσi,t =∑K
∑T
∂lnCK,Ti,t∂σi,t
(BuyCall TGK,Ti,t − SellCall TGK,Ti,t )
+∑K
∑T
∂lnPK,Ti,t∂σi,t
(BuyPut TGK,Ti,t − SellPut TGK,Ti,t )
(1)
1Vega for a options contract is defined as the rate of change of
options price with respect to changein volatility
6
-
Where CK,Ti,t is the price of the call on underlying stock i at
time t with strike price
K and maturity T ; PK,Ti,t is the price for similar put options;
σi,t is the volatility of un-
derlying stock i at time t; BuyCall TGK,Ti,t is the number of
call contracts purchased
by the trader group TG on day t on underlying stock i with
strike price K and ma-
turity T ; and SellCall TGK,Ti,t , BuyPut TGK,Ti,t and SellPut
TG
K,Ti,t are the analogous
quantities for, respectively, the sale of calls and the purchase
and sale of puts by the
trader group TG. For empirical calculations, the partial
derivatives are difficult to com-
pute and hence, (∂lnCK,Ti,t /∂σi,t) is approximated by
(1/CK,Ti,t ).BlackScholesCallV ega
K,Ti,t
and (∂lnPK,Ti,t /∂σi,t) is approximated by (1/PK,Ti,t
).BlackScholesPutV ega
K,Ti,t . We esti-
mate the Vega using 20-day rolling realized volatility measure
based on the Andersen,
Bollerslev, Diebold, and Ebens (2001) measure of realized
volatility 2.
We relate this volatility demand to future realized volatility
in the spot market. Due to
the GARCH type clustering of realized volatility, we control for
lagged realized volatility
up to 5 trading days. We also control for lagged implied
volatility, as it is known to have
predictive ability about realized volatility. Other control
variables being traded volume in
the stock and traded volume in the options market. To eliminate
the problem of scaling,
we use the natural logarithm of the volume measures. We also
specifically control for
absolute value of the delta-weighted traded volume of the
particular traded group TG.
This term is analogous to the equivalent traded quantity in the
spot market.
Information asymmetry is supposed to increase prior to corporate
announcements. Ni
et al. (2008) control for the volatility spike due to
pre-scheduled earnings announcements.
In order to accommodate this, Ni et al. (2008) use dummies for
earnings announcements
as well as interaction terms. The actual empirical specification
for estimating the infor-
mativeness of different trader groups for future volatility is
as follows-
2We also run a robustness test using the sample volatility of
sixty trading days leading up to t forcomputation of the Black
Scholes Vega similar to the one used in Ni et al. (2008). The
results arequalitatively similar. (Results not reported here)
7
-
OneDayRVi,t =α + β1.D TGσi,t−j + β2.D TG
σi,t−j.EADi,t
+ β3.OneDayRVi,t−1 + β4.OneDayRVi,t−1.EADi,t
+ β5.OneDayRVi,t−2 + β6.OneDayRVi,t−2.EADi,t
+ β7.OneDayRVi,t−3 + β8.OneDayRVi,t−3.EADi,t
+ β9.OneDayRVi,t−4 + β10.OneDayRVi,t−4.EADi,t
+ β11.OneDayRVi,t−5 + β12.OneDayRVi,t−5.EADi,t
+ β13.EADi,t + β14.IVi,t−1 + β15.IVi,t−1.EADi,t + β16.abs(D
TG∆i,t−j)
+ β17.abs(D TG∆i,t−j).EADi,t + β18.ln(optV olumei,t−j)
+ β19.ln(optV olumei,t−j).EADi,t + β20.ln(stkV olumei,t−j)
+ β21.ln(stkV olumei,t−j).EADi,t + �i,t
(2)
where OneDayRVi,t is the volatility of the underlying security i
on day t. EADi,t is
a proxy which takes up the value of 1 if date t is an corporate
announcement date for
security i, 0 otherwise. IVi,t is the average implied volatility
of the ATM3 Call and Put
option contract for the security i with shortest maturity on
date t. abs(D TG∆i,t) is the
absolute value of the delta adjusted options market net traded
volume across all expiry
dates and strike prices for the trader group TG for security i
on date t. We scale down the
values of the variables abs(D TG∆i,t) by a factor of one
million. ln(stkV olumei,t) and and
ln(optV olumei,t) are the natural logarithm of the spot and and
options market traded
volume respectively for security i on day t. We estimate the
equation for different values
of j = 1, 2, 3, 4, 5 to interpret about the predictive ability
of volatility demand for j days
ahead realized volatility.
We argue that the same model may be used in case of unscheduled
corporate an-
nouncements also. We use a modified model that uses dummy UAD
for unscheduled
corporate announcements instead of earnings announcement
dummies. Similar to the
3ATM: At the Money contract
8
-
earlier specification for earnings announcement dummy, the
UADi,t is a proxy which
takes up the value of 1 if date t is an unscheduled corporate
announcement date for
security i, 0 otherwise.
OneDayRVi,t =α + β1.D TGσi,t−j + β2.D TG
σi,t−j.UADi,t
+ β3.OneDayRVi,t−1 + β4.OneDayRVi,t−1.UADi,t
+ β5.OneDayRVi,t−2 + β6.OneDayRVi,t−2.UADi,t
+ β7.OneDayRVi,t−3 + β8.OneDayRVi,t−3.UADi,t
+ β9.OneDayRVi,t−4 + β10.OneDayRVi,t−4.UADi,t
+ β11.OneDayRVi,t−5 + β12.OneDayRVi,t−5.UADi,t
+ β13.UADi,t + β14.IVi,t−1 + β15.IVi,t−1.UADi,t + β16.abs(D
TG∆i,t−j)
+ β17.abs(D TG∆i,t−j).UADi,t + β18.ln(optV olumei,t−j)
+ β19.ln(optV olumei,t−j).UADi,t + β20.ln(stkV olumei,t−j)
+ β21.ln(stkV olumei,t−j).UADi,t + �i,t
(3)
4 Data
For our analysis, we use six months (01 Jan 2015 to 30 Jun 2015)
of options market
trading data obtained from the NSE for 159 stocks 4. Our dataset
contains information
regarding 37 million transactions in the options market during
the period of 120 trading
days. We summarize this dataset to create daily demand for
volatility measures and other
control variables. For our analysis, we club algorithmic trades
executed by Custodian and
NCNP group into a single class of agency algorithmic traders.
Prop algorithmic traders
are our best available proxy for HFTs. Our dataset does not
provide estimates for implied
4Actual number of stocks permitted in the derivatives market
during the period was 160. Out of these,one stock did not have
sufficient number of observations at daily level to be included in
our analysis.
9
-
volatility. As such, we run optimization exercises to estimate
the implied volatility using
the options traded price and the Black-Scholes options pricing
model.
Table 1: Summary statistics of the variables used in analysis.
Volatility figures areexpressed in basis points (bps), where 100
bps = 1%
Variable Obs Mean Median Std Dev Min. Max.
OneDayRV [NSE Reported] 17772 267.66 252.54 87.02 103.91
1346.05OneDayRV [Anderson] 17772 208.26 189.94 99.15 55.68
5839.29OneDayRV [Alizadeh] 17772 343.01 297.94 370.62 70.60
42546.70Implied Vol. (Annualized) 17769 3863.51 3705.92 1156.84
1010.09 16476.62Volatility Demand (D Algoσ) 17772 -0.70 -0.13 7.32
-145.39 122.00Volatility Demand (D NAσ) 17772 0.70 0.13 7.32
-122.00 145.39Volatility Demand (D PAσ) 17772 -0.52 -0.07 5.38
-126.46 84.79Volatility Demand (D AAσ) 17772 -0.17 -0.04 4.20
-107.75 73.51
abs(D Algo∆) 17772 0.06 0.02 0.12 0.00 3.16
abs(D NA∆) 17772 0.06 0.02 0.12 0.00 3.16
abs(D PA∆) 17772 0.05 0.01 0.10 0.00 2.78
abs(D AA∆) 17772 0.02 0.01 0.04 0.00 0.88ln(Options Vol) 17772
13.50 13.84 2.11 4.83 19.51ln(Spot Vol) 17772 14.22 14.33 1.32 8.34
20.12
For the estimation of realized volatility, we use 3 alternative
definitions. For the
first definition, we use the daily volatility reported by NSE.
The exchange computes the
volatility using an exponentially weighted moving average (EWMA)
model as σi,t,NSE =√0.96 ∗ σ2i,t−1,NSE + 0.04 ∗ (ln
Closei,tOpeni,t
)2 5. Where σi,t,NSE is the volatility reported by NSE
for i -th security on t-th day, while Openi,t and Closei,t are
the the daily opening and
closing prices for the i -th security on t-th day.
The second definition is based on the method followed by
Andersen et al. (2001). In
this method, realized volatility is calculated from intra-day
returns of every five minutes
as σi,t,Anderson =√∑nt
k=1 (rk,t)2 where rk,t is the intra-day return of the k -th
five-minute
sub-period for the i -th security on t-th day.
The third and final definition is based on the method followed
by Alizadeh, Brandt,
and Diebold (2002). The same measure was used by Ni et al.
