Content uploaded by Marten Risius
Author content
All content in this area was uploaded by Marten Risius on Apr 27, 2020
Content may be subject to copyright.
Research Article
Effects of transparency: analyzing social
biases on trader performance in social
trading
Florian Glaser
1
, Marten Risius
2
1
Goethe University Frankfurt, Theodor-W.-Adorno-Platz 4, 60323 Frankfurt am Main, Germany;
2
University of Mannheim, Mannheim, Germany
Correspondence:
F Glaser, Goethe University Frankfurt, Theodor-W.-Adorno-Platz 4, 60323 Frankfurt am Main, Germany.
Tel: +49 (0)69 798 34679;
E-mail: fglaser@wiwi.uni-frankfurt.de
Abstract
Social Trading platforms combine the trading functionalities of classical online broker
services with the communication and interaction features of social networks. Next to
following other users’ profiles, a main characteristic of social trading platforms is the
possibility to follow other users by automatically copying their trades. By being a
technologically based financial intermediary that enables individual profit maximization,
social trading platforms constitute a contemporary example of financialization. Our
empirical analysis of the behavior of traders on a social trading platform provides new
insights on financialization related questions regarding the influence of transparency and
interaction in delegated investment environments. The disposition effect is a well-studied
behavioral bias of investors and traders. Human investors tend to realize returns of their
winning positions too early and let unfavorable positions accumulate losses for too long.
We find that on social trading platforms the traders’ sensitivity to the disposition effect is
influenced by the amount of attention they receive from their followers who invested capital
into the traders’ strategy. These novel insights propose a link between principal-agent
theory and the disposition effect induced by transparency mechanisms. We extend the
literature on trader-investor interaction channels in social trading networks. The results
obtained in a social network environment are of high relevance for regulators who have a
strong focus on customer protection and financial services regulation. They also provide
guidelines for platform designers, traders, investors and social trading platform operators.
Journal of Information Technology (2016). doi:10.1057/s41265-016-0028-0
Keywords: social trading; digitalization of financial services; financialization; disposition
effect; agency theory; individual trading performance
Introduction
Financialization describes ‘‘the increasing role of financial
motives, financial markets, financial actors and financial
institutions in the operation of the domestic and
international economies’’ (Epstein, 2005: 3), which results
‘‘in a structural transformation of economies, firms (includ-
ing financial institutions), states and households’’ (Aalbers,
2015: 3). It is considered to be the politically inadvertent
solution to current economic problems (Buchanan, 2016)
such as the diminished trust of private persons into
institutional investors (Lagoarde-Segot, 2016). Thereby,
financialization focuses on the benefit of actors within
financial markets (Aalbers, 2008) where profit is generated
through financial channels rather than productive means
(Buchanan, 2016). Symptomatically, personal speculative
trading in form of day trading or foreign exchange transac-
tions has seized unprecedented market shares over the past
years (Lagoarde-Segot, 2016). These developments are con-
sidered to be mainly driven by information technology,
which decreases the cost of traded assets for households
(Philippon, 2015). Therefore, in conjunction with the
accelerated digitization of the financial industry, financial-
ization has provided leeway for the development of market
Journal of Information Technology (2016)
ª2016 Association for Information Technology Trust All rights reserved 0268-3962/16
www.palgrave.com/journals
innovations such as technologically supported (dis-)inter-
mediation (Currie and Lagoarde-Segot, 2016).
In this regard, particularly social trading networks are
becoming increasingly relevant as disintermediating platforms
that provide a high degree of transparency for investors. Social
trading platforms, on the one hand, embody financialization
by providing individuals with further means of generating
profits through purely financial channels like speculative
trading (Doering et al.,2015). Accordingly, the platform www.
zulutrade.com – henceforth Zulutrade, which is subject to our
analysis – reports a total trading volume of more than $8.6
billion since its inception in 2008 (www.zulutrade.com). On
the other hand, social trading also helps to address problems
associated with financialization. Specifically, they satisfy the
increasing need for transparency and trust, either due to reg-
ulatory requirements (e.g., the Markets in Financial Instru-
ments Directive II – MiFID2) or due to the diminishing trust in
incumbent financial service providers (Sapienza and Zingales,
2012; Alt and Puschmann, 2012; Ba and Pavlou, 2002) and
online services in general (Jøsang et al.,2007).
Given the currently strong trend towards disintermediation
of financial services (e.g., Blockchain technology, peer-to-peer
lending), social trading is a candidate for disrupting incum-
bent asset management services. However, social trading is a
novel service for online trading and investing, which has barely
been researched so far (Neumann et al.,2015;Panet al.,2012)
and, hence, is not as well understood as comparable incum-
bent investment delegation services like fund management or
asset management. Research on social trading platforms has
been limited to the analysis of whether traders can gain profits
from social trading platforms at all (Pan et al.,2012;Abbey
and Doukas, 2015;Neumannet al.,2015). We consider social
trading to be relevant for information systems researchers,
because this type of service platform has the potential to
simultaneously address two core aspects of financialization. Its
features enable full transparency for investors and near-real
time resolution control over the investor’s invested capital,
which enables individual investors to maximize their profit.
These features imply both great possibilities (e.g., allocating
capital to skilled and more experiences investors while
bypassing unnecessary intermediaries) as well as risks (e.g.,
loss of invested money due to overly risky investment
strategies). The ability to leverage the full potential of these
electronic markets and financial services, however, depends on
the design of their functionalities and integrated services for
investors to enable trust and alleviate interaction and trans-
action (Ba and Pavlou, 2002). To the best of our knowledge,
we are the first to investigate the influence of the design-
inherent transparency on trader behavior.
Against the backdrop of these developments, we focus our
research on cooperation and influencing channels of inter-
action in online trading environments, specifically in the
context of social trading. The goal of this work is to provide
answers to the underlying research question:
How is trader behavior influenced by real-time interaction
with the investors?
In search for answers, we analyze the behavior of traders on one
of the leading social trading platforms – www.zulutrade.com –
to obtain novel insights and extend the knowledge on behavior
in fully digitized, transparent financial trading environments.
The remainder of the paper is structured as follows. In
‘‘Social trading background’’ section, we start with a review of
social trading platforms in general. Afterwards, we provide an
overview of day trading and foreign exchange trading, which is
the predominant type of investment on social trading
platforms. In ‘‘Related literature and hypotheses development’’
section we review the related literature, derive theoretical
hypotheses and propose our research model which is empir-
ically validated in ‘‘Empirical analysis’’ section. ‘‘Discussion’’
section provides a critical reflection on our research approach.
We conclude with a general overview of the contributions and
limitations of this work in ‘‘Conclusion’’ section.
Social trading background
This section provides a short introduction to social trading to
highlight the features and functionalities that define the novel
online broker service in relation to other social media
platforms and traditional investor-broker relationships as well
as provide an overview over related social trading literature.
Functionalities of social trading platforms
Social trading platforms provide the infrastructure to display
profiles of users (in this context traders) to other users in a
fashion that is similar to social networks. The displayed profile
information contains performance and risk metrics of the
users’ trading activities. Trades of users are visible to every
other user on a per trade basis, i.e. every trading decision is
made public thereby introducing full transparency.
Users can copy other users’ trades with a single click.
Copying means that the same trade is executed in the copying
user’s online broker account. This distinguishing feature is
known as copy-trading. From a technological perspective,
this is achieved by linking accounts. Users’ platform trading
accounts are directly linked to their online broker accounts.
