0% found this document useful (0 votes)
6 views19 pages

Holmberg2020

This paper explores the relationship between emotions, consciousness, and stock market returns, revealing a statistically significant correlation between random numbers from the Global Consciousness Project (GCP) and stock market indices. The findings suggest that variations in GCP data may precede market trades, challenging existing paradigms of consciousness and highlighting the need for further research. The implications of these results extend beyond economics, suggesting a deeper connection between collective consciousness and market behavior.

Uploaded by

Sergei Pashin
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
6 views19 pages

Holmberg2020

This paper explores the relationship between emotions, consciousness, and stock market returns, revealing a statistically significant correlation between random numbers from the Global Consciousness Project (GCP) and stock market indices. The findings suggest that variations in GCP data may precede market trades, challenging existing paradigms of consciousness and highlighting the need for further research. The implications of these results extend beyond economics, suggesting a deeper connection between collective consciousness and market behavior.

Uploaded by

Sergei Pashin
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

Ulf Holmberg

Stock Returns and the Mind


An Unlikely Result that Could Change
Our Understanding of Consciousness
For personal use only -- not for reproduction
Copyright (c) Imprint Academic

Abstract: Emotions and feelings affect economic systems. This is well


known as e.g. stock markets tend to react to sudden political and
emotional events. However, the link between emotions, consciousness,
and economic systems at a deeper level than the aggregate resulting
action of people at large is yet to be explored and understood. In this
paper, a first building block is presented as it is shown that a variable
derived from the random numbers obtained by the Global Conscious-
ness Project is statistically related to various well-known stock market
index returns. The relationship is shown to be non-linear and that
variations in the variable, to some extent, predate the underlying
trade. The results presented are found to be robust and qualitatively
unaffected by the removal of outliers. Apart from the pure economic
value of these findings, the results have truly baffling implications.
This is the case as they confirm some previous unorthodox research
suggesting that consciousness stretches out beyond the locally con-
fined space of our heads and that consciousness can affect hardware-
generated random numbers at a distance. Thus, these results put
doubt on the existing paradigm with regards to consciousness and
highlight the need for further research.

1. Brief Introduction
This paper shows that random numbers generated by the Global
Consciousness Project (GCP) significantly correlate with stock market
returns. This topic is investigated since market prices are the result of
Correspondence:
Ulf Holmberg, PhD, Independent Researcher, Kammakargatan 50, 111 60,
Stockholm, Sweden. Email: [email protected]

Journal of Consciousness Studies, 27, No. 7–8, 2020, pp. 31–49


32 U. HOLMBERG

investors’ collective decisions on worth and thus should be affected by


events similar to the events that are claimed to affect the GCP data.
This suggests that there could exist some correlation between the GCP
data and stock market returns which is empirically tested for with
highly significant results. The relationship found is shown to be non-
linear as it can be approximated using a second- or third-degree
polynomial. Also, and somewhat surprisingly, variations in the GCP
data variable seem to predate the underlying trade. The paper ends
with a section discussing these results and suggests future research.

2. Economics, Market Prices,


and the Will of the Many
For personal use only -- not for reproduction

Within the field of economics, the market price is a term that refers to
the price at which an asset or service can be bought or sold. Economic
Copyright (c) Imprint Academic

theory further suggests that the market price converges to the ‘clearing
market price’, which is the point where the forces of supply and
demand meet. Since the market price depends on the demand and
supply of a good or service, it is also likely to change regularly in
order to adapt to events that may affect the underlying demand and
supply functions.
Sudden and unexpected macroeconomic and political occurrences
could be classified as such events and arguably such events could
cause the market price to be re-evaluated and changed. This since the
events could affect consumers (demand side shifts), producers (supply
side shifts), or even change the ‘rules of the game’ (i.e. regulatory
changes). However, in the absence of such events, the market price
could be viewed as stable and as an emergent result of the ‘sponta-
neous coordination of the plenty’.1 But for the coordination to occur,
economic agents need to ‘know what they want’, have preferences,
and be able to order and rank the choices presented to them (see e.g.
Mas-Colell, Whinston and Green, 1995). All these factors thus
implicitly assume that some form of conscious behaviour as con-
sumers, for instance, maximizes some underlying utility function
while firms simply seek to maximize profits (and have the cognitive
ability to do so).2 The maximization procedures referred to above are

1 For a discussion on the emergent properties of market prices see Wang et al. (2018).
2 Even though such views on economic agents could be said to grossly oversimplify the
complexity of human behaviour, it should be noted that the underlying economic theory
is flexible enough to accommodate other important aspects of human will. Within e.g.
STOCK RETURNS & THE MIND 33

also likely to occur with ‘one’s own self-interest’ first in mind and, as
famously stated by Adam Smith (1776) in his work The Wealth of
Nations, ‘[markets are] led by an invisible hand to promote an end
which was no part of his intention’. Thus, the individual’s conscious
behaviour is likely to play an important part in the determination of
market prices.
It should, however, be noted that there exists no agreement within
the field of economics as to how the underlying functions that
economic agents seek to maximize should be formalized. Most econo-
mists would, however, agree that a ‘consumer preference set’ of some
sort is needed and that parts of such a preference set could be shared
with other economic agents.3 From this it could be understood that
For personal use only -- not for reproduction

market prices are partly dependent on a collective preference set and


that this also should be true for stock market prices. However, it is
noted that stock market prices, at least in theory, are only sensitive to
Copyright (c) Imprint Academic