(2008). In this method,
realized volatility is calculated from daily high, low and
closing prices and estimated as
5or GARCH (1;1) model for the volatility index INDIAVIX
10
-
σi,t,Range =Highi,t−Lowi,t
Closei,twhere Highi,t, Lowi,t and Closei,t are the the daily
high, low and
closing prices for the i -th security on t-th day.
(a) Volatility Estimate (NSE Reported)
(b) Volatility Estimate (Anderson et. al. 2001)
(c) Volatility Estimate (Alizadeh et. al. 2002)
Figure 1: The figure plots average realized volatility around
earnings announcement.The x-axis represents the time line around
the pre-scheduled earnings announcement. 0represents the earnings
announcement date. negative values indicate trading days priorto
announcement and positive values indicate trading days post
announcement.
11
-
(a) Volatility Estimate (NSE Reported)
(b) Volatility Estimate (Anderson et. al. 2001)
(c) Volatility Estimate (Alizadeh et. al. 2002)
Figure 2: figure plots average realized volatility around
unscheduled corporate announce-ment. The x-axis represents the time
line around the corporate announcement. 0 repre-sents the
announcement date. negative values indicate trading days prior to
announce-ment and positive values indicate trading days post
announcement.
The earnings announcement data is obtained form Prowess database
by CMIE (Centre
12
-
for Monitoring Indian Economy). We consider both quarterly as
well as annual earnings
announcements. During our sample period, we have 269
observations of earnings an-
nouncements for our selected list of companies.
For unscheduled corporate announcements we consider the
following corporate actions
- M&A announcement, share buyback, stock split, bonus issue
(stock dividend), joint
venture announcements, special dividend (cash), reverse-split
(consolidation), demerger,
bankruptcy & delisting. We obtain data for the same from the
Thomson Eikon database.
Our dataset consists of 88 such events of unscheduled corporate
announcements.
The plots for average volatility around the announcement dates
depict a clear pattern.
In case of earnings announcement (Fig. 1), the volatility has
spikes on Day 0 (announce-
ment date) and Day 1 (one day after announcement date). This
empirical observation
may be explained due to the nature of the announcement. Most of
these earnings an-
nouncement information come post market hours, which explains
the high volatility on
the next trading day. In case of an unscheduled announcement
(Fig. 2), however, the
information usually comes within market hours, resulting in
prominent volatility spike
only on Day 0 6. Also, we can notice how the volatility
definition affects the shape of
the plot. In case of exchange (NSE) reported volatility, it
seems that the high level of
volatility post announcement is persistent for several trading
days. It is primarily due
to the high weight given to the one-day lagged volatility for
computation of present-day
volatility.
6For a sub-sample of out dataset, where the time stamp of the
news related to the announcement wasavailable, around 70% of the
news item were timed before market closing hours.
13
-
5 Results
For our first set of models, we run fixed effect panel models,
regressing the one-day real-
ized volatility on volatility-demand measures for algorithmic as
well as non-algorithmic
traders. Econometric tests suggest that fixed-effect models fit
the data better than pooled
model used by Ni et al. (2008). We use all three definitions of
realized volatility - NSE
reported volatility (Table 2 & 3), volatility computed using
intraday returns (Andersen
et al., 2001) (Table 4 & 5) and volatility computed by range
estimator (Alizadeh et al.,
2002) (Table 6 & 7). For each definition of realized
volatility, we run separate models us-
ing dummies for pre-scheduled earnings announcements (Table 2, 4
& 6) and unscheduled
corporate announcements (Table 3, 5 & 7).
Each table consists of two panels, where we differentiate our
trader group (TG) as
algorithmic and non-algorithmic traders. By definition, the
volatility-demand measures
(D TGσ) for algorithmic and non-algorithmic traders are equal in
magnitude and opposite
in sign. The absolute value of delta-adjusted traded volume
(abs(D TG∆)) of these
two trader groups will also be same by construction. As such the
two panels exhibit
exactly same results except for the coefficients corresponding
to volatility demand of
these two groups, which have same magnitude but opposite sign.
Apart from the trader-
group (TG) specific terms, we also report the coefficients
corresponding to lagged realized
volatility measures, dummies for announcement and the
interaction terms. Due to space
constraint, we do not report coefficients corresponding to the
additional control variables.
While positive values for the coefficients corresponding to
volatility demand represent
informativeness of the trader group, the negative sign indicates
that the counterparty is
informed.
We vary the value of the parameter j in order to measure the
predictive ability of the
volatility demand. The interaction terms with the announcement
dummies interpret addi-
tional information content prior to announcements. Consistent
with our first hypothesis,
we find that the volatility-demand for non-algorithmic traders
has positive relation with
14
-
Tab
le2:
Res
ult
sof
fixed
effec
tpan
elre
gres
sion
model
tote
stvo
lati
lity
info
rmat
ion
trad
ing
by
algo
rith
mic
and
non
-alg
orit
hm
ictr
ader
sin
the
NSE
opti
ons
mar
ket
contr
olling
for
sched
ule
dea
rnin
gsan
nou
nce
men
ts.
Mea
sure
ofvo
lati
lity
(RV
):Sto
ckvo
lati
lity
rep
orte
dby
NSE
[Sqrt
(0.9
4∗PrevDayVolatility
2+
0.06
∗SameDayReturn
2)
orE
WM
Am
odel
]
jConst.
DTG
σO
neD
ayRV
EA
D
abs(D
TG
∆)
ModelR
2
(t-j)
(t-j)
(t-1
)(t-1
)(t-2
)(t-2
)(t-3
)(t-3
)(t-4
)(t-4
)(t-5
)(t-5
)(t-j)
(t-j)
*EA
D*EA
D*EA
D*EA
D*EA
D*EA
D*EA
D
Trader
Group:
Alg
orithm
icTrader
1-7
.3*
-0.07***
-0.55***
0.95***
0.19**
-0.01
0.05
-0.01
0.13
0.02*
-0.26**
-0.03***
-0.14*
-10.87
-0.25
-4.91
0.9576
(-1.65)
(-3.36)
(-5.55)
(111.41)
(2.14)
(-1.13)
(0.4)
(-0.63)
(0.89)
(1.78)
(-1.98)
(-3.64)
(-1.7)
(-0.68)
(-0.17)
(-0.7)
27.94*
-0.04*
-0.45**
0.96***
0.15*
-0.01
0.09
-0.01
0.26*
0.02*
-0.4***
-0.03***
-0.15*
21.35
-3.39**
1.17
0.9572
(1.8)
(-1.89)
(-2.35)
(123.9)
(1.79)
(-1.13)
(0.72)
(-1.35)
(1.83)
(1.8)
(-3)
(-3.76)
(-1.8)
(1.31)
(-2.26)
(0.06)
36.17
-0.02
-0.41**
0.96***
0.14*
-0.02*
0.08
-0.01
0.21
0.02*
-0.34**
-0.03***
-0.13
17.09
0.54
-6.61
0.9572
(1.4)
(-0.9)
(-2.02)
(124.37)
(1.65)
(-1.75)
(0.64)
(-1.2)
(1.45)
(1.87)
(-2.57)
(-3.49)
(-1.62)
(1)
(0.36)
(-0.49)
410.26**
-0.04*
-0.35**
0.96***
0.16*
-0.02*
0.08
-0.01
0.16
0.01
-0.33**
-0.02**
-0.12
-3.86
4.27***
-11.99
0.9572
(2.37)
(-1.86)
(-2.11)
(124.28)
(1.85)
(-1.72)
(0.66)
(-1.04)
(1.1)
(1.01)
(-2.48)
(-2.54)
(-1.5)
(-0.23)
(2.82)
(-0.95)
513.63***
-0.02
-0.3*
0.96***
0.16*
-0.02*
0.12
-0.01
0.19
0.02*
-0.38***
-0.03***
-0.13
13.25
1-2
4.67*
0.9572
(3.24)
(-0.81)
(-1.9)
(124.2)
(1.88)
(-1.69)
(0.9)
(-1)
(1.35)
(1.83)
(-2.93)
(-3.66)
(-1.6)
(0.79)
(0.69)
(-1.81)
Trader
Group:
Non-A
lgorithm
icTrader
1-7
.3*
0.07***
0.55***
0.95***
0.19**
-0.01
0.05
-0.01
0.13
0.02*
-0.26**
-0.03***
-0.14*
-10.87
-0.25
-4.91
0.9576
(-1.65)
(3.36)
(5.55)
(111.41)
(2.14)
(-1.13)
(0.4)
(-0.63)
(0.89)
(1.78)
(-1.98)
(-3.64)
(-1.7)
(-0.68)
(-0.17)
(-0.7)
27.94*
0.04*
0.45**
0.96***
0.15*
-0.01
0.09
-0.01
0.26*
0.02*
-0.4***
-0.03***
-0.15*
21.35
-3.39**
1.17
0.9572
(1.8)
(1.89)
(2.35)
(123.9)
(1.79)
(-1.13)
(0.72)
(-1.35)
(1.83)
(1.8)
(-3)
(-3.76)
(-1.8)
(1.31)
(-2.26)
(0.06)
36.17
0.02
0.41**
0.96***
0.14*
-0.02*
0.08
-0.01
0.21
0.02*
-0.34**
-0.03***
-0.13
17.09
0.54
-6.61
0.9572
(1.4)
(0.9)
(2.02)
(124.37)
(1.65)
(-1.75)
(0.64)
(-1.2)
(1.45)
(1.87)
(-2.57)
(-3.49)
(-1.62)
(1)
(0.36)
(-0.49)
410.26**
0.04*
0.35**
0.96***
0.16*
-0.02*
0.08
-0.01
0.16
0.01
-0.33**
-0.02**
-0.12
-3.86
4.27***
-11.99
0.9572
(2.37)
(1.86)
(2.11)
(124.28)
(1.85)
(-1.72)
(0.66)
(-1.04)
(1.1)
(1.01)
(-2.48)
(-2.54)
(-1.5)
(-0.23)
(2.82)
(-0.95)
513.63***
0.02
0.3*
0.96***
0.16*
-0.02*
0.12
-0.01
0.19
0.02*
-0.38***
-0.03***
-0.13
13.25
1-2
4.67*
0.9572
(3.24)
(0.81)
(1.9)
(124.2)
(1.88)
(-1.69)
(0.9)
(-1)
(1.35)
(1.83)
(-2.93)
(-3.66)
(-1.6)
(0.79)
(0.69)
(-1.81)
tstatisticsin
pare
nth
ese
s*
p<
0.10,**
p<
0.05,***
p<
0.01
15
-
Tab
le3:
Res
ult
sof
fixed
effec
tpan
elre
gres
sion
model
tote
stvo
lati
lity
info
rmat
ion
trad
ing
by
algo
rith
mic
and
non
-alg
orit
hm
ictr
ader
sin
the
NSE
opti
ons
mar
ket
contr
olling
for
unsc
hed
ule
dco
rpor
ate
annou
nce
men
ts.