The operator is often an online brokerage company. How-
ever, the operator and the broker can be different institutions
as well. Figure 1provides a generalized visualization of the
technical design of a social trading platform.
The resulting dependency between traders and their
followers is comparable to that in traditional delegated
portfolio management. In the traditional portfolio manage-
ment setup investors delegate their capital to portfolio
managers in order to invest the capital on the investors’
behalf. Consequently, the relationship bears a similar moral
hazard as a result of the principal-agent connection between
trader and follower (Bhattacharya and Pfleiderer, 1985;
Stoughton, 1993). That is the portfolio manager (agent)
might have an incentive to maximize his own profit instead
of the investor’s profit. If his incentives are not tightly
linked to the performance of his investment decisions (e.g.,
he receives a yearly share of the total amount invested
regardless of the performance), the incentive to maximize
performance for the investors might not be strong enough.
This can be mitigated by performance bound incentives,
enabling monitoring of the agents behavior and increased
possibilities control (Farnsworth and Taylor, 2006). A
fundamental difference to the traditional principal-agent
relationship is the level of transparency regarding actions
and decisions made by agents (traders). While in the
traditional principal-agent situationaninvestoronlygets
periodic portfolio performance summaries, every single
Effects of transparency F Glaser and M Risius
opening of a position and every decision to close a position
can be monitored in real time by every follower (principal)
on a social trading platform. Due to the direct linking of
accounts, investors can immediately withdraw their capital
from a trader at almost no costs. In traditional asset
management, investors face far longer lead times for fund
withdrawal and comparably less transparency, especially
with respect to costs.
Most social trading platforms primarily focus on foreign
exchange (FX) trading and facilitate day trading by their
tool setup. Therefore, they concur with the general
financialization related increase in speculative trading,
which have resulted in a global daily FX transaction
turnover of $9 trillion (Lagoarde-Segot, 2016). This focus
has historical reasons as online trading communities based
on forums (Antweiler and Frank, 2004)havealong
tradition in FX trading. The digitization has brought these
communities to the next level of automated trade and
strategy copying in the form of social trading platforms.
Furthermore, trading of currencies is less regulated and
markets for major currencies are known to be the most
liquid markets. Thus, adverse price effects like market
impact costs of copied trades are very unlikely to
negatively influence followers’ performance (e.g., in an
illiquid market, the trade of a follower could be executed
at a worse price due to a lack of supply for the traded asset
and hence higher quoted prices).
Day trading and FX trading
Day trading and foreign exchange trading follow different
conventions and standards in comparison to stock or fund
trading. This section provides a general introduction to some
peculiarities in order to enhance the understanding for the
trading environment and the relevant variables.
A fundamental concept in FX trading is the notion of a
‘‘lot’’. A lot is the unit for the size of a trade and one lot
usually refers to a trade size of $100.000. The profit in FX
trading is often measured in pips where one pip represents
the smallest change of the price for a currency pair. Prices of
currency pairs are usually quoted with a precision of four
digits, so the smallest change in price is a change on the
fourth digit by one, i.e. one pip (Garner, 2012).
To increase the returns on investment given these small
changes in prices, investors can leverage their exposure, often
up to a leverage of 1:100 or even more. That means investors
only invest 1/100th of the actual traded amount but fully
participate on the return (positive or negative) on the actual
traded amount. The followers on the platform can decide
about the leverage that is applied when a trade is copied. This
decision can be set to a default value for every copied trade or
on a per trade basis. It follows that considering the level of
returns of a trade is not a valid indicator for the actual return
of a trader’s followers. We can only know with certainty if the
trade had a positive or negative return. The actual realized
return depends on each follower’s individual leverage deci-
sion for their copy of the trade (Garner, 2012).
FX traders are usually day traders, i.e. they are opening and
closing positions within a single day (Bjønnes et al.,2005;
Bjønnes and Rime, 2005; Breedon and Ranaldo, 2013).
Remuneration and incentive schemes
According to Neumann et al. (2015), there are three generic
remuneration schemes in social trading for brokers in general:
Follower-based, profit-based and volume-based. The Follower-
based model compensates traders based on the actual number
of followers. The more followers a trader has, the higher the
compensation, irrespective of the trader’s or followers’ perfor-
mances. Profit-based remuneration means that the traders
participate on the profits of the trades that are executed in the
accounts of their followers. In volume-based models, traders
share the profit from premiums imposed on execution prices
for follower trades with the platform operating broker. The
prices at which a follower trade is executed are slightlyhigher to
compensate the broker and the trader. It follows that the more
trades are executed in follower accounts, the higher the revenue
for a trader. Note that the trades need not be winning trades
(i.e., trades with positive return when they are closed) in order
to let the trader make revenue.
Social Trading Platform
Social Network
Forum
Chat
Messages
Comments
Follow
Trader Account
Performance/Risk Profile
Portfolio Trades
Follower
Account
Performance/Risk Profile
Portfolios Trades
copy
Broker
Financial Markets
Accounts
Operator
copy
Figure 1 Generic social trading platform design.
Effects of transparency F Glaser and M Risius
Incentives on Zulutrade follow a volume-based model. A
trader receives an absolute amount in pips for each lot that is
executed by their followers. Additionally, they earn the return
on their own trades given the case they execute trades with
their own capital.
It is crucial for our latter argumentation that on Zulutrade
traders do not necessarily need to trade with actual money.
They can simply provide paper trades (trades that are
executed without placing any money) in their accounts and
profit from followers that automatically execute these trades
with their capital. Whether traders simply conduct paper
trades or invest actual money naturally has strong implica-
tions for the moral hazard and agency problems hypotheses,
which will be discussed later in ‘‘Subjective biases on
individual trading performance’’ section.
Economic and theoretical aspects of the social trading industry
Social trading platforms developed following the general
dissemination of social media tools and, therefore, constitute
a comparatively young phenomenon. While a large number
of platform providers have arisen, the two most prominent
examples with the largest user numbers are eToro (4.5
million users) and Zulutrade (1 million users) (Robinson,
2014). The platforms differ regarding the traded instruments
(most commonly mentioned are Forex, Indices, Commodi-
ties, Shares, and ETFs). During the initial phase of dissem-
ination, these platforms annually enabled $40 million
revenue through FX trading accounting for 0.5% of the
global market (Falkner and Corsthwaite, 2013).While it is
generally difficult to obtain independently published figures,
Wikifolio reports an annual trading volume of $300 million
in 2014 (Oehler et al.,2016) and Zulutrade claims a
transaction volume of over $917 billion since its inception
in 2007 (Zulutrade, 2016).
Due to the growing adoption and substantial economic
prospects, social trading has already been subject to several
studies. First research addressed concerns regarding the
performance and reliability of social trading. Comparing the
returns of social trading certificates to respective benchmark
market indexes demonstrates a comparable performance of
the traditional and novel investment approaches with
particularly high short-term returns for the latter (Doering
et al.,2015; Oehler et al.,2016). Moreover, the analysis of
stock assessments posted by users on a social trading
platform have been found to provide significant explanatory
power for future stock returns (Chen et al.,2014; Pelster and
Breitmayer, 2016). Regarding the common trading behavior,
classical phenomena like ‘‘loss aversion’’ were successfully
replicated by large-scale analyses of the online financial
communities (Liu et al., 2014). These results document the
relevance and transferability of the findings on social trading
platforms to other trading environments. Due to the unique
opportunity of accessing relational information on social
trading sites, further research has investigated the influence
of social factors on the individual trading behavior. Pan et al.