investors’ beliefs about future firm profits and dividends (Sharpe,


1978). But, since profits and dividends are dependent on the indi-
vidual’s choice to consume, consumer preferences will implicitly
determine future firm profits. This is a mechanism that (theoretically)
should be understood and considered by investors when they form
their expectations about the future and set a firm’s price. Thus, also,
stock market prices are dependent on a collective preference set. But
what about daily stock market price movements?
An individual’s opinion, or some aggregate of such opinions, is
often referred to as sentiment. Sentiment can thus be thought of as a
subset of the drivers of market prices and thus possibly part of some
collective preference set that determines the market price of a firm.
Acknowledging this, I lean on the large bulk of empirical research that
shows that investor and market sentiment is indeed related to daily
stock market returns and note that the idea that market sentiment may
affect daily stock price movements has a long history within the field
of economics.4 As an example, Keynes (1936) associated the stock
market with a ‘beauty contest’ where participants devote their efforts
not to judging the underlying concept of beauty, but instead to

behavioural economics, economists seek to understand the motivations and reasons


behind individuals’ decisions.
3 Examples of this are that prices should be non-negative and that consumers value
having a good higher than not having a good at all.
4 Fisher and Statman (2003) and Daszyńska-Żygadło, Szpulak and Szyszka (2014) find a
positive relationship between investor sentiment and market returns.
34 U. HOLMBERG

‘anticipating what average opinion expects the average opinion to be’.


Also, Shiller (2017) argued for the importance of sentiment as
investors’ optimistic or pessimistic beliefs about the stock markets
may directly influence prices. The interested reader could consult the
excellent introduction in Lansing and Tubbs (2018) for a more
detailed discussion on sentiment and stock market prices while I in
this paper simply acknowledge that sentiment affects stock market
prices and that sentiment may be affected by daily stock market
beliefs.
The above reasoning suggests that changes to the collective
preference set are likely to affect daily stock price movements. The
above also suggests that e.g. emotional and engaging world events
For personal use only -- not for reproduction

could affect the collectively decided worth of a firm. As it is claimed


that the GCP data also may be affected by such events, it is argued
that daily stock market returns and the GCP data could be correlated.
Copyright (c) Imprint Academic

This is an unexplored research avenue and, since the inclusion of the


GCP data rests on some unorthodox research on the nature of con-
sciousness, a supporting discussion on consciousness follows in the
sections below.

3. A Brief Discussion on Consciousness


Consciousness is perhaps one of our greatest mysteries as no one
knows what it is, what it does, or even how it has emerged. The pre-
vailing working hypothesis, in most sciences, is however that con-
sciousness is solely the result of physical arrangements and informa-
tion processing patterns (see e.g. Güzeldere, 1997). This viewpoint
rests on the existence of neural correlates (see e.g. Cotterill, 2001;
Llinás, 2002; Koch, 2004; among others), but how the brain alone can
produce our subjective experiences (such as the feeling of warmth,
cold, or pain) is not yet understood. It is even a philosophical mystery
how non-conscious matter can give rise to sentient beings and this
unsolved philosophical conundrum is often referred to as the ‘hard
problem of consciousness’ (Chalmers, 1995; 2003).
From the above it can be read that our understandings of conscious-
ness are incomplete and that much more research is needed. However,
it could also be understood that most studies on consciousness focus
on explaining an individual’s conscious experience and not the will of
STOCK RETURNS & THE MIND 35

the many.5 Thus aspects related to collective decision making are


often ignored even though one notable exemption exists, the collective
consciousness concept within the field of sociology.
In sociology, a set of shared beliefs, ideas, attitudes, and knowledge
that is common to a social group or society is defined as a society’s
collective consciousness (Durkheim, 1893/1997). As consumers are
individuals, and as individuals act and consume within social groups
and economies, it could be argued that the collective consciousness à
la Durkheim affects the collective preference set of consumers and,
through it, market prices. Perhaps this is a concept of consciousness
that can be used for understanding changes in the collective consumer
preference set in economics. Perhaps it could also be said that the
For personal use only -- not for reproduction

collective consciousness of Durkheim should be correlated with the


GCP data, at least if the society is defined in a way that is aligned with
the definition underlying the GCP data (i.e. globally). However, the
Copyright (c) Imprint Academic

GCP data is special as it rests on the assumption that human con-


sciousness can stretch out beyond our heads and affect random
number generators at a distance. As this idea is not part of the most
prevalent theories on consciousness, a discussion on alternative
theories is provided in the following section.