Mea
sure
ofvo
lati
lity
(RV
):Sto
ckvo
lati
lity
rep
orte
dby
NSE
[Sqrt
(0.9
4∗PrevDayVolatility
2+
0.06
∗SameDayReturn
2)
orE
WM
Am
odel
]
jConst.
DTG
σO
neD
ayRV
UA
D
abs(D
TG
∆)
ModelR
2
(t-j)
(t-j)
(t-1
)(t-1
)(t-2
)(t-2
)(t-3
)(t-3
)(t-4
)(t-4
)(t-5
)(t-5
)(t-j)
(t-j)
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
Trader
Group:
Alg
orithm
icTrader
1-6
.81
-0.09***
-0.28*
0.94***
1.01***
-0.01
-1.03***
-0.01
0.39
0.02**
-0.27
-0.03***
-0.04
26.68
-0.2
-11.97
0.9575
(-1.54)
(-4.8)
(-1.95)
(110.84)
(8.11)
(-0.65)
(-3.98)
(-0.81)
(1.47)
(2.04)
(-1.43)
(-3.95)
(-0.32)
(0.87)
(-0.13)
(-0.74)
28.93**
-0.04*
-0.25
0.96***
1.04***
-0.01
-1.46***
-0.02
0.81***
0.02**
-0.31
-0.03***
-0.03
3.6
-3.7**
-5.59
0.9572
(2.02)
(-1.9)
(-1.55)
(123.27)
(8.52)
(-0.48)
(-5.72)
(-1.61)
(2.98)
(1.97)
(-1.61)
(-4.06)
(-0.25)
(0.12)
(-2.46)
(-0.28)
36.85
-0.02
-0.29
0.96***
1.03***
-0.01
-1.38***
-0.01
0.81***
0.02**
-0.4**
-0.03***
-0.01
9.22
0.51
-16.03
0.9572
(1.56)
(-1.09)
(-1.51)
(123.71)
(8.39)
(-1.24)
(-5.68)
(-1.25)
(3.08)
(2.07)
(-2.02)
(-3.83)
(-0.05)
(0.36)
(0.34)
(-1.02)
411.2***
-0.04**
0.26
0.96***
1.02***
-0.01
-1.25***
-0.01
0.67***
0.02
-0.35*
-0.02***
-0.04
30.42
4.4***
12.58
0.9572
(2.58)
(-2.25)
(1.05)
(123.61)
(8.23)
(-1.21)
(-4.97)
(-1.22)
(2.65)
(1.36)
(-1.83)
(-2.91)
(-0.32)
(1)
(2.91)
(0.53)
513.44***
-0.02
0.18
0.96***
1.03***
-0.01
-1.23***
-0.01
0.63**
0.02**
-0.34*
-0.03***
-0.02
39.01
11.44
0.9572
(3.2)
(-0.96)
(0.81)
(123.54)
(8.28)
(-1.2)
(-5.1)
(-1.16)
(2.52)
(2.06)
(-1.81)
(-3.95)
(-0.13)
(1.26)
(0.68)
(0.09)
Trader
Group:
Non-A
lgorithm
icTrader
1-6
.81
0.09***
0.28*
0.94***
1.01***
-0.01
-1.03***
-0.01
0.39
0.02**
-0.27
-0.03***
-0.04
26.68
-0.2
-11.97
0.9575
(-1.54)
(4.8)
(1.95)
(110.84)
(8.11)
(-0.65)
(-3.98)
(-0.81)
(1.47)
(2.04)
(-1.43)
(-3.95)
(-0.32)
(0.87)
(-0.13)
(-0.74)
28.93**
0.04*
0.25
0.96***
1.04***
-0.01
-1.46***
-0.02
0.81***
0.02**
-0.31
-0.03***
-0.03
3.6
-3.7**
-5.59
0.9572
(2.02)
(1.9)
(1.55)
(123.27)
(8.52)
(-0.48)
(-5.72)
(-1.61)
(2.98)
(1.97)
(-1.61)
(-4.06)
(-0.25)
(0.12)
(-2.46)
(-0.28)
36.85
0.02
0.29
0.96***
1.03***
-0.01
-1.38***
-0.01
0.81***
0.02**
-0.4**
-0.03***
-0.01
9.22
0.51
-16.03
0.9572
(1.56)
(1.09)
(1.51)
(123.71)
(8.39)
(-1.24)
(-5.68)
(-1.25)
(3.08)
(2.07)
(-2.02)
(-3.83)
(-0.05)
(0.36)
(0.34)
(-1.02)
411.2***
0.04**
-0.26
0.96***
1.02***
-0.01
-1.25***
-0.01
0.67***
0.02
-0.35*
-0.02***
-0.04
30.42
4.4***
12.58
0.9572
(2.58)
(2.25)
(-1.05)
(123.61)
(8.23)
(-1.21)
(-4.97)
(-1.22)
(2.65)
(1.36)
(-1.83)
(-2.91)
(-0.32)
(1)
(2.91)
(0.53)
513.44***
0.02
-0.18
0.96***
1.03***
-0.01
-1.23***
-0.01
0.63**
0.02**
-0.34*
-0.03***
-0.02
39.01
11.44
0.9572
(3.2)
(0.96)
(-0.81)
(123.54)
(8.28)
(-1.2)
(-5.1)
(-1.16)
(2.52)
(2.06)
(-1.81)
(-3.95)
(-0.13)
(1.26)
(0.68)
(0.09)
tstatisticsin
pare
nth
ese
s*
p<
0.10,**
p<
0.05,***
p<
0.01
16
-
Tab
le4:
Res
ult
sof
fixed
effec
tpan
elre
gres
sion
model
tote
stvo
lati
lity
info
rmat
ion
trad
ing
by
algo
rith
mic
and
non
-alg
orit
hm
ictr
ader
sin
the
NSE
opti
ons
mar
ket
contr
olling
for
sched
ule
dea
rnin
gsan
nou
nce
men
ts.
Mea
sure
ofvo
lati
lity
(RV
):A
nder
son
(200
1),
esti
mat
eof
real
ized
vola
tility
usi
ng
intr
a-day
five
-min
ute
retu
rnof
the
secu
rity
.
jConst.