(2012) found higher returns for trades copied from platform-
dependent expert traders compared to individually self-
initiated trades. However, they also found preliminary
evidence that the assessment of expertise does not only
depend on objective performance measures but also on social
cues (e.g., follower numbers). In a more targeted analysis,
Wohlgemuth et al. (2016) consequently found that expert
trustworthiness depends on objective performance measures
(i.e., profitable trades, returns, maximum losses, and risk
level) as well as on social cues (i.e., profile pictures, displaying
the real name, and interaction frequency). In order to
improve the selection of experts, researchers proposed an
algorithmic system that helps to identify and visualize expert
traders based on their performance, risk, and consistency
history (Lee and Ma, 2015). Furthermore, platform enabled
social cues on individual (i.e., detailed decision information
from peers) and network levels (i.e., aggregated decisions
from the crowd) have been found to improve the risk
management of otherwise risk averse investors (Zhao et al.,
2015) by reducing under-diversification (Baghestanian et al.,
2015). This effect, however, partially depends on the way in
which the related peer information is framed. Prominently
displaying the highest earning trader can actually increase the
risk affinity of the other users (Baghestanian et al.,2015).
This is particularly relevant, since especially traders with a
good past performance publicly present their strategies
(Amman and Schaub, 2016), which might move people to
underestimate risks. Interestingly, however, it is not the
information but rather the sentiment of the trader commu-
nication that drives the followers’ investment behavior
(Amman and Schaub, 2016).Thus, in general it can be seen
that social trading platforms offer a unique opportunity for
researchers to investigate the impact of social characteristics
on trading behavior, which we will elaborate in the subse-
quent section.
Related literature and hypotheses development
Measuring individual trading performance
The goal of this paper is to identify influences on the decision
making of traders in fully transparent trading environments.
We first consider how to measure trading performance and
how to identify known influences and agency-related biases
that affect traders in general and might have an impact on
traders on social trading platforms as well.
A common metric for the skill or success of any trader is
the ratio of trades with positive returns (henceforth winratio)
(Dewally et al.,2009). Considering that day trading and
especially FX trading allow high levels of leverage, followers
can yield a much higher return on a trade if they copy the
trade and apply a high individual leverage. Accordingly, in an
expert interview the customer support of Zulutrade men-
tioned winratio a key measure for customers to focus on
when assessing the quality of traders. Therefore, Zulutrade
recommends new platform users – irrespective of their
experience in financial markets – to select traders mainly by
the overall winratio, among other variables like age of the
account.
Moreover, the winratio depends solely on the decision of
the traders to close any of the open trades that provide a
positive return independent of the level of the return. Put
differently, the winratio performance measure reacts imme-
diately to any changes in the decision makers’ behavior but
still captures the quality of the trader decisions (number of
positive trades). Return levels do not capture this behavioral
decision dimension as they can hardly be influenced by
traders. This is essential in this case, as we investigate exactly
Effects of transparency F Glaser and M Risius
these immediate changes in the behavior and decision
making of traders. The individual trader’s performance and
decision making is not only based on skill and ability levels
but has been found to be affected through certain behavioral
and social characteristics, as we elaborate in the following.
Subjective biases on individual trading performance
It has long been proposed by the efficient market hypothesis
that stock prices objectively reflect all related information at
the given time (Fama, 1970,1998). New information is
instantly processed by the markets, which establishes a price
equilibrium and precludes misspecification of stocks (Lo,
2004). In this sense, prices are objectively justified and
returns are independent from subjective characteristics of the
trader. Criticism of this perspective has been fueled by events
like financial crises due to overrated market estimations
(Shiller, 2015) and empirical findings that show the non-
compliance of stock price movements with company funda-
mentals (Cornell, 2013). This has given rise to the assump-
tion of behavioral finance that irrational human behavior can
affect stock market reactions. As one of the most prominent,
(Keynes, 2013) argued that subjective impacts like sentiment
or conventional valuations affect human long-term expecta-
tions and investment decisions beyond rational fact-based
judgement. The establishment of financialization in academic
finance research corresponds with this development to a
certain degree. Academic finance relied on the ‘representative
idiom’ which assumes an objectively identifiable financial
reality – like the efficient market hypothesis – represented by
an empirical equivalent (Lagoarde-Segot, 2016). Behavioral
finance, however, considers an objectivist world view to be
somewhat outdated (Orlean, 2012). Correspondingly, finan-
cialization researchers follow the ‘performative idiom’, which
allows for external factors to influence multiple realities of
economic agents (Lagoarde-Segot, 2016). Consequently,
researchers call for multidisciplinary approaches to purpose-
fully address financialization and behavioral finance (Aalbers,
2015; Orlean, 2012). Therefore, we refer to findings from
behavioral finance on subjective effects to evaluate the
individual trading performance. Social trading platforms
have been found to be a particularly viable environment to
study subjective effects on trading behavior, because of their
key features regarding transparent social network structures
and immediate communicative exchanges (e.g., Chen et al.,
2014; Liu et al.,2014; Wohlgemuth et al.,2016). In line with
the aforementioned social trading research, we consider
errors of human judgement due to the attention from others
(i.e., behavioral and relational biases) to be particularly
relevant (Shiller, 2005).
Behavioral biases and patterns of day traders
A trader can close trades prematurely to guarantee capturing
a positive return. If an open position would yield a negative
return, traders can decide to wait if the position yields a
positive return later on. In behavioral finance, these two
behavioral patterns are known as the ‘‘disposition effect’’. I t
states that investors are likely to close a winning trade too
early and to avoid closing a trade at a loss for too long
(Shefrin and Statman, 1985). It is one of the most prominent
stylized facts in behavioral finance and empirically validated
over time, different asset classes, countries and known to
affect professional and novice investors alike (Weber and
Camerer, 1998). It is sometimes also referred to as ‘‘sign
preference effect’’ (Ben-David and Hirshleifer, 2012). Figure 2
depicts the different scenarios regarding the disposition effect
and the timing of the closing of a trade which makes the
difference (for a more detailed overview of the relevance and
difficulty of timing specifically for FX trading, we recom-
mend Rosov and Foster (2014)). In the context of social
trading, the disposition effect has been found to be reduced
through mirror trading of expert strategies (Liu et al.,2014).
Since, however, pure mirror trading approaches only account
for two percent of the trades initiated over the platforms (Pan
et al., 2012), the respectively mitigating effects can be
considered to be rather low.
Given the previous discussion regarding behavioral per-
formance drivers, for social trading platforms we hypothesize:
H1 The timing of trades significantly affects the trader
performance.
H2 Letting losing positions accumulate losses and pre-
maturely closing winning positions biases the trader
performance.
Apart from the behavioral drivers of trader performance,
the broker’s portfolio focus/diversity has been found to
influence the investment behavior. For example, Chen and
Lai (2015) compared concentrated investment styles (i.e.,
investors focusing on only a few different assets) to less
concentrated, i.e., overconfident ones (i.e. investors feeling
able to assess many different assets simultaneously and with
Trade duraon
Return
Return
Return
Opmal closing point Disposion
(Keep losses running)
Disposion
(Sign preference)
Equivalent contribuon to winrao (bothposive) but different realized return levels
Trade duraon Trade duraon
Realized return
level
Figure 2 Disposition effect related scenarios of timing the closing of a trade.
Effects of transparency F Glaser and M Risius
the same quality). They find that fund managers exhibit
higher performance if they concentrate on few stocks.
Furthermore, Hiraki et al. (2015) show that industry-
concentrated funds outperform more diversified peers.
