The SP500 Empathetic Machine: a view of what might lie ahead for those that want to build a position in Equities

1. Page 1
SP500 Quantamental Insights
Artificial Intelligence at the Service of Asset Managers and Decision Makers
By EyeHigh
25th Sept 2020
SP500 and the Empathetic Machine
What can we say about the
underlying trend of the stock
market given today’s
insurmountable degree of
uncertainty? To get a grasp of
what might lie ahead lets first
refer to the Manufacturing ISM
(Chart 1; business expectations
survey).
Before that, let us briefly review a
few thoughts on the current
macroeconomic environment. In
our view, global retail sales have
clearly been impacted by a
positive shock. This has mainly
resulted from the lockdowns
earlier this year that shifted
demand from hard-to-reach
services to tech-goods. Global
industrial production has
recorded unsustainably high
growth rates, a result partly
anticipated by the ISM. The
obvious risk is that this survey is
tackling the future with an
extrapolated optimism (Charts
1&2) that is likely about to roll
over.
So, does the level of the ISM pose
a threat to equities? Or are the
recent market corrections already
a reflection of what is to come in
terms of the ISM? In the latter
case we are too late. To ascertain
these issues, we project two
hypothetical paths for the ISM
and Fed rates (Ch 2). Both suggest
turn in business expectations,
with Path 1 representing the most
benign scenario.
Empathetic Machine
Algorithms provide us with
monthly averages. These
averages represent a bias
which makes them difficult
to use for market timing. In
addition, our basic model
relies on simulated paths for
only 2 variables-paths that
are clearly disputable.
Despite these limitations,
the resulting market reaction
function is valuable, as is it
based on common sense
aimed at pointing in the right
direction on average.
Anticipating a move like the
one experienced in Feb20 is
almost a near-impossible
task for a model of this sort.
No algorithm is equipped an
output with such a deviation
from average forecasts.
By contrast, it is much easier
to assess market reactions
AFTER a shock has taken
place; the past exhibits
plenty of recoveries that
share the same patterns.
These patterns result from
investors repetitively
pondering 2 issues: 1) The
expected macro landscape
(Are valuations justified?)
2) The greed/fear balance
coupled with the market
tendency to overreact and
mean revert.
THIS MAY CORRESPOND TO
A DOUBLE BOTTOM IN A
MILDLY UPWARD TREND

2. Page 2
SP500 Quantamental Insights
Artificial Intelligence at the Service of Asset Managers and Decision Makers
By EyeHigh
SP500 and the Ai Empathetic Machine
Given our time horizon we assume
no change in rates. We suspect
that the loss of growth momentum
(lockdowns) and the fiscal
stimulus/elections will be central
for the ISM survey.
The election is an exogenous
variable. The result is unknown as
it is the fiscal program that will be
approved by the winning party.
Our goal is to uncover the
probability of a significant macro
downturn in spite of these
limitations. The algorithm will
provide average “biases” (not
exact forecasts) which we will
interpret.
These are the main conclusions of
our two algorithms:
1) If the ISM were currently rolling
over, algo A & B both suggest that
we are immersed in a 1-2 month
downside risk window.
2) Strikingly, the most severe ISM
correction (Path 2) barely changes
that conclusion. We suspect that
this is so because in both cases the
ISM gets back above 50 in a
reasonably short time span (Ch 2).
3) This market hiccup would
translate to double bottom within
a mildly positive market trend
(Charts 3 and 4).
4) The main difference between of
the two paths lies in the period
dic20 to feb21 (when path A
reaches 60; slightly higher
returns).
5) The time to buy is within the
next few weeks.
Empathetic Machine
So, if recoveries share a lot
more common features
then, from a machine
learning perspective, they
are a convenient starting
point for analysis. Now,
three weeks into a
correction, might well be
the right time to pose these
questions.
For the sake of accuracy
and timing we would need
to drill down to a weekly
data frequency. The idea
would be to feed “highly
emotional” data into the
machine and still get
reasonable readings of the
underlying strength of the
market.
We fed the algorithm with
VIX, Skew, Put/Call, and
similar emotional data
together with a low load of
macro fundamentals. We
targeted cycles of 8 weeks
into the future and allow
the machine some freedom
(not much) to choose a
suitable mix to respond.
Our empathetic machine
“feels” (Chart 5) that the
fear/greed balance will be
tilted towards fear for
another 2-3 weeks. This,
then, is the opportunity
window if we are to build a
long position in equities.
Chart 5: Empathetic Machine Underlying Strength

3. DISCLAIMER:
This document does not constitute an offer or a solicitation to engage in any trading strategy or to purchase or sell any financial instruments. Given its
general nature, the information included in this document does not contain all the elements that may be relevant for a recipient to make an informed
decision in relation to any strategies or financial instruments discussed herein. The information and opinions contained in this document have been
obtained from sources believed to be reliable, but we make no representation, express or implied, that such information and opinions are accurate or
complete. Other financial institutions or persons may have different opinions or draw different conclusions from the same facts or ideas analyzed in
this document. Any scenarios, assumptions, historical or simulated performances, indicative prices or examples of potential transactions or returns are
included for illustrative purposes only, and we give no assurance that any favorable scenarios described are likely to happen, nor that it is possible to
trade on the terms described herein or that any of the potential returns illustrated can be achieved.
We have used historical information in order to provide an illustration of how certain parameters may have performed over a defined period. This
document also contains certain performance data based on back-testing, i.e., calculations of the hypothetical performance of a strategy, index or asset
as if it had actually existed during a defined period of time and may in certain circumstances contain simulated performance information where the
index or asset described has recently been established or issued.
The scenarios, simulations, development expectations and forecasts contained in this document are for illustrative purposes only. This type of
information has inherent limitations which you must consider carefully. While the information has been prepared in good faith in accordance with our
own internal models and other relevant sources, an analysis based on different models or assumptions may yield different results. Therefore, this
analysis may vary significantly from an analysis obtained from other sources or market participants. Further, please note that such analysis is based on
a number of working assumptions that may not be capable of duplication in actual trading terms. Unlike actual performance records, hypothetical or
simulated performances, returns or scenarios may not necessarily reflect certain market factors such as liquidity constraints, fees and transaction costs.
Actual historical or backtested past performance does not constitute an indication of future results.
In providing this document, we give no financial, legal, tax or any other type of advice to recipients. Certain strategies and/or potential transactions
discussed in this document involve the use of derivatives, which may be complex in nature and may give rise to substantial risk, including the risk of
partial or total loss of any investment. We make no representation as to whether any of the strategies discussed herein may be suitable for investors’
financial needs, circumstances or requirements. Investors must make their own assessment of the strategies and/or potential transactions, using such
professional advisors as they make require. We accept no liability for any direct or consequential losses arising from any action taken in connection
with the information contained in this document.