DTG
σO
neD
ayRV
EA
D
abs(D
TG
∆)
ModelR
2
(t-j)
(t-j)
(t-1
)(t-1
)(t-2
)(t-2
)(t-3
)(t-3
)(t-4
)(t-4
)(t-5
)(t-5
)(t-j)
(t-j)
*EA
D*EA
D*EA
D*EA
D*EA
D*EA
D*EA
D
Trader
Group:
Alg
orithm
icTrader
1-8
6.54***
-0.35***
-1.7***
0.19***
0.04
0.06***
0.28***
0.07***
0.01
0.04***
0.32***
0.01
-0.5***
191.92***
28.88***
63.72**
0.3608
(-4.4)
(-4.03)
(-3.95)
(22.66)
(0.45)
(7.68)
(2.71)
(9.71)
(0.09)
(5.62)
(3.11)
(1.61)
(-5.72)
(2.7)
(4.43)
(2.08)
237.64*
-0.21**
1.02
0.22***
0.02
0.06***
0.31***
0.07***
0.02
0.04***
0.31***
0.01
-0.52***
251.15***
-2.51
159.78*
0.3551
(1.91)
(-2.49)
(1.21)
(28.81)
(0.21)
(6.93)
(3.14)
(9.41)
(0.23)
(5.58)
(3.04)
(1.59)
(-5.95)
(3.43)
(-0.39)
(1.74)
334.9*
-0.02
-1.53*
0.22***
0.04
0.06***
0.23**
0.07***
0.03
0.04***
0.31***
0.01
-0.51***
160.87**
-12.46*
97.34*
0.3549
(1.78)
(-0.25)
(-1.69)
(29.02)
(0.43)
(8.14)
(2.31)
(8.38)
(0.39)
(5.39)
(3.01)
(1.44)
(-5.79)
(2.09)
(-1.92)
(1.65)
432.43*
-0.04
-2.33***
0.22***
0.04
0.06***
0.23**
0.07***
00.03***
0.31***
0.01
-0.5***
144.94*
18.41***
88.62
0.3557
(1.65)
(-0.44)
(-3.2)
(29.07)
(0.49)
(8.1)
(2.28)
(8.92)
(-0.02)
(3.92)
(3.02)
(1.48)
(-5.66)
(1.95)
(2.84)
(1.64)
590.19***
-0.03
-0.27
0.22***
0.03
0.06***
0.28***
0.07***
-0.02
0.04***
0.33***
0.02*
-0.56***
74.9
3.74
29.04
0.3542
(4.61)
(-0.41)
(-0.39)
(28.89)
(0.29)
(8.13)
(2.83)
(9.21)
(-0.21)
(5.71)
(3.17)
(1.95)
(-6.11)
(0.99)
(0.58)
(0.48)
Trader
Group:
Non-A
lgorithm
icTrader
1-8
6.54***
0.35***
1.7***
0.19***
0.04
0.06***
0.28***
0.07***
0.01
0.04***
0.32***
0.01
-0.5***
191.92***
28.88***
63.72**
0.3608
(-4.4)
(4.03)
(3.95)
(22.66)
(0.45)
(7.68)
(2.71)
(9.71)
(0.09)
(5.62)
(3.11)
(1.61)
(-5.72)
(2.7)
(4.43)
(2.08)
237.64*
0.21**
-1.02
0.22***
0.02
0.06***
0.31***
0.07***
0.02
0.04***
0.31***
0.01
-0.52***
251.15***
-2.51
159.78*
0.3551
(1.91)
(2.49)
(-1.21)
(28.81)
(0.21)
(6.93)
(3.14)
(9.41)
(0.23)
(5.58)
(3.04)
(1.59)
(-5.95)
(3.43)
(-0.39)
(1.74)
334.9*
0.02
1.53*
0.22***
0.04
0.06***
0.23**
0.07***
0.03
0.04***
0.31***
0.01
-0.51***
160.87**
-12.46*
97.34*
0.3549
(1.78)
(0.25)
(1.69)
(29.02)
(0.43)
(8.14)
(2.31)
(8.38)
(0.39)
(5.39)
(3.01)
(1.44)
(-5.79)
(2.09)
(-1.92)
(1.65)
432.43*
0.04
2.33***
0.22***
0.04
0.06***
0.23**
0.07***
00.03***
0.31***
0.01
-0.5***
144.94*
18.41***
88.62
0.3557
(1.65)
(0.44)
(3.2)
(29.07)
(0.49)
(8.1)
(2.28)
(8.92)
(-0.02)
(3.92)
(3.02)
(1.48)
(-5.66)
(1.95)
(2.84)
(1.64)
590.19***
0.03
0.27
0.22***
0.03
0.06***
0.28***
0.07***
-0.02
0.04***
0.33***
0.02*
-0.56***
74.9
3.74
29.04
0.3542
(4.61)
(0.41)
(0.39)
(28.89)
(0.29)
(8.13)
(2.83)
(9.21)
(-0.21)
(5.71)
(3.17)
(1.95)
(-6.11)
(0.99)
(0.58)
(0.48)
tstatisticsin
pare
nth
ese
s*
p<
0.10,**
p<
0.05,***
p<
0.01
17
-
Tab
le5:
Res
ult
sof
fixed
effec
tpan
elre
gres
sion
model
tote
stvo
lati
lity
info
rmat
ion
trad
ing
by
algo
rith
mic
and
non
-alg
orit
hm
ictr
ader
sin
the
NSE
opti
ons
mar
ket
contr
olling
for
unsc
hed
ule
dco
rpor
ate
annou
nce
men
ts.
Mea
sure
ofvo
lati
lity
(RV
):A
nder
son
(200
1),
esti
mat
eof
real
ized
vola
tility
usi
ng
intr
a-day
five
-min
ute
retu
rnof
the
secu
rity
.
jConst.
DTG
σO
neD
ayRV
UA
D
abs(D
TG
∆)
ModelR
2
(t-j)
(t-j)
(t-1
)(t-1
)(t-2
)(t-2
)(t-3
)(t-3
)(t-4
)(t-4
)(t-5
)(t-5
)(t-j)
(t-j)
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
Trader
Group:
Alg
orithm
icTrader
1-6
6.55***
-0.39***
-4.73***
0.18***
0.42***
0.05***
1.95***
0.07***
1.36***
0.03***
3.35***
0.01**
-3.44***
451.68***
32.41***
182.3***
0.4283
(-3.58)
(-4.84)
(-8.65)
(22.74)
(2.9)
(6.43)
(13.14)
(9.24)
(8.11)
(4.47)
(19.41)
(2.05)
(-19.48)
(3.54)
(5.34)
(2.67)
262.89***
-0.12
-5.67***
0.21***
0.5***
0.05***
1.87***
0.07***
1.34***
0.03***
3.44***
0.01**
-3.5***
-494.68***
-3.07
-542.03***
0.4187
(3.37)
(-1.45)
(-9.02)
(28.6)
(3.38)
(6.64)
(12.1)
(9)
(7.82)
(4.37)
(19.95)
(2)
(-19.48)
(-3.91)
(-0.5)
(-6.43)
352.35***
-0.01
-5.45***
0.21***
0.6***
0.05***
1.99***
0.07***
0.89***
0.03***
3.72***
0.01*
-3.63***
-577.86***
-10.4*
-451.66***
0.4189
(2.81)
(-0.16)
(-6.92)
(28.66)
(4.17)
(6.93)
(13.25)
(8.55)
(5.06)
(4.34)
(21.06)
(1.91)
(-19.99)
(-5.2)
(-1.69)
(-6.75)
455.54***
-0.09
-0.8
0.21***
0.26*
0.05***
1.94***
0.06***
1.57***
0.03***
3.51***
0.01**
-3.73***
78.89
23.56***
-764.88***
0.4192
(2.99)
(-1.1)
(-0.85)
(28.68)
(1.77)
(6.75)
(13.12)
(8.46)
(9.3)
(3.48)
(20.32)
(2.07)
(-20.86)
(0.62)
(3.83)
(-7.75)
589.24***
0.01
-0.01
0.21***
0.38***
0.05***
1.94***
0.06***
1.43***
0.03***
3.55***
0.02**
-3.82***
-225.28*
5.44
-323.53***
0.4161
(4.8)
(0.07)
(-0.01)
(28.43)
(2.59)
(6.86)
(13.06)
(8.6)
(8.4)
(4.61)
(20.64)
(2.56)
(-20.27)
(-1.72)
(0.88)
(-5)
Trader
Group:
Non-A
lgorithm
icTrader
1-6
6.55***
0.39***
4.73***
0.18***
0.42***
0.05***
1.95***
0.07***
1.36***
0.03***
3.35***
0.01**
-3.44***
451.68***
32.41***
182.3***
0.4283
(-3.58)
(4.84)
(8.65)
(22.74)
(2.9)
(6.43)
(13.14)
(9.24)
(8.11)
(4.47)
(19.41)
(2.05)
(-19.48)
(3.54)
(5.34)
(2.67)
262.89***
0.12
5.67***
0.21***
0.5***
0.05***
1.87***
0.07***
1.34***
0.03***
3.44***
0.01**
-3.5***
-494.68***
-3.07
-542.03***
0.4187
(3.37)
(1.45)
(9.02)
(28.6)
(3.38)
(6.64)
(12.1)
(9)
(7.82)
(4.37)
(19.95)
(2)
(-19.48)
(-3.91)
(-0.5)
(-6.43)
352.35***
0.01
5.45***
0.21***
0.6***
0.05***
1.99***
0.07***
0.89***
0.03***
3.72***
0.01*
-3.63***
-577.86***
-10.4*
-451.66***
0.4189
(2.81)
(0.16)
(6.92)
(28.66)
(4.17)
(6.93)
(13.25)
(8.55)
(5.06)
(4.34)
(21.06)
(1.91)
(-19.99)
(-5.2)
(-1.69)
(-6.75)
455.54***
0.09
0.8
0.21***
0.26*
0.05***
1.94***
0.06***
1.57***
0.03***
3.51***
0.01**
-3.73***
78.89
23.56***
-764.88***
0.4192
(2.99)
(1.1)
(0.85)
(28.68)
(1.77)
(6.75)
(13.12)
(8.46)
(9.3)
(3.48)
(20.32)
(2.07)
(-20.86)
(0.62)
(3.83)
(-7.75)
589.24***
-0.01
0.01
0.21***
0.38***
0.05***
1.94***
0.06***
1.43***
0.03***
3.55***
0.02**
-3.82***
-225.28*
5.44
-323.53***
0.4161
(4.8)
(-0.07)
(0.01)
(28.43)
(2.59)
(6.86)
(13.06)
(8.6)
(8.4)
(4.61)
(20.64)
(2.56)
(-20.27)
(-1.72)
(0.88)
(-5)
tstatisticsin
pare
nth
ese
s*
p<
0.10,**
p<
0.05,***
p<
0.01
18
-
Tab
le6:
Res
ult
sof
fixed
effec
tpan
elre
gres
sion
model
tote
stvo
lati
lity
info
rmat
ion
trad
ing
by
algo
rith
mic
and
non
-alg
orit
hm
ictr
ader
sin
the
NSE
opti
ons
mar
ket
contr
olling
for
sched
ule
dea
rnin
gsan
nou
nce
men
ts.