Specifically for FX trading, Jacobs and Weber (2015) find
that distraction (induced by following different assets) of
traders results in a negative influence on their performance.
Hence, we hypothesize:
H3 Focusing on particular instruments improves trader
performance compared to diversified investment.
Lastly, behavioral finance research has investigated the role
of the level of trading activity on performance. The promi-
nent work of Barber and Odean (2002) shows that individual
online retail investors tend to perform worse when they
increase their trading frequency. However, Abbey and
Doukas (2015) find for individual currency traders that an
increased turnover is associated with higher performance.
This is in line with the findings of Dewally et al. (2009) for
non-Household traders. An extended review of profiles of
traders on Zulutrade revealed that traders are often experi-
enced FX traders. They either have a professional background
in finance and trading or document a strong dedication to
trading as part time job or additional source of personal
income. Hence, we assume that the sample comprises
primarily non-Household investors and hypothesize:
H4 An increased trading experience positively affects the
trader performance.
Moral hazard in delegated investment setups
In social trading environments, traders and investors are tied
together in a classical principal-agent setup (Holmstrom and
Milgrom, 1991). The setup is comparable to delegated
portfolio management (Bhattacharya and Pfleiderer, 1985;
Fernandes et al.,2010). In general, agency theory explains the
forces on relationships in transactional environments where
either party is following personal interests and uncertainty
prevails due to information asymmetry, i.e., imperfect
monitoring capabilities (Pavlou et al.,2007). The agent
performs actions on behalf of the principal. The principal
cannot perfectly monitor the agent’s behavior but has to rely
on the incentives and motives that are guiding the agent’s
decision on the investor’s behalf. For example, an investor
has invested a share of the personal capital into a fund of a
professional manager. The principal cannot fully monitor the
motives and incentives guiding the professional investor but
has to rely on the incentive of the fund manager who is
rewarded based on the performance. For portfolio manage-
ment these relationships are extensively studied in different
literature streams. For instance, Evans (2008) reports the
decrease of agency costs and decrease of turnover to be
associated with higher personal investments of fund man-
agers in their own funds. A general review of relationships
between compensation, incentives and risk based on an
extensive theoretical analysis is provided by Ross (2004). The
decisive factor for the quality of the principal-agent rela-
tionship is the moral hazard. The conceptual similarity of
asset management to social trading environments (followers
invest their capital in traders’ strategies) introduces a
principal-agent relationship similar to the classical asset
management. Hence, they face a comparable issue of moral
hazard due to the previously described incentive schemes put
in place to remunerate traders (Neumann et al.,2015). The
moral hazard they face can be described as follows: in order
to maximize their own profit they could simply increase their
trading activity as they are rewarded based on lot sizes that
are executed in form of copied trades in their followers’
broker accounts. Irrespective of the outcome of a trade, the
trader would benefit from the transaction. However, traders
can earn more if they attract more capital by improving the
track record, i.e. the overall performance. For example, one
could try to improve the winratio in terms of the number of
trades that yield a positive return. One could also face a
moral pressure to increase performance when followers
invest more money and trust in the trader’s abilities.
If the trading performance is poor, investors can imme-
diately withdraw their funds and the trader’s revenue would
drop accordingly. In this regard, the platform mechanics on a
social trading platform are particularly unique, offering
solutions for that kind of moral hazard related issues:
First, monitoring capabilities of investors are fundamen-
tally different compared to asset management. Every investor
can monitor the trader’s behavior at any point in time on a
per trade basis in every detail. According to principal-agent
theory, the capability of the principal to monitor the agent is
negatively related to the moral hazard of the agent.
Second, with linked brokerage accounts, investors’ reac-
tion time is reduced to near-real time. Penalties for undesired
behavior can thus be imposed immediately by every investor.
A withdrawal of invested capital is the most radical. Leaving a
comment on the public profile of the trader or providing a
rating visible to others are less intense. These so called social
feedback mechanisms (Ba and Pavlou, 2002; Pavlou et al.,
2007) have been found to improve user generated content in
online community setups (Siering and Muntermann, 2013).
Third, the incentives of volume-based compensation should
intuitively expose traders to the temptation of hazardously
maximizing their monetary outcome. Consequently, traders
would increase the order quantity in order to increase their
compensation per lot copy-traded in a follower account.
Fourth, traders that are not invested with their own money
simply have an incentive profile that is similar to payout
profiles of buying a stock option. The trader can earn positive
returns with no monetary risks besides the premium (in this
case invested time in trading on the platform).
In summary, a trader on Zulutrade has strong incentives
(volume-based remuneration) to behave morally hazardous.
However, the trader is aware of being monitored in real-time
and in detail, while the reaction time of the followers can also
be near real-time. Hence, principal-agent theory suggests that
the trader is more likely to behave in the interest of the
follower. These considerations lead us to the following
hypothesis:
H5 Changes in the amount of invested capital improves a
trader’s performance.
Investors can use the rating scale on a trader’s profile to
(dis-)approve of recent decisions. Changes of the average
rating of a trader indicate the general activity of investors on
the profile of the trader. A change of the rating therefore
shows an increased monitoring activity of investors. Conse-
quently, we hypothesize:
Effects of transparency F Glaser and M Risius
H6 Changes in social feedback lead to a better trading
performance.
An overview of the previously derived research model is
depicted by Figure 3. We attach the variables that we
operationalize to test our hypotheses to the theoretical concepts.
In the next section we explain the data set and the
methodology, which we applied for empirical validation of
our proposed model.
Empirical analysis
Data set
Through an advanced data scraping method, we were able to
extend the focus of previous studies on social trading or
investment delegation decision beyond the single trade focus,
by simultaneously collecting a data set that includes both
trader profiles and trade information.
Data regarding traders on Zulutrade is publicly available
and was gathered in an automated and structured way: We
scraped all trader profiles on a daily basis between 10 pm and
12 pm UTC + 2. The raw profile data set comprises profile
snapshots of around 33.000 traders for each day between
06/2013 and 01/2014. Historical trades for every unique
trader profile were obtained subsequently, resulting in almost
9 million trades over the years 2007 to 2014.
To reduce data noise and guarantee high data quality, the
original sample of profiles and trades were pre-processed as
follows: Initially, trade data was reduced to observations
within the time window for which we obtained the corre-
sponding profile data. A total of 3.79 million trades and
around 33.000 profiles per day remained. It is worth noting
that this vast number of daily accounts contains a high share
of inactive or test accounts, which we dropped consequently.
The platform is open for everyone to register and start
trading, i.e., providing signals. On the one hand, this
unrestricted access policy leads to a broad diversity of
traders, i.e. professional and less experienced traders. On the
other hand, many users register only for testing purposes and
churn after a short period of time.
Based on a rule of thumb, we only consider traders that
have been trading on more than two days within the
observation window and for which the distance between their
first and their last trade is more than four weeks to exclude
test accounts. This reduces noisy data from the profiles set
that actually might have one or more followers but only
remain active for the four weeks test phase of a demo
account. After removing all traders that are not active for at
least four weeks, the final sample contains 201 traders over
216 days. Lastly, it is worth noting that – since we are
working with lagged variables – the subset of traders for
which we do not observe at least three consecutive trading
days are dropped during the estimation procedure. Hence the
results presented are estimated with n =178 traders for
T=107 distinct days.