4. Alternative Theories on Consciousness


and the GCP Data
There exist several alternative theories of consciousness that allow the
possibility that the mind stretches out beyond our heads and it should
also be noted that physics permits this possibility. This as the so-called
‘observer effect’ in quantum mechanics (a well-established physical
property of matter) describes that the observation of a quantum
phenomenon changes the phenomenon observed and studied. Even
though this does not necessarily require a conscious observer, the
observer effect seems to suggest that only the measurement of an
object (or event) initiates the transition from the ‘possible’ to the

5 Perhaps the problem with explaining what consciousness is originates from the prob-
lems faced in the definition of the concept. Consciousness could, for instance, be
defined as the state of being aware of and responsive to one’s surroundings, but since
such a definition (or similar versions of it) is imprecise, the term has also been defined
in terms of sentience alone, e.g. awareness, qualia, and subjectivity.
36 U. HOLMBERG

‘actual’ as the famous ‘wave function’ collapses.6 This thus suggests


that human measurement done at a distance affects quantum systems
at a distance.7 That consciousness can extend outside a human head
and interact with (say) a random number generator has also been
studied within the research field of parapsychology by, for example,
Nelson, Jahn and Dunne (1986), Radin et al. (2006), and Dunne and
Jahn (2007). The results from their studies suggest that consciousness
can do so and, taken together, it is noted that some research results
allow for the possibility of consciousness stretching out beyond our
heads.
Thus, some research findings suggest that consciousness has
properties that cannot be described purely using reductionist material
For personal use only -- not for reproduction

sciences alone, at least not as they are understood to date. Such


findings have thus resulted in several alternative hypotheses and
theories of consciousness, theories that attempt to ‘close the gap’
Copyright (c) Imprint Academic

between philosophy and material sciences. In, for example, Donald


Hoffman’s ‘conscious agents’ theory (2008; 2014), consciousness,
rather than space-time and physical objects, is fundamental. Even
though Hoffman is thought of as a consciousness realist, this is a
theory of consciousness shared by many modern philosophical
idealists (see e.g. Kastrup, 2018). Other notable examples include
electromagnetic theories of consciousness (see e.g. Pockett, 2012;
McFadden, 2002)8 and quantum brain dynamics theories (see e.g.
Atmanspacher, 2004; Van den Noort, Lim and Bosch, 2016).9
Some, but not all, of these alternative theories allow for conscious-
ness stretching out beyond the human head and some view conscious-
ness as fundamental, putting consciousness studies at the forefront of
academic exploration. From this it is noted that some of the alternative

6 Physicists have found that even passive observation of quantum phenomena can change
the measured result (see e.g. Buks et al., 1998).
7 It is noted that this interpretation of the observer effect is controversial within the field
of physics.
8 Electromagnetic theories of consciousness branch off into a ‘Cemi’ version and a
‘quantum mind’ version. The former proposes that digital information from neurons is
integrated to form a conscious electromagnetic information field in the brain (see e.g.
McFadden, 2002), and the latter that electric dipoles of water molecules constitute a
quantum field, referred to as the cortical field, with corticons as the quanta of the field.
Thus, the two alternative theories are linked in a sense.
9 One viewpoint here is that the brain may be viewed as a ‘quantum computer’ and that
all psychological phenomena, including consciousness, can be explained using the pro-
cesses of quantum computing (e.g. Penrose, 1994).
STOCK RETURNS & THE MIND 37

theories allow for the possibility of consciousness affecting matter at a


distance and that machines harvesting quantum technology could be
affected by consciousness from a distance. Resting on such findings,
Roger D. Nelson developed the Global Consciousness Project (GCP)
to investigate if this human–machine interaction also could be true on
a global scale.
The GCP is an international and multidisciplinary collaboration that
generates and collects random number data continuously from a net-
work of physical random number generators at 70 locations around the
world. The random numbers are generated using quantum tunnelling
techniques and the hypothesis underlying the GCP is that events
which elicit widespread emotion or draw the simultaneous attention of
For personal use only -- not for reproduction

large numbers of people may affect the output of the hardware-


generated random numbers in a statistically significant way. The idea
is thus that, if the mind can stretch out beyond our heads and affect
Copyright (c) Imprint Academic

random number generators at a distance, it could be true that the mind


could do so unconsciously and unintentionally and that large
emotional events will thus affect hardware-generated random numbers
in a way that gets ‘picked up’ and made visible in the numbers genera-
ted from it. The GCP has produced remarkable results as the random
numbers seem to be influenced by large global emotional events
(Nelson and Bancel, 2011). Even though the GCP and the data
generated from the project are subject to much debate, one thing is
clear: the events that are claimed to be picked up by the GCP hard-
ware could also affect consumers’ collective preference set and thus
also stock market prices and stock market returns.

5. A Suggestive Link between


Stock Markets and the GCP Data
From the above it could be argued that stock market prices and the
data collected by the GCP should covary since the GCP data, as
claimed, is affected by large global emotional events and since such
events should also affect stock market prices and the returns obtained
from them. Thus, I here seek a ‘non-interaction-based’ statistical
codependence between the data obtained from these seemingly
unrelated sources.
However, the way in which the GCP has conducted their studies
would not answer the research question posted herein. This since the
GCP studies rely on large global emotional events (such as disasters,
war, or political events) to get measurable effects. But as pointed out
38 U. HOLMBERG

by its critics, this has been one of the project’s weaknesses since the
events chosen could be selected in order to fit the data. This criticism
has been addressed by the GCP as they have constructed a predefined
protocol on how to choose the events to be studied. For this study,
however, such criticism is irrelevant as I seek a time series correlation
that does not rely on the importance of single large emotional events.
I chose to study daily stock market returns which thus also requires
the need for a daily aggregate variable from the GCP data. The aggre-
gate GCP data variable derived relies on the the large bulk of publicly
available GCP data (collected every second) and converts it into daily
observations that aims to capture significant deviations from what
should be expected from random numbers. To this end, I use the
For personal use only -- not for reproduction

unfiltered Z-scores calculated every 15 minutes, and derive a daily


observation by taking the 24-hour maximum of these Z-scores. Such a
measure should arguably capture any large shifts in the random
Copyright (c) Imprint Academic