Mea
sure
ofvo
lati
lity
(RV
):A
liza
deh
(200
2),
esti
mat
eof
real
ized
vola
tility
com
pute
dth
rough
diff
eren
ceb
etw
een
the
stock
’sin
trad
ayhig
han
dlo
wpri
cediv
ided
by
the
clos
ing
stock
pri
ce.
jConst.
DTG
σO
neD
ayRV
EA
D
abs(D
TG
∆)
ModelR
2
(t-j)
(t-j)
(t-1
)(t-1
)(t-2
)(t-2
)(t-3
)(t-3
)(t-4
)(t-4
)(t-5
)(t-5
)(t-j)
(t-j)
*EA
D*EA
D*EA
D*EA
D*EA
D*EA
D*EA
D
Trader
Group:
Alg
orithm
icTrader
1-5
38.61***
-1.54***
-3.31*
0.02**
0.16
0.03***
0.22
0.04***
-0.02
0.02**
0.18
-0.01
-0.25
222.62
43.88
75.19
0.0718
(-6.29)
(-3.92)
(-1.71)
(2.56)
(1.01)
(3.54)
(1.29)
(5.46)
(-0.1)
(2.01)
(1.05)
(-1.62)
(-1.6)
(0.7)
(1.49)
(0.55)
2-2
17.8**
-1.3***
1.3
0.04***
0.25
0.02***
0.2
0.04***
-0.04
0.02**
0.18
-0.01
-0.28*
296.21
-22.99
236.81
0.067
(-2.55)
(-3.37)
(0.34)
(5.03)
(1.62)
(3.02)
(1.19)
(4.97)
(-0.21)
(2.14)
(1.09)
(-1.55)
(-1.81)
(0.91)
(-0.79)
(0.57)
3-1
58.24*
-0.32
-5.56
0.04***
0.26*
0.03***
0.19
0.03***
-0.08
0.01*
0.17
-0.01
-0.27*
202.93
2.64
192.44
0.0661
(-1.86)
(-0.82)
(-1.37)
(5.26)
(1.74)
(4.27)
(1.14)
(4.24)
(-0.47)
(1.73)
(1.01)
(-1.55)
(-1.68)
(0.59)
(0.09)
(0.71)
4-1
05.76
-0.39
-3.07
0.04***
0.24
0.03***
0.19
0.04***
-0.04
0.01
0.12
-0.01*
-0.27*
-25.4
45.56
35.61
0.0659
(-1.25)
(-1)
(-0.93)
(5.26)
(1.59)
(4.44)
(1.1)
(5.24)
(-0.24)
(0.93)
(0.71)
(-1.84)
(-1.68)
(-0.08)
(1.55)
(0.14)
5146.99*
-0.49
-0.02
0.04***
0.25
0.03***
0.2
0.04***
-0.05
0.02**
0.2
-0.01
-0.3*
103.65
38.85
50.8
0.0648
(1.74)
(-1.27)
(0)
(5.24)
(1.64)
(4.52)
(1.22)
(5.49)
(-0.31)
(2.15)
(1.17)
(-1.62)
(-1.88)
(0.31)
(1.33)
(0.18)
Trader
Group:
Non-A
lgorithm
icTrader
1-5
38.61***
1.54***
3.31*
0.02**
0.16
0.03***
0.22
0.04***
-0.02
0.02**
0.18
-0.01
-0.25
222.62
43.88
75.19
0.0718
(-6.29)
(3.92)
(1.71)
(2.56)
(1.01)
(3.54)
(1.29)
(5.46)
(-0.1)
(2.01)
(1.05)
(-1.62)
(-1.6)
(0.7)
(1.49)
(0.55)
2-2
17.8**
1.3***
-1.3
0.04***
0.25
0.02***
0.2
0.04***
-0.04
0.02**
0.18
-0.01
-0.28*
296.21
-22.99
236.81
0.067
(-2.55)
(3.37)
(-0.34)
(5.03)
(1.62)
(3.02)
(1.19)
(4.97)
(-0.21)
(2.14)
(1.09)
(-1.55)
(-1.81)
(0.91)
(-0.79)
(0.57)
3-1
58.24*
0.32
5.56
0.04***
0.26*
0.03***
0.19
0.03***
-0.08
0.01*
0.17
-0.01
-0.27*
202.93
2.64
192.44
0.0661
(-1.86)
(0.82)
(1.37)
(5.26)
(1.74)
(4.27)
(1.14)
(4.24)
(-0.47)
(1.73)
(1.01)
(-1.55)
(-1.68)
(0.59)
(0.09)
(0.71)
4-1
05.76
0.39
3.07
0.04***
0.24
0.03***
0.19
0.04***
-0.04
0.01
0.12
-0.01*
-0.27*
-25.4
45.56
35.61
0.0659
(-1.25)
(1)
(0.93)
(5.26)
(1.59)
(4.44)
(1.1)
(5.24)
(-0.24)
(0.93)
(0.71)
(-1.84)
(-1.68)
(-0.08)
(1.55)
(0.14)
5146.99*
0.49
0.02
0.04***
0.25
0.03***
0.2
0.04***
-0.05
0.02**
0.2
-0.01
-0.3*
103.65
38.85
50.8
0.0648
(1.74)
(1.27)
(0)
(5.24)
(1.64)
(4.52)
(1.22)
(5.49)
(-0.31)
(2.15)
(1.17)
(-1.62)
(-1.88)
(0.31)
(1.33)
(0.18)
tstatisticsin
pare
nth
ese
s*
p<
0.10,**
p<
0.05,***
p<
0.01
19
-
Tab
le7:
Res
ult
sof
fixed
effec
tpan
elre
gres
sion
model
tote
stvo
lati
lity
info
rmat
ion
trad
ing
by
algo
rith
mic
and
non
-alg
orit
hm
ictr
ader
sin
the
NSE
opti
ons
mar
ket
contr
olling
for
unsc
hed
ule
dco
rpor
ate
annou
nce
men
ts.
Mea
sure
ofvo
lati
lity
(RV
):A
liza
deh
(200
2),
esti
mat
eof
real
ized
vola
tility
com
pute
dth
rough
diff
eren
ceb
etw
een
the
stock
’sin
trad
ayhig
han
dlo
wpri
cediv
ided
by
the
clos
ing
stock
pri
ce.
jConst.