The trade data sample is aggregated on a daily basis. Trades
that were opened and trades that were closed are counted
separately for each day. Figures that are containing the
numbers of both opened and closed trades on the same day
are marked with the suffix ‘‘Total’’. To increase internal
validity, we dropped the numerically negligible non-FX asset
classes (e.g., Gold, Oil or the S&P 500 Index). Lastly, due to
our focus of socially biased trading behavior, we only
consider paper-traders who are not affected by additional
Performance Experience
(Active Days)
Moral Hazard
(Capital Invested)
Timing
(Duration)
Behavior / Style
Social Relationship
Controls
Social Feedback
(User Rating )
Disposition Effect
(Distance Highest)
H
5
H
6
H
1
H
2
Focus
(Number of Pairs)
H
3
Activity
(Number of Trades)
H
4
Figure 3 Overview of relationships according to our research model. Variables used to capture the proposed construct are mentioned in parentheses for each
construct.
Effects of transparency F Glaser and M Risius
financial return strategies. In total, the data sample com-
prised 428,153 single trades for the aforementioned 178
traders over a period of 107 days.
Variable operationalization
To approximate social feedback we included ‘‘Rating’’ as the
average rating from all followers who have invested capital
into a traders’ strategy ranging from 0 to 5.
Moral hazard is measured through ‘‘Capital’’, which repre-
sents the total amount of USD that is invested in the traders’
strategy and is observed on a daily basis. As typically found in
scale-free networks, capital has a skewed power law distribution,
because a small number of traders capture the major share of the
total capital invested (Newman and Park, 2003). Nonetheless,
50% of the traders are entrusted several thousand $USD. To
avoid potential biasing effects of the skewed distribution to a
linear regression model, we considered (lagged) first order
differences for our further analysis (Wooldridge, 2010).
Performance of a trader is measured by the ‘‘winratio’’ as
discussed in previous chapters. Winratio is defined on a daily
basis for each trader. It represents the share of trades with a
positive return that a trader closed on a particular day. The share
was derived from the unprocessed dataset where winning trades
are flagged with a dummy that is set to 1 if the trade yields a
positive return on close and is set to 0 in the case of a loss. Thus,
the winratio can take values between 0 (i.e., only trades with
negative or zero returns) and 1 (i.e., only trades with positive
returns) for each trader for any given day (Table 1).
To measure the disposition effect, we considered ‘‘Dis-
tanceHighest’’ as the number of days between the maximum
return that the trader could have realized with that particular
trade and the point in time the trade was actually closed by
the trader. This information is retrospectively reported by the
platform operator for each single trade. Thus, this variable
captures the effects of keeping less favorable trades open in
the hope for higher returns in the future which is in line with
the known aversion to realize a loss. A ‘‘DistanceHighest’’
around 0 means to close the trade around the (current)
optimum, which can also be premature, i.e. closing a trade as
long as its return is increasing.
With ‘‘Duration’’ we capture a trader’s timing represented
by the time between the opening time of a trade and its closing
time denoted in days. We compute the average duration over
all trades that are closed on a daily basis for every trader. Last,
to include the behavioral patterns in terms of a trader’s focus
and activity, we considered ‘‘PairsTotal’’ (count of unique
currency pairs traded) and ‘‘TradesTotal’’ (the total number of
trades handled, i.e., opened and closed). Both are calculated
for each trader and each day of the sample.
The pairwise correlations (Table 2) between these mea-
sures are generally low except for the average duration of
trades paired with the distance to highest price, which are
inherently related. Otherwise, the cross-correlations are well
within acceptable limits.
Methodology
To obtain answers to our research question, we conducted a
fixed-effects panel regression. The final model is represented
by Equation (1).
Winratioit ¼aiþb1DCapitalit þb2DCapitalit1
þb3DCapitalit2þb4DRatingit þb5DRatingit 1
þb6DistanceHighestit þb7Durationit
þb8PairsTotalit þb9TradesTotalit
þb10TradesTotalit 1þb11Experienceit þeti
ð1Þ
Table 1 Descriptive statistics of operationalized variables
Variable N Mean Median Minimum Maximum SD
Winratio 13,852 0.81 1.00 0.00 1.00 0.32
Capital 13,852 237,558.45 4400.92 0.00 12,374,490.00 902,516.37
Rating 13,852 2.35 3.00 0.00 5.00 1.80
DistanceHighest 13,852 0.76 0.00 0.00 30.00 2.48
Duration 13,852 1.92 0.58 0.00 91.03 3.96
PairsTotal 13,852 3.61 3.00 2.00 29.00 2.43
TradesTotal 13,852 16.04 13.00 2.00 90.00 12.22
Experience 13,852 596.63 552.31 5.19 1858.06 388.73
Table 2 Pairwise correlations of variables
Variable 12345678
1 Winratio 1.00
2 Capital -0.01 1.00
3 Rating -0.07 0.10 1.00
4 DistanceHighest -0.35 0.00 -0.01 1.00
5 Duration -0.15 -0.03 -0.03 0.69 1.00
6 PairsTotal -0.17 0.06 0.04 -0.02 -0.05 1.00
7 TradesTotal -0.02 0.07 0.06 0.00 -0.01 0.19 1.00
8 Experience 0.04 0.02 0.10 0.08 0.10 -0.03 -0.04 1.00
Effects of transparency F Glaser and M Risius
This approach enabled us to control for (1) any invariant,
a priori idiosyncratic differences among the traders (i.e.,
personal traits, situational effects) through the restricted
fixed effects assumption, (2) for a temporal learning trend
(i.e., experience), and (3) the (dummy encoded) day specific
changes (which are not explicitly displayed in the tables or
equation for the sake of clarity) additionally to the consid-
eration of the hypothesized focal effects of trader style and
social influences (Wooldridge, 2010).
These aspects were controlled for considering that each
trader in the sample could be endowed with unique tools
or distinguishing personal traits. We selected a fixed-effects
model, to control for individual performance altering
effects because the present work focuses on systematic
patterns common for all traders (Wooldridge, 2010). Time
dummy variables for each day were included to rule out
time-specific effects that systematically influence traders
(e.g., seasonal effects, holidays, weather changes) during
certain periods of time (Wooldridge, 2010). Considering,
that traders learn over time and improve their perfor-
mance (Chiang et al.,2011), we controlled for the
progressive learning of a trader through the number of
days since the trader’s first trade on the platform (i.e.,
experience).
Lastly, we included lagged variables of Capital, Rating, and
the number of trades to enable inferences of chronological
causality (i.e., repeating occurrence of events that follow each
other with reasonable time delay) As reported by Abbey and
Doukas (2015), a day trader’s trading activity in a previous
period influences the performance of a trader in the current
period. Regarding the interpretation of the metrics, we do
not consider the (generally rather persistent) levels of these
variables, but the first order differences in comparison to the
previous period. For example, DCapitalit 1denotes the
absolute change of capital invested in trader i between
period t-2 and t-1. We tested models with higher lag
lengths but could not find any remarkable differences
regarding model fit or specification analytical metrics.
Besides, additional lags were insignificant.
Robustness considerations
During the course of the description of our analysis, we
documented additional approaches taken to test our results
against outliers, skewed distributions of explanatory variables
and sub-sampling issues. In this section we briefly take
account of decisions regarding our econometric estimation
procedure that could systematically bias the results.
First of all, a test for unobserved effects in the individual
residuals rejects the null hypothesis and, thus, justifies the
modelling of individual effects through a random- or fixed-
effects panel regression (Wooldridge, 2010). The choice of fixed
effects estimation over random effects estimation was based on
the rejection of the null hypothesis in the Hausman test
procedure for misspecification (Hausman and Taylor, 1981).
Lastly, we considered serial-correlation and heteroscedasticity
patterns of error terms commonly found in panel regressions.