numbers since even subtle unexpected shifts in the data would be


captured in the measure.10
Note that the number of active random number generators tends to
vary over time but that this is of little concern as I utilize the informa-
tion from the Z-scores obtained from the column ‘All Egg Composite’
from the Daily Tables section on the GCP webpage.11 This data
retrieval process results in a time series of maximum daily Z-scores
spanning from the 9th August 2019 all the way back to the 1st January
1999. Note also that all ‘bad data’ are removed.12
The aggregated daily GCP variable is labelled Max[Z] and if the
GCP data are affected by events disturbing the will of the many, it will
(arguably) be captured in the Max[Z] variable. Also, and as discussed
above, such events should affect stock market returns which thus
suggests that stock returns and Max[Z] could be correlated. A statis-
tically significant correlation could also suggest that there is some
underlying quantity determining both stock market returns and the
Max[Z] series as illustrated in Figure 1 (in the figure, the postulated

10 Most GCP studies use XOR filtered data in their studies. Such a filtering procedure will
exclude events that could cause spurious effects (e.g. temperature changes) but, as the
procedure removes some possibly important data, I here use its unfiltered version.
11 Note that the 15-minute intervals underlying the maximum Z-score calculations begin at
00:00:00 UTC. Please visit the GCP website for further details http://noosphere.
princeton.edu/.
12 In particular, all dates with reported bad data are removed. Also, three dates with
unusually high maximum, minimum, and average values are removed.
STOCK RETURNS & THE MIND 39

correlation between the two variables is illustrated by the intersection


between stock market returns and GCP data effects).
For personal use only -- not for reproduction

Figure 1. The GCP data and stock market returns could have a testable
intersection.
Copyright (c) Imprint Academic

But how could such an intersection be formalized? This is as yet


unexplored territory and no known functional form linking Max[Z]
and stock market returns exists. In the absence of a functional form, I
lean on the Taylor theorem (Taylor, 1715) which allows for an
approximation of any functional relationship with a polynomial linear
function.13 Here, it is thus only postulated that Max[Z] could have an
effect on stock market returns in some unknown way, and that
equation (1) can be used as an approximation of this relationship:
k k
ri,t = αi + ΣjΣk(γi,j,kr t – j + βi,j,kMax[Z] t – j) (1)
where rt is stock returns, Max[Z] is the maximum value over 24 hours,
i are different stock market indexes, and where k = (1,2,…K) is the
number of polynomial terms while j is the lag structure under study.
The best fit and order of the polynomial in equation (1) is an empirical
question and treated as such.

6. Results
In this section, the relationship between Max[Z] and stock market
returns (r) is analysed, i.e. the intersection in Figure 1.14 The returns
are obtained from the Dow Jones Global Equity Index as this index is

13 A Taylor series is a series expansion of a function about a point that allows for an
approximation of functional dependence.
14 Returns are defined as rt = (Pt – Pt – 1)/Pt – 1 where Pt is the price/index value at time t.
40 U. HOLMBERG

ideal for capturing global stock market movements and daily changes
in global sentiment. In other words, this index is a good fit for testing
the existence of an intersection as in Figure 1.15
Requiring the data to be balanced, missing data are excluded such
that only dates on which there are both Max[Z] values and some
activity on the Dow Jones Global Equity Index are used. In Table 1
below, descriptive data for the variables used are displayed and, as can
be seen, stock market returns exhibit the ‘usual’ stock market return
properties while Max[Z] is largely skewed and exhibits excess
kurtosis.

Max[Z] Open Close


For personal use only -- not for reproduction

Mean 3.0197 0.0002 0.0002


Median 2.7100 0.0000 0.0004
Maximum 82.1000 0.0665 0.0907
Copyright (c) Imprint Academic

Minimum 1.7600 –0.0861 –0.0713


Std. Dev. 3.9546 0.0098 0.0098
Skewness 15.9442 –0.7297 –0.2778
Kurtosis 273.2198 12.2483 10.7302
Note: The number of observations (N) for all variables is 5236.

Table 1. Descriptive of the data.

As discussed in the previous section, the exact functional form


describing the relationship between stock market returns and Max[Z]
is not known, but a polynomial as in equation (1) can be used to proxy
such a relationship. The return series (r) used in this study are derived
both from daily ‘close’ values (the last reported value of the day,
which is the usual reporting standard within the industry) but also on
daily ‘open’ values (from the first observation every day). This since I
want to allow for the possibility that Max[Z] could predate stock
market returns.
After significant statistical trials, a second- and third-degree
polynomial function seems to fit the data best, dependent on if the
analysis is done on today’s or yesterday’s Max[Z]. The obtained
estimates are derived using Ordinary Least Squares (OLS) and, due to
the possibility of heteroskedastic and/or autocorrelated residuals, the
HAC-Newey-West estimator (Newey and West, 1987) for standard

15 The index measures the performance of stocks that trade globally, targeting 95%
coverage of markets open to foreign investment. It is float market cap weighted and
quoted in USD.
STOCK RETURNS & THE MIND 41

errors is used.16 All significance tests are performed using the t-


statistic together with HAC standard errors and in Table 2 the results
using the full sample stretching back to March 1999 are presented.17
In Table 2, γr is the coefficient describing dependence on previous
stock market returns (the autocorrelation component) and the β:s the
dependence with Max[Z] as in equation (1). As expected, the auto-
correlation component (γr) is highly significant but so is the Max[Z]
dependence for both daily ‘open’ and ‘close’ returns. Also, both return
series are affected by the present date’s Max[Z] and yesterday’s
Max[Z] such that changes in Max[Z] (partly) seem to predate actual
stock market returns. This even after possible additional ‘excluded
variable bias’ is investigated by letting both return series be dependent
For personal use only -- not for reproduction

on an additional autocorrelation component. Apart from the implica-


tion the results have on our understanding of consciousness (it appears
to confirm the validity of the GCP data), this is a profound result in
Copyright (c) Imprint Academic

the face of the ‘no arbitrage’ assumption in economics.