DTG
σO
neD
ayRV
UA
D
abs(D
TG
∆)
ModelR
2
(t-j)
(t-j)
(t-1
)(t-1
)(t-2
)(t-2
)(t-3
)(t-3
)(t-4
)(t-4
)(t-5
)(t-5
)(t-j)
(t-j)
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
Trader
Group:
Alg
orithm
icTrader
1-3
40.48***
-0.91***
-47.85***
0.02***
-2.39***
0.01
10.85***
0.04***
2.13***
0.01
7.34***
-0.01
-4.05***
5261.84***
40.22*
1863.72***
0.4947
(-5.4)
(-3.17)
(-25.26)
(4.2)
(-11)
(1.6)
(63.03)
(6.38)
(8.67)
(0.9)
(36.84)
(-0.97)
(-16.66)
(11.67)
(1.88)
(7.83)
2-3
.48
-0.44
-19.63***
0.04***
-1.91***
0.01*
10.88***
0.04***
1.47***
0.01
7.67***
0-3
.25***
1484.45***
-26.55
-4047.59***
0.4609
(-0.05)
(-1.49)
(-8.42)
(6.46)
(-8.39)
(1.95)
(58.76)
(6.07)
(5.73)
(1.05)
(37.46)
(-0.78)
(-12.4)
(3.16)
(-1.2)
(-13.51)
3-1
.2-0
.04
-8.84***
0.04***
-1.76***
0.01**
10.91***
0.03***
1.71***
07.21***
0-3
.74***
-1910.04***
3.5
-1240.98***
0.4561
(-0.02)
(-0.12)
(-3.06)
(6.45)
(-7.82)
(2.29)
(61.34)
(5.56)
(6.65)
(0.85)
(33.6)
(-0.81)
(-13.84)
(-4.74)
(0.16)
(-5.28)
457.75
-0.36
0.43
0.04***
-2.4***
0.01**
10.77***
0.04***
2.08***
07.94***
-0.01
-3.29***
4471.13***
61.11***
-3604.67***
0.4639
(0.9)
(-1.23)
(0.13)
(6.5)
(-10.79)
(2.39)
(61.15)
(6.15)
(8.19)
(0.44)
(37.64)
(-0.88)
(-12.77)
(9.56)
(2.76)
(-10.17)
5130.15**
-0.22
-8.75***
0.04***
-2.94***
0.01**
11.36***
0.04***
2.03***
0.01
7.93***
-0.01
-2.97***
7348.68***
37.56*
685.45***
0.4617
(2.03)
(-0.76)
(-2.68)
(6.46)
(-12.92)
(2.43)
(63.85)
(6.24)
(7.97)
(1.06)
(38.69)
(-0.89)
(-11.08)
(15.5)
(1.7)
(2.94)
Trader
Group:
Non-A
lgorithm
icTrader
1-3
40.48***
0.91***
47.85***
0.02***
-2.39***
0.01
10.85***
0.04***
2.13***
0.01
7.34***
-0.01
-4.05***
5261.84***
40.22*
1863.72***
0.4947
(-5.4)
(3.17)
(25.26)
(4.2)
(-11)
(1.6)
(63.03)
(6.38)
(8.67)
(0.9)
(36.84)
(-0.97)
(-16.66)
(11.67)
(1.88)
(7.83)
2-3
.48
0.44
19.63***
0.04***
-1.91***
0.01*
10.88***
0.04***
1.47***
0.01
7.67***
0-3
.25***
1484.45***
-26.55
-4047.59***
0.4609
(-0.05)
(1.49)
(8.42)
(6.46)
(-8.39)
(1.95)
(58.76)
(6.07)
(5.73)
(1.05)
(37.46)
(-0.78)
(-12.4)
(3.16)
(-1.2)
(-13.51)
3-1
.20.04
8.84***
0.04***
-1.76***
0.01**
10.91***
0.03***
1.71***
07.21***
0-3
.74***
-1910.04***
3.5
-1240.98***
0.4561
(-0.02)
(0.12)
(3.06)
(6.45)
(-7.82)
(2.29)
(61.34)
(5.56)
(6.65)
(0.85)
(33.6)
(-0.81)
(-13.84)
(-4.74)
(0.16)
(-5.28)
457.75
0.36
-0.43
0.04***
-2.4***
0.01**
10.77***
0.04***
2.08***
07.94***
-0.01
-3.29***
4471.13***
61.11***
-3604.67***
0.4639
(0.9)
(1.23)
(-0.13)
(6.5)
(-10.79)
(2.39)
(61.15)
(6.15)
(8.19)
(0.44)
(37.64)
(-0.88)
(-12.77)
(9.56)
(2.76)
(-10.17)
5130.15**
0.22
8.75***
0.04***
-2.94***
0.01**
11.36***
0.04***
2.03***
0.01
7.93***
-0.01
-2.97***
7348.68***
37.56*
685.45***
0.4617
(2.03)
(0.76)
(2.68)
(6.46)
(-12.92)
(2.43)
(63.85)
(6.24)
(7.97)
(1.06)
(38.69)
(-0.89)
(-11.08)
(15.5)
(1.7)
(2.94)
tstatisticsin
pare
nth
ese
s*
p<
0.10,**
p<
0.05,***
p<
0.01
20
-
future realized volatility, indicating non-algorithmic traders
are informed regarding future
realized-volatility whereas algorithmic traders are not. The
results are consistent across
all definitions of realized volatility and for both type of
announcements- pre-scheduled
earnings announcements (Table 2, 4 & 6) and unscheduled
corporate announcements
(Table 3, 5 & 7). The fact that the coefficients have
similar sign and significance level for
the two type of announcements, seem to support our second
hypothesis that volatility
information based trading have similar implications with regard
to both scheduled and
unscheduled announcements. Similar to Ni et al. (2008), we find
that the interaction
term for the volatility demand and the EAD dummy is positive
(for non-algo traders)
suggesting options trading volume prior to earnings announcement
date has additional
information regarding future realized volatility. But unlike Ni
et al. (2008), we find that
predictive ability of options trading volume does not extend
till five trading days, rather
it is hardly significant beyond two trading days. We do observe
a change in the level of
significance for the lagged variables based on the definition of
one day realized volatility.
The exchange (NSE) reported realized volatility measures are
based on a EWMA type
modeling, where one-day lagged volatility has significant
weight. While using this mea-
sure as our dependent variable, we find that only the first lag
of the realized volatility
term remains significant, while higher order lags become
insignificant in most cases. Prior
to announcement dates, however, lagged terms do provide
additional information 7.
For our next set of models we split the algorithmic trader group
into proprietary
algorithmic traders and agency algorithmic traders as these two
groups differ fundamen-
tally in the way they employ algorithms. Proprietary algorithmic
traders are primarily
high-frequency traders who use their advantage of speed to
execute a large number of
relatively small-sized trades in very small time. Agency
algorithmic traders provide trade
execution services for other investors. Results indicate that
co-efficients corresponding to
volatility demand for both these trader groups are negative,
indicating none of them have
7A possible argument can be made that it is the surprise
component of the earnings announcementthat drives the volatility
spikes, where surprise is defined as the difference in earnings
levels from thelevels foretasted by analysts. We also run
robustness tests by sub-sampling the dataset for high and
lowearnings surprise (results not reported). However the results
remain consistent in both cases.
21
-
prior information regarding future volatility. Similar to our
first set of models, we use
all three definitions of volatility for both scheduled (Table 8,
10 & 12) and unscheduled
announcements (Table 9, 11 & 13). Institutional investors
are usually known to trade on
information. However, when institutional investors use
algorithms to execute trades on
their behalf, the agency algorithms split the orders to ensure
minimal price impact. As
such trades executed by agency algorithmic traders on behalf of
informed investors do
not convey information.
6 Conclusion
The exponential growth of algorithmic traders in the financial
markets demands a better
understanding of the role played by these machine traders. A lot
of recent literature
has been devoted towards their role in the spot market,
especially in issues related to
the provisioning of liquidity. However, the extent of literature
devoted to the role of
algorithmic traders in the derivative markets is considerably
lesser. Existing literature
seems to suggest that algorithmic traders react much faster to
public information. We do
not find any literature exploring whether algorithmic traders
have information regarding
future volatility. The benefit of leverage suggests that
informed investors are better off
using that information in the derivatives market compared to the
spot market. The non-
linear payoff structure suggests that options are ideal
securities for utilizing any volatility
related information. Using the framework provided by Ni et al.
(2008) we inspect if
algorithmic traders have information regarding future realized
volatility.
We use a large dataset obtained from the National Stock Exchange
of India which
provides identifiers for trades executed by algorithmic traders.
We use six months of
intraday data (Jan-Jun 2015) for both stock and options market 8
for 159 stocks to create
daily demand for volatility for various trader groups and relate
that to future realized
volatility in the spot market. We find that non-algorithmic
traders are informed about
8Number of trades executed in the NSE stock options market
during this period is more than 37 mn
22
-
Tab
le8:
Res
ult
sof
fixed
effec
tpan
elre
gres
sion
model
tote
stvo
lati
lity
info
rmat
ion
trad
ing
by
pro
pri
etar
yan
dag
ency
algo
rith
mic
trad
ers
inth
eN
SE
opti
ons
mar
ket
contr
olling
for
sched
ule
dea
rnin
gsan
nou
nce
men
ts.
Mea
sure
ofvo
lati
lity
(RV
):Sto
ckvo
lati
lity
rep
orte
dby
NSE
[Sqrt
(0.9
4∗PrevDayVolatility
2+
0.06
∗SameDayReturn
2)
orE
WM
Am
odel
]
jConst.