Our panel data is not suffering from being a ‘‘short’’ panel set
and hence the standard tests for serial correlation are feasible
(Wooldridge, 2010). Both Breusch-Godfrey and Durbin-Wat-
son tests for serial correlation both reject the null-hypothesis
indicating serial correlation. Hence, we need tocorrect for these
effects by using robust standard errors (Arellano, 1987).
Empirical results
The results of our empirical model validation with standard-
ized beta coefficients are presented in Table 3.
Overall, the results support our underlying assumption that
on social trading platforms, the trader performance is affected
by behavioral and interaction features of the platform
(adjusted R
2
=13.77%, p\.001). In particular, the results
also support the previously hypothesized relationships.
Regarding the effect of timing (hypothesis 1) and disposition
(hypothesis 2), we see that both features significantly affect
trader performance. Considering the correlation of both
metrics, this means that the duration of a trade is only positive
as long as the position is running in the favorable direction, i.e.
increasing return. As soon as the position is running in the
wrong direction, i.e., moving away from highest price and
possibly into the loss zone, the duration of the trade is no
Table 3 Coefficients are reported as beta coefficients (also frequently referred to as standardized coefficients)
Dependent variable: winratio (t)
Hypotheses Variable Coefficient p value Sign.
H1: Timing Duration 0.1294 0.0001 ***
H2: Disposition DistanceHighest -0.4053 0.0000 ***
H3: Focus PairsTotal -0.2043 0.0002 ***
H4: Activity TradesTotal -0.0346 0.3036
TradesTotal (t-1) 0.0616 0.0115 **
H5: Moral DCapital 0.3212 0.5363
DCapital (t-1) 0.8889 0.0013 ***
DCapital (t-2) 0.5724 0.0674 *
H6: Social feedback DRating 0.2219 0.0043 ***
DRating (t-1) -0.0231 0.8382
Controls Experience 0.0771 0.0294 **
Dummies Time included
Adjusted R
2
0.1377
Fstatistics 63.0005 0.0000 ***
Effects of transparency F Glaser and M Risius
longer favorable. The median of DistanceHighest was 0,
indicating that the majority of trades were closed as long as the
trade was running in the favorable direction. Put differently,
the trade was closed at the so far highest return of that trade.
Thus, the results confirm hypotheses H1 and H2.
As suggested by our third hypothesis H3, the number of
traded pairs measuring the focus of the trader is influencing
the performance outcome. The more focused a trader, with
respect to asset classes, the better the performance of the
trades.
Hypothesis H4 is in accordance with the findings of Abbey
and Doukas (2015), showing that a higher level of activity in
the previous period indicates a higher performance in the
current period. To test for collinearity issues, we exclude the
lag of TradesTotal and find the results to be stable.
TradesTotal in the current period remains insignificant.
In line with hypotheses H5, we find an increase in the capital
invested in the preceding days to influence the daily winratio.
We assume that two forms of influence account for this
observation. First, it indicates that the trader assumes to be
monitored in the current period, i.e., immediately after new
investments, and is trying to please the new investor with
positive returns. Second, it shows that a change in investments
on the previous day causes the trader to become more prone to
the disposition effect. This finding shows the link between real-
time monitoring capabilities of a principal and the behavioral
bias of the trader. It needs to be noted that the effect is only
valid with respect to a change in the previous period.
While the results indicate support regarding the proposed
social feedback hypothesis H6, the absence of a lagged effect
does not allow inferring (chronological) causality from the
finding. Although we see a significant coefficient for the
current period first difference, we cannot establish a clear
chronological order. Rating is obtained on an end of day
basis, whereas winratio is determined within that same day.
Thus, apart from the theoretical reasoning, we cannot
provide final empirical evidence that the rating increased
during the day due to the good performance of the trade, or
if the trader’s performance is influenced by a higher rating in
comparison to the previous day.
Discussion
The goal of this study was to investigate how trader
performance can be enhanced through the transparency
and social interactions on social trading platforms. For this
purpose, we collected and analyzed a panel of 178 trader
profiles and their respective trade information over more
than 100 days from a leading social trading platform –
Zulutrade. Contrary to assumptions of the efficient market
hypothesis, our results generally show that social trader
performance is subject to behavioral (i.e., Disposition) and
relational biases (i.e., Moral-Hazard, Social Feedback) com-
parable to incumbent investment environments. Further-
more, we are able to closely consider time-dependent
interacting effects of the prominent disposition and moral
hazard biases that impair trader returns. The disposition
effect describes the tendency of a trader to prematurely close
winning trades or overrun losing trades. Our results show
that the disposition effect intensifies with a anteriorly
increase in moral hazard expressed through capital entrusted
to a trader. In other words, paying more attention to a trader
enhances the trader’s urge to avoid losses even if it means
passing on optimal returns. The reported effects remain
stable under simultaneous control for known performance
drivers such as trade duration, trade frequency, lagged trade
frequency, investment focus, and experience of the trader.
Lastly, one needs to acknowledge that the present empirical
results do not necessarily indicate the assumed causal
dependency between monitoring proxies and winratio. The
positive relationship states that decreasing ratings or capital is
associated with a decreasing winratio. Alternatively, it could
be assumed that the effect is moderated through additional
variables. This could, for example, be that a reduced
monitoring from users decreases the trader’s effort or the
willingness to trim publicly displayed performance metrics. It
is also possible that traders focus on higher returns instead of
higher winratios. Therefore, they need to take more risk and
hence suffer from a higher share of trades they need to close
with a loss. These additional explanations, however, do not
repeal the relationship but offer venues for further research
on social trading.
Conclusion
While the efficient market hypothesis assumes stock prices to
be purely rationality based, trader performance in offline-
settings has been found to be affected by a set of behavioral
and social biases. Due to their key features of transparent
communication and social relations, social trading platforms
offer the unique opportunity to study behavioral and social
impacts on trading behavior. To the best of our knowledge,
we are the first to combine profile and trade data from a
social trading platform to analyze comparable effects in an
upcoming online real-time environment of trader investor
relationships. Thereby, we confirm the disposition effect for
traders in the online environment of a social trading
platform. This contributes to financialization research, con-
sidering that social trading represents an associated phe-
nomenon in terms of technologically disintermediating
platforms that enable maximizing individual financial profits
through speculative trading. In particular, we exploit unique
characteristics of the winratio performance measure and
observe that traders tend to be more prone to the disposition
effect in the current period if their rating or the invested
capital has changed in the previous period. We conclude that
traders try to improve their performance metrics when they
receive more attention from the community, i.e., capital
investments in the previous period or changes in the average
user rating. The reported effects remain stable while con-
trolling for established performance drivers such as trade
duration, trade frequency, lagged trade frequency, attention
level and experience of the trader.
These insights are relevant from a theoretical as well as
practical perspective. Foremost, we extend the theoretical
knowledge regarding the drivers of the prominent and well-
studied disposition effect in an innovative online trading
environment. Thereby, we further supplement findings from
behavioral finance contradicting the efficient market hypoth-
esis. While social trading platforms differ substantially from
the common trading environment regarding levels of trans-
parency and types of relationships, the disposition effect is
stable across settings. Even more so, the augmented exposi-
tion to investor attention seems to enhance the disposition
Effects of transparency F Glaser and M Risius
effect while decreasing the trader performance significantly.