What is meant by the no arbitrage assumption is that economic
systems have an implicitly built-in non-arbitrage property and that all
true arbitrage profits are ‘removed’ as soon as the market becomes
aware of them. This has been observed in the past, and the Monday/
Weekend effect (Cross, 1973) is probably the most famous. This was
a phenomenon in financial markets in which stock returns on Mon-
days were often significantly lower than those of the immediately
preceding Friday. The prevalence of the phenomenon is much debated
as it disappeared when the results where made public (as price
mechanisms corrected) but later re-emerged. Regardless, the results
presented herein are derived from publicly available data and the
estimates can thus be found and derived by anyone. If arbitrage profits
can be made from these results, this quantified effect should be traded
away as soon as the results are made public as they can only exist
because no one yet knows about the effect.

16 Note that if the usual standard errors are used the qualitative nature of the results is
unaffected.
17 Data on the Dow Jones Global Equity Index began during March 1999.
42 U. HOLMBERG

Open on Open on Open on Max[Z]t


Max[Z]t Max[Z]t – 1 and Max[Z]t – 1
α –0.0016** –0.0011*** –0.0027***
γr 0.0781*** 0.0782*** 0.0779***
βMax[Z]t 0.0007*** – 0.0007***
βMax[Z]t2 –2.13E-05** – –2.14E-05**
βMax[Z]t3 1.60E-07** – 1.71E-07**
βMax[Z]t-1 – 0.000463*** 0.000457***
βMax[Z]t-12 – –6.11E-06*** –6.11E-06***
R2 0.73% 0.81% 0.90%
Close on Close on Close on Max[Z]t
Max[Z]t Max[Z]t – 1 and Max[Z]t – 1
α –0.0012** –0.0006 –0.0018***
γr 0.1297*** 0.1297*** 0.1314***
For personal use only -- not for reproduction

βMax[Z]t 0.0005** – 0.0005**


βMax[Z]t2 –1.57E-05* – –1.64E-05**
βMax[Z]t3
Copyright (c) Imprint Academic

1.09E-07 – 1.23E-07*
βMax[Z]t-1 – 0.0003** 0.0003*
βMax[Z]t-12 – –4.31E-06** –3.83E-06*
R2 1.79% 1.78% 1.91%
Significance levels: * 10%, ** 5%, and *** 1%.
Note: The sample size (N) is 5237.

Table 2. Estimates for rt.

Returning to the results in Table 2, it is noted that the predating effect


obtained can also be found from the significant results from the ‘open’
series on the current date’s Max[Z]. This as the open return series is
obtained directly when stock markets open in New York, while the
Max[Z] result is obtained several hours later during the day and first
when the 24h time period has passed.18 Furthermore, looking at the
coefficient of determination (the R2 value) it is found that about 1%
for daily open values and just below 2% of daily close values can be
explained using the specifications in Table 2. Not much, but well in
line with previous research with regards to the autocorrelated nature of
stock market returns. However, and more importantly given the
research hypothesis, the R2 value increases between 0.2% and 0.3%
after the inclusion of the Max[Z] dependence. This result is obtained
by comparing the coefficient of determination with and without the
polynomial structure describing the Max[Z]t and Max[Z]t – 1 depend-

18 The 15-minute intervals, from which the daily maximum Z value is calculated (the
Max[Z] variable), begins at 00:00:00 UTC and ends 24 hours thereafter.
STOCK RETURNS & THE MIND 43

ence but, as most variance still remains unexplained, the results open
up an obvious avenue for future research. Note also that the significant
dependence on Max[Z] is unaffected by the inclusion of the lagged
returns and that this variable is only included for the residuals to be
well behaved (and in order to reduce missing variable bias).
The results presented in Table 2 also carry with them some clues
regarding the shape between Max[Z] and stock market returns. This
since it is found that the polynomial in equation (1) has a positive first
term and a negative second term, a result that holds true for both the
Max[Z]t dependence as well as for Max[Z]t – 1. Also, if the current
date’s Max[Z] dependence is studied, both on a stand-alone basis and
in conjunction with Max[Z]t – 1, a third and positive polynomial term is
For personal use only -- not for reproduction

found. This can be seen from the first and last column in Table 2. The
results also suggest that small to moderate values of Max[Z] coincide
with positive returns, which is a result that corresponds nicely to the
Copyright (c) Imprint Academic