DTG
σO
neD
ayRV
EA
D
abs(D
TG
∆)
ModelR
2
(t-j)
(t-j)
(t-1
)(t-1
)(t-2
)(t-2
)(t-3
)(t-3
)(t-4
)(t-4
)(t-5
)(t-5
)(t-j)
(t-j)
*EA
D*EA
D*EA
D*EA
D*EA
D*EA
D*EA
D
Trader
Group:
Prop
Alg
orithm
icTrader
1-8
.5*
-0.05**
-0.82***
0.96***
0.2**
-0.02
0.05
-0.01
0.14
0.02*
-0.28**
-0.03***
-0.15*
-14.1
-3.27*
-8.26
0.9576
(-1.93)
(-2.06)
(-4.95)
(110.72)
(2.3)
(-1.42)
(0.4)
(-0.7)
(0.99)
(1.78)
(-2.14)
(-3.62)
(-1.83)
(-0.9)
(-1.82)
(-1.1)
28.02*
-0.05*
-0.75***
0.96***
0.15*
-0.01
0.1
-0.01
0.24*
0.02*
-0.37***
-0.03***
-0.16*
16.88
-4.07**
-30.34
0.9572
(1.82)
(-1.9)
(-2.92)
(123.97)
(1.75)
(-1.07)
(0.78)
(-1.38)
(1.74)
(1.78)
(-2.83)
(-3.76)
(-1.93)
(1.05)
(-2.3)
(-1.26)
34.95
-0.02
-0.27
0.96***
0.14
-0.02*
0.09
-0.01
0.2
0.02
-0.34***
-0.03***
-0.13
18.73
-2.99*
-2.91
0.9572
(1.13)
(-0.86)
(-1.18)
(124.34)
(1.64)
(-1.73)
(0.71)
(-0.78)
(1.41)
(1.52)
(-2.58)
(-3.58)
(-1.64)
(1.1)
(-1.68)
(-0.2)
49.08**
-0.04
-0.39*
0.97***
0.18**
-0.02*
0.08
-0.01
0.15
0.01
-0.33**
-0.02***
-0.13
-6.11
2.14
-23.16
0.9572
(2.1)
(-1.48)
(-1.67)
(124.28)
(1.98)
(-1.75)
(0.62)
(-0.99)
(1.08)
(1.23)
(-2.44)
(-2.85)
(-1.6)
(-0.37)
(1.2)
(-1.39)
513.88***
-0.04
-0.32
0.96***
0.17*
-0.02*
0.11
-0.01
0.2
0.02*
-0.39***
-0.03***
-0.13
8.54
1.42
-44.42**
0.9572
(3.31)
(-1.41)
(-1.27)
(124.24)
(1.95)
(-1.69)
(0.85)
(-1.01)
(1.4)
(1.84)
(-2.96)
(-3.66)
(-1.61)
(0.51)
(0.83)
(-2.48)
Trader
Group:
Agency
Alg
orithm
icTrader
1-7
.65*
-0.11***
-0.74***
0.95***
0.15*
-0.01
0.08
-0.01
0.15
0.02*
-0.29**
-0.03***
-0.13*
4.48
-1.08
50.87
0.9576
(-1.73)
(-3.15)
(-4.68)
(114.06)
(1.77)
(-1.26)
(0.64)
(-0.63)
(1.06)
(1.8)
(-2.24)
(-3.61)
(-1.65)
(0.27)
(-0.26)
(1.5)
28.42*
-0.03
-0.07
0.96***
0.15*
-0.02
0.08
-0.01
0.26*
0.02*
-0.4***
-0.03***
-0.13
21.7
-5.12
20.66
0.9572
(1.91)
(-0.99)
(-0.22)
(123.86)
(1.77)
(-1.44)
(0.6)
(-1.17)
(1.83)
(1.84)
(-3.05)
(-3.71)
(-1.64)
(1.32)
(-1.22)
(0.43)
37.23*
-0.02
-0.57
0.96***
0.13
-0.02*
0.09
-0.01
0.2
0.02**
-0.35***
-0.03***
-0.12
29.64*
8.38**
95.36*
0.9573
(1.65)
(-0.46)
(-1.44)
(124.42)
(1.47)
(-1.75)
(0.68)
(-1.31)
(1.44)
(1.99)
(-2.63)
(-3.51)
(-1.52)
(1.75)
(1.99)
(1.75)
410.82**
-0.04
-0.35
0.96***
0.15*
-0.02*
0.1
-0.01
0.16
0.01
-0.35***
-0.02***
-0.11
1.11
15.93***
2.6
0.9572
(2.5)
(-1.22)
(-1.32)
(124.21)
(1.75)
(-1.68)
(0.76)
(-1.02)
(1.14)
(1.24)
(-2.63)
(-2.93)
(-1.39)
(0.07)
(3.75)
(0.07)
514.4***
0.01
-0.48**
0.96***
0.16*
-0.02*
0.11
-0.01
0.19
0.02*
-0.38***
-0.03***
-0.13
26.56*
8.23*
8.91
0.9572
(3.45)
(0.39)
(-2.03)
(124.14)
(1.87)
(-1.69)
(0.88)
(-0.97)
(1.33)
(1.85)
(-2.9)
(-3.71)
(-1.62)
(1.65)
(1.95)
(0.23)
tstatisticsin
pare
nth
ese
s*
p<
0.10,**
p<
0.05,***
p<
0.01
23
-
Tab
le9:
Res
ult
sof
fixed
effec
tpan
elre
gres
sion
model
tote
stvo
lati
lity
info
rmat
ion
trad
ing
by
pro
pri
etar
yan
dag
ency
algo
rith
mic
trad
ers
inth
eN
SE
opti
ons
mar
ket
contr
olling
for
unsc
hed
ule
dco
rpor
ate
annou
nce
men
ts.
Mea
sure
ofvo
lati
lity
(RV
):Sto
ckvo
lati
lity
rep
orte
dby
NSE
[Sqrt
(0.9
4∗PrevDayVolatility
2+
0.06
∗SameDayReturn
2)
orE
WM
Am
odel
]
jConst.
DTG
σO
neD
ayRV
UA
D
abs(D
TG
∆)
ModelR
2
(t-j)
(t-j)
(t-1
)(t-1
)(t-2
)(t-2
)(t-3
)(t-3
)(t-4
)(t-4
)(t-5
)(t-5
)(t-j)
(t-j)
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
Trader
Group:
Prop
Alg
orithm
icTrader
1-7
.98*
-0.09***
-0.27
0.95***
1.03***
-0.01
-1.12***
-0.01
0.46*
0.02**
-0.3
-0.03***
-0.02
29.31
-3.06*
-11.81
0.9575
(-1.81)
(-3.24)
(-1.41)
(110.17)
(8.35)
(-0.9)
(-4.4)
(-0.89)
(1.75)
(2.04)
(-1.6)
(-3.93)
(-0.17)
(0.98)
(-1.74)
(-0.59)
29.06**
-0.06**
-0.11
0.96***
1.05***
0-1
.41***
-0.02
0.76***
0.02*
-0.31
-0.03***
-0.03
4.14
-4.36**
-4.12
0.9572
(2.06)
(-2.16)
(-0.64)
(123.3)
(8.62)
(-0.44)
(-5.45)
(-1.62)
(2.75)
(1.96)
(-1.63)
(-4.06)
(-0.25)
(0.14)
(-2.46)
(-0.17)
35.68
-0.02
-0.5*
0.96***
1.01***
-0.01
-1.36***
-0.01
0.82***
0.02*
-0.42**
-0.03***
05.25
-2.94*
-24.9
0.9572
(1.3)
(-0.94)
(-1.88)
(123.7)
(8.25)
(-1.22)
(-5.66)
(-0.84)
(3.15)
(1.72)
(-2.11)
(-3.92)
(0.02)
(0.2)
(-1.66)
(-1.28)
49.8**
-0.05*
0.05
0.96***
1.03***
-0.01
-1.31***
-0.01
0.7***
0.02
-0.33*
-0.03***
-0.03
30.41
2.05
11.78
0.9572
(2.27)
(-1.82)
(0.14)
(123.62)
(8.33)
(-1.25)
(-5.21)
(-1.16)
(2.65)
(1.6)
(-1.73)
(-3.24)
(-0.22)
(0.99)
(1.15)
(0.45)
513.55***
-0.04
0.19
0.96***
1.03***
-0.01
-1.22***
-0.01
0.61**
0.02**
-0.32*
-0.03***
-0.03
50.3
1.33
17.68
0.9572
(3.23)
(-1.41)
(0.64)
(123.58)
(8.28)
(-1.19)
(-5.03)
(-1.17)
(2.45)
(2.08)
(-1.7)
(-3.95)
(-0.22)
(1.63)
(0.78)
(0.96)
Trader
Group:
Agency
Alg
orithm
icTrader
1-7
.05
-0.14***
-1.04***
0.95***
0.94***
-0.01
-1***
-0.01
0.45*
0.02**
-0.25
-0.03***
-0.07
21.58
-0.67
-33.49
0.9575
(-1.59)
(-4.14)
(-2.88)
(113.52)
(7.53)
(-0.81)
(-3.96)
(-0.82)
(1.75)
(2.06)
(-1.34)
(-3.93)
(-0.59)
(0.7)
(-0.16)
(-1.09)
29.49**
-0.02
-0.75*
0.96***
1.05***
-0.01
-1.28***
-0.01
0.59**
0.02**
-0.31*
-0.03***
0-2
.17
-5.6
-71.28
0.9572
(2.15)
(-0.72)
(-1.86)
(123.28)
(8.39)
(-0.82)
(-5.25)
(-1.4)
(2.33)
(2.02)
(-1.66)
(-4.02)
(0.01)
(-0.07)
(-1.34)
(-1.24)
37.93*
-0.02
-0.28
0.96***
1.05***
-0.01
-1.35***
-0.02
0.7***
0.02**
-0.32*
-0.03***
-0.03
15.83
8.87**
6.24
0.9572
(1.81)
(-0.68)
(-0.53)
(123.75)
(8.6)
(-1.24)
(-5.5)
(-1.37)
(2.62)
(2.19)
(-1.65)
(-3.84)
(-0.22)
(0.62)
(2.1)
(0.15)
411.89***
-0.05
0.29
0.96***
1***
-0.01
-1.22***
-0.01
0.69***
0.02
-0.33*
-0.03***
-0.08
14.16
17.16***
-88.59
0.9572
(2.74)
(-1.48)
(0.81)
(123.57)
(7.97)
(-1.18)
(-5.04)
(-1.19)
(2.75)
(1.6)
(-1.69)
(-3.32)
(-0.61)
(0.5)
(4.05)
(-1.49)
514.28***
00.52
0.96***
1.03***
-0.01
-1.24***
-0.01
0.63**
0.02**
-0.34*
-0.03***
-0.01
45.13
8.34**
20.09
0.9572
(3.42)
(0.12)
(0.9)
(123.5)
(8.3)
(-1.19)
(-5.16)
(-1.13)
(2.52)
(2.08)
(-1.78)
(-4)
(-0.12)
(1.54)
(1.98)
(0.36)
tstatisticsin
pare
nth
ese
s*
p<
0.10,**
p<
0.05,***
p<
0.01
24
-
Tab
le10
:R
esult
sof
fixed
effec
tpan
elre
gres
sion
model
tote
stvo
lati
lity
info
rmat
ion
trad
ing
by
pro
pri
etar
yan
dag
ency
algo
rith
mic
trad
ers
inth
eN
SE
opti
ons
mar
ket
contr
olling
for
sched
ule
dea
rnin
gsan
nou
nce
men
ts.
