You are here:Forum » Cause Effect and Dr Kaplan's Risk Matrix

# Cause Effect and Dr Kaplan's Risk Matrix

Posted on 22 January 2012

PROBABILITY CLOUDS, RISK REWARD, MORENO RISK MATRIX AND PERFORMANCE MEASUREMENT

Andrew Moreno

Phil Jones wrote:

> cause-effect

I was thinking over your statement regarding Dr Kaplans Risk Matrix, I had some thoughts,

CAUSE EFFECT PRESUPPOSITIONS

In NLP as you know, there's the idea of cause effect presuppositions.

This is that X causes y and then a statement can be made that

presupposes this either implicitly or explicitly.

PHYSICS MODELS OF PROBABILITY CLOUDS

In physics in the photoelectric effect, if you shoot a photon

at an atom, according to the cause effect presupposition description,

the photon would cause an electron to jump energy levels, thus if

aggregated, this would create an electric current.

However the physics models don't allow for an electron to be in

a certain place around the nucleus for the photon or groups of

photons to hit the electron in a cause effect presupposition manner

for the electron to shift energy levels.

The physics models only at this stage allow for the electron to

be in a certain orbit or energy state around the nucleus in a balance

of probabilities. The electron orbits in the physics models are

described as clouds of probabilities where the probability is high

that they will be at a certain place in the cloud. The premise that

they will be in a certain place at a certain time isn't certain,

according to the physics models.

So according to your above statement, if you have an organisation, there

could be cause effect statements that describe that certain things

cause risks but it isn't certain that that certain causes caused the

risks.

BALANCE OF PROBABILITIES

More precisely would be that certain things, according to a balance

of probabilities, cause risks and this could be described in a

cause effect presupposed sentence. However for a specific event

that tries to specifically quantify that x created y risk then

that wouldn't be possible according to an analogy to the physics

models.

In the Supreme Court guide that I have, it mentions that in the

Rules of Evidence there is a standard of proof based on a "balance

of probabilities". In a civil lawsuit there might be imperfect

information or information that is unquantifiable. This overlaps

with the field of study of epistemology, which NLP is partly based

on.

EBS - UNDERSTANDING RISK VERSUS MANAGING RISK

In the Heriot Watt EBS module on Strategic Risk Management they

outline that understanding risk is different from managing risk.

MORENO RISK MATRIX FOR UNDERSTANDING/MANAGING RISK

So to build a two dimensional matrix similar to Dr Kaplan's in

his position paper,

type | randomness | balance of probabilities

type

understanding

risk | monte carlo | probability clouds, pattern matching

managing | risk management systems | discretionary decision making

risk | pricing copulas

Some hedge fund managers manage risk using internal decision strategies

or discretionary techniques that are based on probabilities and pattern

matching. I recently saw a piece of software in Equis corp's catalogue

that uses probability calculus for example.

However I saw on the medical TV show House the MD's debating on whether

there existed cognitive pattern matching. They concluded that it didnt

exist. However computers can do that using image recognition mathematics.

I discussed image recognition with Industry Canada a few years ago.

Some mathematicians try to understand risk using random monte carlo

simulations. My understanding is that these mathematicians usually aren't

as strong in discretionary trading however I may be mistaken as I don't

have enough data to support this.

I happened to meet a former professional chess player who traded options

and he mentioned he knew the mathematics it was based on. Dr Schiro

mentioned that in calculus, it's useful to use educated

guesses to derive solutions. However this approach uses probabilities

and not monte carlo simulations. I'm not sure if it's possible to

play chess or solve calculus equations using monte carlo simulations.

In the financial crisis, there was/is a disconnect in that many

hedge fund managers relied on risk management systems to trade when

they should have combined using discretionary techniques. One example

is the use of trading systems based on pricing copulas that broke down

during the financial crisis.

[According to Institutional Investor magazine in 2011 many hedge funds

are below their high water mark, which means they won't be earning

performance fees for this year, this shows correlation which sort of

corresponds to the cause effect presuppositions - in this case it might

be that many hedge funds invested by momentum investing, which means

in a sideways or bear market their performance decreases.]

On the opposite note, some hedge fund managers used purely discretionary

techniques to trade. Case in point is the recent failure of a certain

investment bank within the last 2 months that failed even though Basel

2 or 3 was in effect. The firm took on too much risk on a sovereign debt

trade that was discretionary.

Dr Kaplan wrote in his "dimensions paper" that one of the ways to

deal with unknown risks is through hedging which can be a discretionary

or non discretionary endeavour.

THE HUMAN ELEMENT TO RISK/REWARD

So, risk/reward management systems should probably have some element of

human based real time discretion in managing risk. Dr Kaplan, in his

letter to the FT, wrote that the rebalancing accounting used by

financial engineers in calculating the risk reward profile of

investment positions should be taught in business schools.

Maybe this could lead to new job descriptions as computers are

often seen as tools to reduce the level and numbers of human input

into decision making, which is probably a mistaken proposition.

RISK/REWARD QUANTIFICATION IN IMPERFECT SITUATIONS

However the constraint is how to quantify the discretionary

risk/reward management decisions of employees in unquantifiable

situations where there is imperfect information or the results are

not apparent, either immediately or in the future.

One way that was used in the past is to use track records and

control charts. However there are constraints to this approach.

There are probably other methods.

Another way is social proof, which is used on the Internet.

Humans have a sense of what creates risk and reward, at least

that's the sense I have of the people I have met and seen.

But there is a lag time.

I was discussing with a friend that certain fables have a quality

to them. The opposite of risk management is quality I think.

In control charts, computers probably couldn't quantify results

properly. There is a new push for search engines to search the deep

web, not just metadata and web pages.

SHIFT FROM FINANCIAL REPORTING TO FINANCIAL DECISIONS

In the Heriot Watt EBS module on Financial Risk Management

it stated that at one time financial reporting was valued highest

however that had shifted to financial decision making. Whether

that will be the same during/after the financial crisis remains

to be seen.

So maybe there will be a shift back to financial reporting or at

least hybrid methods. Dr Schiro mentioned that some financial analysts

in his former firm could "eyeball" financial statements to find

the relationships between numbers. This is a form of discretionary

decision making. I can do this also using financial statements of

my trading records. I adjusted my trading according to the results

I saw.

According to Dr Bandler if one person can have an ability, it can be

taught and replicated. That's the hope at least.

One of the things I developed was a way to target synapse generation

in certain areas of the brain. For instance, for the ability to eyeball

charts, it's possible to develop synapses in the visual cortex, the back

of the head, so that any cognition can be improved and any deficits

of synapses can be remedied.

One of the reasons I wrote this letter is because I have a model of

self diagnostics, which I think is a key part to risk/reward. Self

diagnostics is invaluable in medical situations. My doctor mentioned

that many doctors can't self diagnose. This model can provide insight

in unquantifiable situations and uses a balance of probabilities

approach also. I hope to write on it soon.

IMPLICATIONS FOR PERFORMANCE MEASUREMENT

Risk management models development will help to reduce systemic

risk and provide a structure for new job categories so that all areas of

the organization and society can have an impetus towards risk/reward

management, as opposed to purely operational job categories which

are on the decline due to development of intelligent computers.

Humans have an innate sense of risk reward and this can be leveraged to

improve organisational performance and give people a sense of

courage to push into new territory.

Regards,

Andrew Moreno