small but positively observed trend growth in stock market returns


(the first row for the second and third columns in Table 1). But, if
Max[Z] increases to large enough values, there is a negative effect on
stock market returns, possibly due to large negative global emotional
events that also have a large impact on the GCP data. As Max[Z]
grows even larger, the effect on stock market returns turns positive
again and the results also suggest that this third polynomial term effect
rapidly loses its ability to affect stock market returns as no significant
term is found on yesterday’s Max[Z].
But, as shown in the descriptive data in Table 1, the Max[Z] variable
is both skewed and exhibits large kurtosis which could result in non-
normal residuals if the parameters are obtained through OLS. How-
ever, it is noted that OLS estimates are still a reasonable estimator in
the face of non-normal errors. In particular, the Gauss-Markov
Theorem states that the ordinary least squares estimate is the best
linear unbiased estimator of the regression coefficients as long as the
errors have zero mean, are uncorrelated, and have constant variance.
This seems to be the case for the estimates underlying the results in
Table 2.19 Also, heteroskedasticity and autocorrelation consistent
standard errors (HAC) have been used when testing for significance
such that the results could be considered robust to such issues.
However, as ‘outliers’ could drive the qualitative nature of the
results, only dates on which Max[Z] < 10 are considered in a separate

19 Best meaning optimal in terms of minimizing mean squared error.


44 U. HOLMBERG

analysis.20 The statistical results excluding such large values are pre-
sented in Table 3 and, as can be seen, significant correlations remain.
The ‘open’ return series maintains its dependence on Max[Z]t while its
dependence on Max[Z]t – 1 shifts and becomes linear. Also, a combined
specification suggests such dependence on the ‘open’ series, while for
the ‘close’ series the predating dependence vanishes. These results
thus suggest that large Max[Z] values could, in part, be the cause of
the non-linearity and also that the results support the qualitative con-
clusion that a statistical dependence between these variables exists.

Open on Open on Open on Max[Z]t


Max[Z]t Max[Z]t – 1 and Max[Z]t – 1
α –0.00126** –0.0013 –0.0146**
For personal use only -- not for reproduction

γr 0.0784*** 0.0780*** 0.0780***


βMax[Z]t 0.0096** – 0.0099**
Copyright (c) Imprint Academic

βMax[Z]t2 –0.0021** – –0.0022**


βMax[Z]t3 0.0001** – 0.0001**
βMax[Z]t-1 – 0.0005** 0.0005*
βMax[Z]t-12 – – –
R2 0.77% 0.67% 0.84%
Close on Close on Close on Max[Z]t
Max[Z]t Max[Z]t – 1 and Max[Z]t – 1
α –0.0013* – –
γr 0.1321*** – –
βMax[Z]t 0.0005* – –
βMax[Z]t2 – – –
βMax[Z]t3 – – –
βMax[Z]t-1 – – –
βMax[Z]t-12 – – –
R2 1.82% – –
Significance levels: * 10%, ** 5%, and *** 1%.
Note: The sample size (N) is 5207.

Table 3. Estimates for rt given that Max[Z] < 10.

Summing up, the results in Table 3 confirm the qualitative nature of


the results in Table 2 as they suggest that the estimates are not driven
by outliers and that they are statistically sound. Also, both tables
suggest that an increase in Max[Z] in general is affiliated with
increased returns, but that larger Max[Z] values signal negative returns

20 A Max[Z] value of 10 is highly unlikely and thus a good proxy for extreme value and
outlier detection.
STOCK RETURNS & THE MIND 45

to come. It is also found that large changes in Max[Z] predate stock


price movements and it is noted that this last point could have obvious
market implications. In addition to the above, these results seem to
confirm the validity of many GCP studies and suggest that random
numbers, at least if generated using quantum tunnelling techniques,
can be influenced by intention and global emotions at a distance.
The results presented so far herein suggest that the intersection
argued for in Figure 1 is likely to exist. But perhaps the results are due
to some spurious and causal relationship or due to chance? The former
could be true, as the intersection illustrated in Figure 1 also suggests
that an underlying larger quantity could determine both stock market
returns and the Max[Z] series. In fact, as the term spurious correlation
For personal use only -- not for reproduction

is generally understood in statistics, it describes a non-causal correla-


tion that can be spuriously created by an antecedent which (in this
case) causes both Max[Z] and r. Thus, this is of little concern as it
Copyright (c) Imprint Academic

could be argued that such an underlying variable could exist and


should be explored in future research. The second point could, how-
ever, be a source of concern, even though the likelihood of the results
being due to chance is well below 1/100 based on the significance
tests made.21
This latter concern is adequately addressed by applying the same
statistical techniques on a wide range of well-known equity indexes all
around the globe. The dependence structure follows equation (1) and
is inspired by the obtained dependence in Table 2 but modified in the
spirit of parsimony. Returns are calculated using the indexes’ ‘close’
value and, as can be seen in Table 4, significant correlations are again
found with regards to the relationship between daily stock (r) returns
and Max[Z]. In fact, almost all stock market return series (11 of the 12
studied) are found to be significantly correlated with Max[Z], which
supports the results presented in Table 2. Notably, the ‘timing’ of the
variable’s dependence is found to differ, which possibly could be
attributed to the index’s geographical location and thus reflects the
difference in UTC closing times. This expiation is supported by the
findings as different geographical ‘blocks’ are subject to the same
dependence structure at large. The pure American return indexes
(S&P, Dow Jones, Nasdaq, and Ibovespa) correlate with the current
day’s Max[Z] while the European stock market indexes and most

21 The tables lowest P-value related to Max[Z] is for ‘open’ returns on Max[Z]t –1. Here, the
P-value is about 0.0003, i.e. a one in 3,333.333 result if due to chance.
46 U. HOLMBERG

Asian indexes correlate with Max[Z]t – 1.22 The Japanese Nikkei 225 is
an exception but, taken together, these dependences hint towards the
importance of time.