Mea
sure
ofvo
lati
lity
(RV
):A
nder
son
(200
1),
esti
mat
eof
real
ized
vola
tility
usi
ng
intr
a-day
five
-min
ute
retu
rnof
the
secu
rity
.
jConst.
DTG
σO
neD
ayRV
EA
D
abs(D
TG
∆)
ModelR
2
(t-j)
(t-j)
(t-1
)(t-1
)(t-2
)(t-2
)(t-3
)(t-3
)(t-4
)(t-4
)(t-5
)(t-5
)(t-j)
(t-j)
*EA
D*EA
D*EA
D*EA
D*EA
D*EA
D*EA
D
Trader
Group:
Prop
Alg
orithm
icTrader
1-8
9.33***
-0.29**
-2.46***
0.19***
0.05
0.06***
0.27***
0.07***
0.02
0.04***
0.31***
0.01
-0.5***
166.19**
30.34***
46.18
0.360
(-4.55)
(-2.49)
(-3.36)
(22.7)
(0.62)
(7.72)
(2.67)
(9.7)
(0.24)
(5.62)
(3.07)
(1.58)
(-5.74)
(2.38)
(3.96)
(1.43)
233.5*
-0.23**
-0.22
0.22***
0.03
0.06***
0.31***
0.07***
0.03
0.04***
0.32***
0.01
-0.53***
230.45***
-11.16
75.77
0.355
(1.7)
(-1.97)
(-0.2)
(28.91)
(0.33)
(7)
(3.07)
(9.42)
(0.31)
(5.54)
(3.11)
(1.55)
(-6.02)
(3.19)
(-1.48)
(0.72)
331.22
-0.15
-1.16
0.22***
0.04
0.06***
0.23**
0.07***
0.04
0.04***
0.31***
0.01
-0.52***
150.95**
-23.4***
90.35
0.3551
(1.59)
(-1.29)
(-1.13)
(28.99)
(0.47)
(8.17)
(2.33)
(8.44)
(0.43)
(5.41)
(3.03)
(1.43)
(-5.88)
(1.96)
(-3.1)
(1.43)
426.72
-0.04
-10.22***
0.04
0.06***
0.25**
0.07***
0.02
0.03***
0.29***
0.01
-0.52***
138.19*
9.77
147.14**
0.3551
(1.36)
(-0.34)
(-0.97)
(29.05)
(0.5)
(8.11)
(2.47)
(8.98)
(0.27)
(4.01)
(2.82)
(1.46)
(-5.85)
(1.87)
(1.29)
(2.11)
589.12***
-0.07
0.57
0.22***
0.03
0.06***
0.28***
0.07***
-0.02
0.04***
0.32***
0.02*
-0.55***
89.66
2.4
75.05
0.3542
(4.57)
(-0.65)
(0.51)
(28.91)
(0.3)
(8.12)
(2.83)
(9.21)
(-0.21)
(5.72)
(3.13)
(1.96)
(-6.04)
(1.19)
(0.32)
(0.95)
Trader
Group:
Agency
Alg
orithm
icTrader
1-9
8.14***
-0.54***
-1.85***
0.19***
0.02
0.06***
0.23**
0.07***
0.04
0.04***
0.32***
0.01
-0.51***
123.47*
39.12**
-61.52
0.3596
(-5.02)
(-3.56)
(-2.68)
(22.91)
(0.21)
(7.55)
(2.28)
(9.58)
(0.43)
(5.55)
(3.09)
(1.48)
(-5.82)
(1.71)
(2.13)
(-0.42)
237.77*
-0.27*
3.47**
0.22***
00.06***
0.32***
0.07***
0.01
0.04***
0.31***
0.01
-0.51***
265.24***
-3.83
407.59*
0.3552
(1.93)
(-1.83)
(2.4)
(28.86)
(0.03)
(6.98)
(3.21)
(9.35)
(0.11)
(5.55)
(3)
(1.59)
(-5.79)
(3.62)
(-0.21)
(1.92)
340.15**
0.16
-1.38
0.22***
0.04
0.06***
0.25**
0.07***
-0.04
0.04***
0.34***
0.01
-0.51***
185.49**
-8.85
741.82***
0.355
(2.05)
(1.1)
(-0.79)
(29.05)
(0.41)
(8.13)
(2.49)
(8.32)
(-0.43)
(5.44)
(3.31)
(1.48)
(-5.85)
(2.45)
(-0.48)
(3.03)
430.59
-0.03
-5.03***
0.22***
0.03
0.06***
0.18*
0.07***
00.03***
0.35***
0.01
-0.52***
135.14*
50.82***
116.59
0.3559
(1.57)
(-0.2)
(-4.25)
(29.06)
(0.39)
(8.11)
(1.85)
(9)
(0.03)
(3.97)
(3.29)
(1.47)
(-5.87)
(1.79)
(2.73)
(0.71)
590.54***
0.02
-0.96
0.22***
0.02
0.06***
0.27***
0.07***
-0.02
0.04***
0.33***
0.02*
-0.56***
48.67
16.91
-80.55
0.3543
(4.66)
(0.13)
(-0.92)
(28.89)
(0.24)
(8.13)
(2.77)
(9.22)
(-0.19)
(5.74)
(3.2)
(1.95)
(-6.14)
(0.67)
(0.91)
(-0.47)
tstatisticsin
pare
nth
ese
s*
p<
0.10,**
p<
0.05,***
p<
0.01
25
-
Tab
le11
:R
esult
sof
fixed
effec
tpan
elre
gres
sion
model
tote
stvo
lati
lity
info
rmat
ion
trad
ing
by
pro
pri
etar
yan
dag
ency
algo
rith
mic
trad
ers
inth
eN
SE
opti
ons
mar
ket
contr
olling
for
unsc
hed
ule
dco
rpor
ate
annou
nce
men
ts.
Mea
sure
ofvo
lati
lity
(RV
):A
nder
son
(200
1),
esti
mat
eof
real
ized
vola
tility
usi
ng
intr
a-day
five
-min
ute
retu
rnof
the
secu
rity
.
jConst.
DTG
σO
neD
ayRV
UA
D
abs(D
TG
∆)
ModelR
2
(t-j)
(t-j)
(t-1
)(t-1
)(t-2
)(t-2
)(t-3
)(t-3
)(t-4
)(t-4
)(t-5
)(t-5
)(t-j)
(t-j)
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
*U
AD
Trader
Group:
Prop
Alg
orithm
icTrader
1-7
0.43***
-0.45***
-2.87***
0.18***
0.45***
0.05***
1.71***
0.07***
1.49***
0.03***
3.53***
0.01**
-3.37***
556.18***
31.71***
492.87***
0.4265
(-3.79)
(-4.1)
(-3.87)
(22.76)
(3.07)
(6.5)
(11.38)
(9.24)
(8.82)
(4.47)
(20.72)
(2.01)
(-19.05)
(4.45)
(4.46)
(5.79)
259.18***
-0.21*
-3.52***
0.21***
0.46***
0.05***
1.97***
0.07***
1.27***
0.03***
3.48***
0.01**
-3.69***
-539.85***
-9.95
-707.83***
0.4174
(3.17)
(-1.83)
(-5.48)
(28.62)
(3.1)
(6.65)
(12.68)
(9.01)
(7.38)
(4.35)
(20.27)
(1.97)
(-20.58)
(-4.26)
(-1.39)
(-6.83)
349.27***
-0.11
-9.7***
0.21***
0.58***
0.05***
1.94***
0.07***
0.92***
0.03***
3.8***
0.01*
-3.64***
-706.88***
-20.93***
-747.86***
0.4212
(2.65)
(-1.02)
(-8.91)
(28.69)
(4.07)
(6.98)
(13.11)
(8.65)
(5.3)
(4.37)
(21.55)
(1.89)
(-20.04)
(-6.29)
(-2.93)
(-9.28)
448.32***
-0.09
-3.26**
0.21***
0.29**
0.05***
1.93***
0.06***
1.42***
0.03***
3.66***
0.01**
-3.63***
186.76
14.16**
-651.16***
0.4181
(2.6)
(-0.82)
(-2.43)
(28.63)
(2)
(6.75)
(13.04)
(8.53)
(8.39)
(3.56)
(21.17)
(2.03)
(-20.27)
(1.45)
(1.98)
(-6.02)
588.08***
0.01
-1.75
0.21***
0.43***
0.05***
1.89***
0.06***
1.