Additionally, we extend the literature on principal-agent
theory in the digitized financial service world. We show that
full transparency and exertion of control does not necessarily
introduce positive effects. The behavior of the agent is
influenced by recognition of being monitored. Hence, the
principal could inadvertently induce unfavorable behavior of
the agent through monitoring. Thus, we also obtain insights
on the effects of financialization related issues regarding the
impact of increased transparency and deregulation (Bucha-
nan, 2016). Moreover, we provide insights on trader behavior
and the impact of social trading service features. Making
traders and their performance visible to the community and
enabling social feedback from investors in terms of ratings,
helps to make traders act more conservatively and temporar-
ily increase the winratio.
Therefore, platform operators can refer to these findings to
improve or measure the impact of different incentive struc-
tures and monitoring capabilities provided on their platform.
Based on these results, we argue that operators need to decide
which type of trading behavior they would like to encourage on
their platforms and design the features accordingly. If, on the
one hand, high risk investment strategies with potentially
maximized returns are desirable, platforms should offer only
restrictive attention signaling features such as subjective user
ratings. On the other hand, however, if they want to encourage
rather risk-aware trading, social feedback mechanisms help to
establish such behaviors. Furthermore, regulators are provided
with insights regarding recently emerging financial services. As
customer protection is frequently postulated and enforced by
regulatory authorities, our findings can provide helpful
guidance for future regulatory discussions. As such, we are
able to demonstrate effects of an augmented transparency of
trader behavior for regulatory and investment entities. Specif-
ically, when investors can openly compare success measures of
traders, it causes traders to trim their performance metrics eve n
though it might not yield optimal returns. Thereby, we also
address the major global aspect of financialization regarding
market deregulation (Buchanan, 2016).
However, these contributions need to be considered in the
light of the study’s limitations, which simultaneously provide
venues for future research. In its current state, the findings
might be subject to a sampling or survivorship bias. Due to our
filtering methodology of only using traders who have con-
ducted at least two trades on the platform, one might argue that
more skilled traders remained in the sample which explains
higher winratios. However, it needs to be noted that such a
sample bias would imply that the obtained coefficients are
more likely to underestimate the actual effect size. Considering
that skilled traders should be less prone to react to the external
and, at least for their strategy, irrelevant influences like changes
of capital invested or changing of ratings. Especially since
traders on Zulutrade are completely independent of the
amount invested in their strategy. This is different for funds
and traditional asset management setups, where traders would
benefit from the volume of the portfolio they manage.
However, further research is needed to investigate the
assumption that this sample bias rather underestimates the
effect sizes.
Regarding the generalizability of our findings, it needs to
be considered that the data was obtained from one social
trading platform, which is structurally different from other
platforms (e.g., imposes less controls and constraints regard-
ing admission criteria or track record evaluation for traders).
Consequently, the environment could encourage rather
hazardous trading behavior.
Lastly, further research in this regard could focus on
changing the design of functionalities for communication
between traders and followers or more elaborate recommen-
dation systems for traders or trading strategies or incorporate
other dimensions like changes of risk taking behavior in the
presence of higher levels of followers and entrusted capital.
This would enable to analyze more closely how risk taking is
influenced by social presence, interaction or social function-
alities in online trading environments.
References
Aalbers, M. (2008). The Finacialization of Home and the Mortage Market Crisis,
Competition & Change 12(2): 148–166.
Aalbers, M. (2015). Corporate Financialization, The International Encyclopedia of
Geography: People, the Earth, Environment, and Technology, Oxford, UK: Wiley.
Abbey, B.S. and Doukas, J.A. (2015). Do Individual Currency Traders Make
Money? Journal of International Money and Finance 56: 158–177.
Alt, R. and Puschmann, T. (2012). The Rise of Customer-Oriented Banking –
Electronic Markets Are Paving the Way for Change in the Financial Industry,
Electronic Markets 22(4): 203–215.
Amman, M. and Schaub, N. (2016). Social Interaction and Investing: Evidence
form an Online Social Trading Network [WWW document] https://www.rsm.
nl/fileadmin/home/Department_of_Finance__VG5_/PAM2016/Final_Papers/
Nic_Schaub.pdf (accessed 25 August 2016).
Antweiler, W. and Frank, M.Z. (2004). Is All That Talk Just Noise? The
Information Content of Internet Stock Message Boards, The Journal of Finance
59(3): 1259–1294.
Arellano, M. (1987). Computing Robust Standard Errors for Within-groups
Estimators, Oxford Bulletin of Economics and Statistics 49(4): 431–434.
Ba, S. and Pavlou, P.A. (2002). Evidence of the Effect of Trust Building
Technology in Electronic Markets: Price Premiums and Buyer Behavior, MIS
Quarterly 26(3): 243–268.
Baghestanian, S., Gortner, P.J. and van der Weele, J. (2015). Peer Effects and
Risk Sharing in Experimental Asset Markets, SAFE Working Paper No. 67.
Barber, B. M. and Odean, T. (2002). Online Investors: Do the Slow Die First?
Review of Financial Studies 15(2): 455–487.
Ben-David, I. and Hirshleifer, D. (2012). Are Investors Really Reluctant to
Realize Their Losses? Trading Responses to Past Returns and the Disposition
Effect, Review of Financial Studies 25(8): 2485–2532.
Bhattacharya, S. and Pfleiderer, P. (1985). Delegated Portfolio Management,
Journal of Economic Theory 36(1): 1–25.
Bjønnes, G.H. and Rime, D. (2005). Dealer Behavior and Trading Systems in
Foreign Exchange Markets, Journal of Financial Economics 75(3): 571–605.
Bjønnes, G.H., Rime, D. and Solheim, H.O. (2005). Liquidity Provision in the
Overnight Foreign Exchange Market, Journal of International Money and
Finance 24(2): 175–196.
Breedon, F. and Ranaldo, A. (2013). Intraday Patterns in FX Returns and Order
Flow, Journal of Money, Credit and Banking 45(5): 954–965.
Buchanan, B.G. (2016). The Way We Live Now: Financialization and Securiti-
zation, Research in International Business and Finance Advance Online
Publication 23 January, doi:10.1016/j.ribaf.2015.11.019.
Chen, H., De, P., Hu, Y. and Hwang, B.-H. (2014). Wisdom of Crowds: The
Value of Stock Opinions Transmitted Through Social Media, Review of
Financial Studies 27(5): 1367–1403.
Chen, X. and Lai, Y.-J. (2015). On the Concentration of Mutual Fund Portfolio
Holdings: Evidence from Taiwan, Research in International Business and Finance
33: 268–286.
Chiang, Y.-M., Hirshleifer, D., Qian, Y. and Sherman, A.E. (2011). Do Investors
Learn from Experience? Evidence from Frequent IPO Investors, Review of
Financial Studies 24(5): 1560–1589.
Cornell, B. (2013). What Moves Stock Prices: Another Look, Journal of Portfolio
Management 39(3): 32–39.
Effects of transparency F Glaser and M Risius
Currie, W.L. and Lagoarde-Segot, T. (2016). Call For Papers: Finacialization and
Information Technology, Journal of Information Technology.
Dewally, M., Ederington, L. and Fernando, C. (2009). Determinants of Trader
Profits in Commodity Futures Markets, Review of Financial Studies 26(10):
2548–2683.
Doering, P., Neumann, S. and Paul, S. (eds.) (2015). A Primer on Social Trading
Networks: Institutional Aspects and Empirical Evidence.
Epstein, G.A. (2005). Financialization and the world economy, Cheltenham, UK:
Edward Elgar Publishing.
Evans, A.L. (2008). Portfolio Manager Ownership and Mutual Fund Perfor-
mance, Financial Management 37(3): 513–534.