USA, USA, Dow USA, Nasdaq Brazil,


S&P 500 Jones Ibovespa
α –0.0003 –0.0003 –0.0003 –0.0018
γ –0.0707*** –0.0663*** –0.0329* –0.0103
Americas

βMax[Z]t 0.0002 0.0002 0.0003 0.0009**


βMax[Z]t2 –3.60E-06* –3.73E-06* –4.23E-06* –3.40E-05**
βMax[Z]t3 – – – 2.81E-07**
R2 0.55% 0.52% 0.14% 0.01%
UK, France, Germany, Switzerland,
For personal use only -- not for reproduction

FTSE 100 CAC 40 CDAX SIX


α –0.0011** –0.0009 –0.0012* –0.0011***
γ –0.0360** –0.0272 0.0013 0.0344*
Europe
Copyright (c) Imprint Academic

βMax[Z]t-1 0.0005*** 0.0004* 0.0005** 0.0005***


βMax[Z]t-12 –6.22E-06*** –6.05E-06** –6.84E-06** –6.51E-06***
R2 0.26% 0.16% 0.12% 0.28%
Singapore, Hong Kong, China, Japan,
SGX Hang Seng Shanghai Nikkei 225 +
α –0.0004 –0.0007 0.0005 –0.000706
γ 0.0395** –0.0108 0.0137 –0.029901
Asia

βMax[Z]t-1 0.0002 0.0004* –4.65E-05 0.000309**


βMax[Z]t-12 –3.47E-06* –5.24E-06** –3.36E-07 –3.27E-06*
R2 0.21% 0.70% 0.05% 0.16%
Significance levels: * 10%, ** 5%, and *** 1%.
Note that + indicates that the regression was done on non-lagged values of
Max[Z]. Note also that the sample size (N) per regression is 5236.

Table 4. Estimates for rt on alternative indexes.

7. Conclusions and Future Research


In this paper, it was shown that the random numbers generated by the
GCP covary with global stock market returns. This topic was investi-
gated after it was claimed that such a dependence should exist since
stock market returns are the result of investors’ collective decisions of
the worth of firms. As such a collective decision should follow the
usual market pricing mechanisms in economics, it was argued that
stock market prices should be affected by events similar to events that
are claimed to affect the GCP data. In an empirical part of this paper,

22 Max[Z] is calculated using a fixed 24h rolling window.


STOCK RETURNS & THE MIND 47

such a dependence was found to exist for several different world stock
market indexes.
The relationship was found using a newly defined variable, derived
from the GCP data. This new variable was constructed in a way that
captures large changes in the GCP data over the past 24 hours. This
variable was labelled Max[Z] and was found to be correlated with
global stock market returns, as well as various local well-known stock
market indexes. It was also shown that the relationship between
Max[Z] and stock market returns can be approximated with a second-
and third-degree polynomial and that the polynomial dependence is
dependent on whether today’s Max[Z] values are used or if the
Max[Z]t – 1 is studied. The results thus suggest that yesterday’s GCP
For personal use only -- not for reproduction

data can be used to describe today’s stock market movements, which


is a profound result as it could open up arbitrage profit opportunities.
How the market will react to this new information will thus be
Copyright (c) Imprint Academic

interesting to follow.
The results also validate some of the claims made by the GCP, and
since the GCP data rest on the assumption that consciousness stretches
out beyond our heads and can affect hardware-generated random
numbers at a distance, the current paradigm with regards to conscious-
ness needs to be discussed. The results actually invite a discussion on
alternative theories of consciousness as the results presented herein
cannot be understood using the prevailing scientific understanding of
consciousness alone. Perhaps the results are better understood through
the lens of some alternative theories as some of these theories allow
for the possibility of consciousness stretching out beyond our heads.
Furthermore, since the random numbers collected by the GCP use
quantum tunnelling techniques to obtain the random numbers,
quantum brain dynamics theories or electromagnetic theories of con-
sciousness are obvious candidates in a pursuit of a deeper under-
standing of the results. Here, this is left as an interesting avenue for
future research.
On a less grandiose level, a better formalization of the functional
form linking large random numbers and stock market returns could be
studied and better daily GCP data variables could be constructed and
tailored to fit individual markets. Also, a more thorough study of the
relationship between different indexes could be done and such a study
should also adjust for time differences in reporting and possible
exchange rate effects. Furthermore, pure market sentiment and market
volatility could be related to the Max[Z] variable and could be studied,
and so could any relationships found between different markets. If
48 U. HOLMBERG

these dependences can be found, a latent and underlying hidden factor


could possibly be obtained and, if this were the case, research that
seeks to explain such a latent hidden variable could be conducted. The
importance of time also adds to future research dimensions and it
would be interesting to see if financial futures markets are affected by
Max[Z] and, if so, at which point in time. Finally, various macro-
economic variables could be constructed and possibly correlated with
variables aggregated out of the GCP data. In short, these novel
findings open multiple avenues for future research.

References
Atmanspacher, H. (2015) Quantum theory and consciousness: An overview with
For personal use only -- not for reproduction

selected examples, Discrete Dynamics in Nature and Society, 1, pp. 51–73.


Buks, E., Schuster, R., Heiblum, M., Mahalu, D. & Umansky, V. (1998)
Dephasing in electron interference by a ‘which-path’ detector, Nature, 391
Copyright (c) Imprint Academic

(6670), pp. 871–874.


Chalmers, D.J. (1995) Facing up to the problem of consciousness, Journal of
Consciousness Studies, 2 (3), pp. 200–219.
Chalmers, D.J. (2003) Consciousness and its place in nature, in Stich, S. &
Warfield, F. (eds.) Blackwell Guide to the Philosophy of Mind, Malden, MA:
Blackwell.
Cotterill, R. (2001) Cooperation of the basal ganglia, cerebellum, sensory
cerebrum and hippocampus: Possible implications for cognition, consciousness,
intelligence and creativity, Progress in Neurobiology, 64 (1), pp. 1–33.
Cross, F. (1973) The behavior of stock prices on Fridays and Mondays, Financial
Analyst Journal, 4, pp. 67–69.
Daszyńska-Żygadło, K., Szpulak, A. & Szyszka, A. (2014) Investor sentiment,
optimism and excess stock market returns: Evidence from emerging markets,
Business and Economic Horizons, 10, pp. 362–373.
Dunne, B. & Jahn, R. (2007) Experiments in remote human/machine interaction,
Journal of Science and Healing, 3, pp. 272–272.
Durkheim, E. (1893/1997) The Division of Labour in Society, Halls, W.D. (trans.),
Coser, L.A. (intro.), New York: Free Press.
Fisher, K.L. & Statman, M. (2003) Consumer confidence and stock returns, The
Journal of Portfolio Management, 30 (1), pp. 115–127.
Güzeldere, G. (1997) The Nature of Consciousness: Philosophical Debates, pp. 1–
67, Cambridge, MA: MIT Press.
Hoffman, D. (2008) Conscious realism and the mind–body problem, Mind and
Matter, 6 (1), pp. 87–121.
Hoffman, D. (2014) The origin of time in conscious agents, Cosmology, 18, pp.
494–520.
Kastrup, B. (2018) The universe in consciousness, Journal of Consciousness
Studies, 25 (5–6), pp. 125–155.
Keynes, J.M. (1936) The General Theory of Employment, Interest and Money,
London: Macmillan.
Koch, C. (2004) The Quest for Consciousness, Englewood, CO: Roberts &
Company.
STOCK RETURNS & THE MIND 49

Lansing, K.J. & Tubbs, M. (2018) Using sentiment and momentum to predict
stock returns, FRBSF Economic Letter, Federal Reserve Bank of San Francisco.
Llinás, R. (2002) I of the Vortex: From Neurons to Self, Cambridge, MA: MIT
Press.
Mas-Colell, A., Whinston, M. & Green, J. (1995) Microeconomic Theory, Oxford:
Oxford University Press.
McFadden, J. (2002) The conscious electromagnetic information (cemi) field
theory: The hard problem made easy?, Journal of Consciousness Studies, 9 (8),
pp. 45–60.
Nelson, R., Jahn, R. & Dunne, B. (1986) Operator-related anomalies in physical
systems and information processes, Journal of the Society for Psychical
Research, 53 (803), pp. 261–285.
Nelson, R. & Bancel, P. (2011) Effects of mass consciousness in random data
during global events, Explore, 7, pp. 373–383.
Newey, W.K. & West, K.D. (1987) A simple, positive semi-definite, hetero-
For personal use only -- not for reproduction

skedasticity and autocorrelation consistent covariance matrix, Econometrica, 55


(3), pp. 703–708.
Penrose, R. (1994) Physics and the mind, in Longair, M. (ed.) The Large, the
Copyright (c) Imprint Academic

Small and the Human Mind, pp. 93–143, Cambridge: Cambridge University
Press.
Pockett, S. (2012) The electromagnetic field theory of consciousness: A testable
hypothesis about the characteristics of conscious as opposed to non-conscious
fields, Journal of Consciousness Studies, 19 (11–12), pp. 191–223.
Radin, D., Nelson, R., Dobins, Y. & Houtkooper, J. (2006) Assessing the evidence
for mind–matter interaction effects, Journal of Scientific Exploration, 20 (3), pp.
361–374.
Sharpe, W.F. (1978) Investments, Upper Saddle, River, NJ: Prentice-Hall.
Shiller, R.J. (2017) Narrative economics, American Economic Review, 107 (4), pp.
967–1004.
Smith, A. (1776) An Inquiry into the Nature and Causes of the Wealth of Nations,
Book IV, Chapter II, paragraph IX, London: W. Strahan & T. Cadell.
Taylor, B. (1715) Methodus Incrementorum Directa et Inversa [Direct and
Reverse Methods of Incrementation], in Latin, pp. 21–23, Prop. VII, Thm. 3,
Cor. 2, translated into English by Bruce, I. (trans.), London: William Inny.
Van den Noort, M., Lim, S. & Bosch, P. (2016) On the need to unify neuroscience
and physics, Neuroimmunology and Neuroinflammation, 3, pp. 271–273.
Wang, L., Ahn, K., Kim, C. & Ha, C. (2018) Agent-based models in financial
market studies, Journal of Physics: Conference Series, [Online], doi:
10.1088/1742-6596/1039/1/012022.

Paper received May 2019; revised December 2019.

You might also like