Falkner, R. and Corsthwaite, D. (2013). Mirror Trading and Social Media Marketing:
Reulatory and Business Risks [WWW document] http://de.slideshare.net/black
swanpartners/black-swan-partners-reedreed-smithsocialtradingmarketingseminar
booklet (accessed 29 August 2016).
Fama, E.F. (1970). Efficient Capital Markets: A Review of Theory and Empirical
Work, The Journal of Finance 25(2): 383–417.
Fama, E.F. (1998). Market Efficiency, Long-Term Returns, and Behavioral
Finance, Journal of Financial Economics 49(3): 283–306.
Farnsworth, H. and Taylor, J. (2006). Evidene on the Compenstation of Portfolio
Managers, The Journal of Financial Research 29(3): 305–324.
Fernandes, J.L.B., Pen
˜a, J.I. and Tabak, B.M. (2010). Delegated Portfolio
Management and Risk-Taking Behavior, The European Journal of Finance 16(4):
353–372.
Garner, C. (2012). Currency Trading in the Forex and Futures Markets, New Jersey,
USA: FT Press.
Hausman, J. and Taylor, W. (1981). Panel Data and Unobservable Individual
Ects, Econometrica 49: 1377–1398.
Hiraki, T., Liu, M. and Wang, X. (2015). Country and Industry Concentration
and the Performance of International Mutual Funds, Journal of Banking &
Finance 59: 297–310.
Holmstrom, B. and Milgrom, P. (1991). Multitask Principal-Agent Analyses:
Incentive Contract, Asset Ownership, and Job Design, Journal of Law,
Economics, & Organization 7(Special Issue): 24–52.
Jacobs, H. and Weber, M. (2015). On the Determinants of Pairs Trading
Profitability, Journal of Financial Markets 23: 75–97.
Jøsang, A., Ismail, R. and Boyd, C. (2007). A Survey of Trust and Reputation
Systems for Online Service Provision, Decision Support Systems 43(2): 618–644.
Keynes, J.M. (2013). The state of long-term expectation (4th ed.), New York, USA:
Cambridge University Press.
Lagoarde-Segot, T. (2016). Financialization: Towards a New Research Agenda.
International Review of Financial Analysis, International Review of Financial
Analysis advance online publication 3 April, doi: 10.1016/j.irfa.2016.03.007.
Lee, W. and Ma, Q. (2015). Whom to Follow on Social Trading Services?
A System to Support Discovering Expert Traders, in Proceedings of the10th
International Conference on Digital Information Management (ICDIM),
pp. 188–193.
Liu, Y.-Y., Nacher, J. C., Ochiai, T., Martino, M. and Altshuler, Y. (2014).
Prospect Theory for Online Financial Trading, PLoS ONE 9(10): 1–7.
Lo, A.W. (2004). The Adaptive Markets Hypothesis: Market Efficiancy from an
Evolutionary Perspective, Journal of Portfolio Management 30(5): 15–29.
Neumann, S., Paul, S. and Doering, P. (2015) A Primer on Social Trading:
Remuneration Schemes, Trading Strategies and Return Characteristics, in
Proceedings of the EFMA Annual Meetings 2015.
Newman, M.E.J. and Park, J. (2003). Why Social Networks are Different from
Other Types of Networks, Physical Review E 68(3): 1–8.
Oehler, A., Horn, M. and Wendt, S. (2016). Benefits from Social Trading?
Empirical Evidence for Certificates on Wikifolios, International Review of
Financial Analysis 46: 202–210.
Orlean, A. (2012). Knowledge in Finance: Objective Value versus Convention.
Northamption, MA: Edward Elgar Publishing.
Pan, W., Altshuler, Y. and Pentland, A. (2012) Decoding Social Influence and
the Wisdom of the Crowd in Financial Trading Network, in Privacy, Security,
Risk and Trust (PASSAT), 2012 International Conference on and 2012
International Confernece on Social Computing (SocialCom), pp. 203–209.
Pavlou, P.A., Liang, H. and Xue, Y. (2007). Understanding and Mitigating
Uncertainty in Online Exchange Relationships: A Principal–Agent Perspective,
MIS Quarterly 31(1): 105–136.
Pelster, M. and Breitmayer, B. (2016) Swarm Intelligence? Stock Opinions of the
Crowd and Stock Returns [WWW document] http://dx.doi.org/10.2139/ssrn.
2787744 (accessed 25 August 2016).
Philippon, T. (2015). Has the U.S. Finance Industry Become Less Efficient?
American Economic Review 105(4): 1408–1438.
Robinson, D. (2014). Return of the daytrader: Can You Earn a Living by Copying
Other Investors? [WWW document] http://www.telegraph.co.uk/finance/
personalfinance/investing/11123162/Return-of-the-daytrader-can-you-earn-a-
living-by-copying-other-investors.html (accessed 29 August 2016).
Rosov, S. and Foster, F.D. (2014). Customer Foreign Exchange Orders: When
Timing Really Does Matter, Australian Journal of Management 39(3): 351–368.
Ross, S.A. (2004). Compensation, Incentives, and the Duality of Risk Aversion
and Riskiness, The Journal of Finance 59(1): 207–225.
Sapienza, P. and Zingales, L. (2012). A Trust Crisis, International Review of
Finance 12(2): 123–131.
Shefrin, H. and Statman, M. (1985). The Disposition to Sell Winners Too Early
and Ride Losers Too Long: Theory and Evidence, The Journal of Finance 40(3):
777.
Shiller, R.J. (2005). Irrational Exuberance (2nd ed.), New Jersey, USA: Princeton
University Press.
Shiller, R.J. (2015). Irrational Exuberance (3rd ed.), New Jersey, USA: Princeton
University Press.
Siering, M. and Muntermann, J. (2013). How to Identify Tomorrow’s Most
Active Social Commerce Contributors? Inviting Starlets to the Reviewer Hall of
Fame, in Proceedings of the International Conference on Information Systems
2013, Milan, Italy.
Stoughton, N.M. (1993). Moral Hazard and the Portfolio Management Problem,
The Journal of Finance 48(5): 2009–2028.
Weber, M. and Camerer, C.F. (1998). The Disposition Effect in Securities
Trading: An Experimental Analysis, Journal of Economic Behavior & Organi-
zation 33(2): 167–184.
Wohlgemuth, V., Berger, E.S. and Wenzel, M. (2016). More than Just Financial
Performance: Trusting Investors in Social Trading, Journal of Business Research
69(11): 4970–4974.
Wooldridge, J.M. (2010). Econometric Analysis of Cross Section and Panel Sata
(2nd ed.), Cambridge, USA: MIT Press.
Zhao, J.C., Fu, W.-T., Zhang, H., Zhao, S. and Duh, H. (2015). To Risk or Not
to Risk? in The 18th ACM Conference (Vancouver, BC, Canada), pp. 95–104.
Zulutrade. (2016). Zulutrade Homepage [WWW document] http://www.
zulutrade.com/index (accessed 29 August 2016).
About the Authors
Florian Glaser is a Ph.D. candidate at Goethe University,
Frankfurt, Germany. His research interests are in the area of
innovations in electronic financial networks in general and in
particular in algorithmic trading and blockchain system
design.
Marten Risius is a Postdoc at the University of Mannheim,
Germany. His research interests are in the areas of digital-
ization, social media management, and social media analyt-
ics. His articles have been published in JSIS, Information &
Management, and in the conference proceedings of the ECIS,
the HICSS, and the AMCIS.
Effects of transparency F Glaser and M Risius
