Realising Virtual Trading:

yardbellAI and Robotics

Nov 14, 2013 (3 years and 6 months ago)

61 views

Realising Virtual Trading:

What Price Virtual Reality?


Soha Maad
a
, Meurig Beynon
a
, Samir Garbaya
b


a
Department of Computer Science, University of Warwick,Coventry CV4 7AL, UK

e
-
mail: soha@dcs.warwick.ac.uk, wmb@dcs.warwick.ac.uk,

b
Laboratoire de Robotique de Paris, 10
-
12 Avenue de l'Europe, 78140 velizy, France

e
-
mail: garbaya@robot.uvsq.fr


OUTLINE


Motivation/issues


Our aim


The case study


textual/2D/VR simulation


performance metrics


Empirical Modelling (EM) & Virtual Reality (VR)


VR for different context


Conclusion


Future research

MOTIVATION/ISSUES (1)

Old Trading Model

New Trading Model

Financial Markets brings

buyers and sellers in one

physical place

Online trading

Electronic Communication Networks

Screen based trading

New trading systems

Extension of trading duration

Opening of new trading channels

Shift

Competition pressure

Computerization is about to overtake markets that traditionally

depended on physical presence to bring buyers and sellers


together in one place

MOTIVATION/ISSUES (2)

Providing appropriate interfaces

for trading environments is a

challenging task

?

MOTIVATION / ISSUES (3)

broker

investor

dealer

financial

consultant

Trading system

The behavior of different trading parties constitutes

a complex environment which is difficult to capture

in a computer model, a computer simulation,


or a textual description

Trading signals

MOTIVATION/ISSUES (4)

Computer aided support to research in the area of


financial market microstructure



Market

microstructure

is

the

academic

name

of

the

branch

of

financial

economics

that

investigates

trading

and

the

organization

of

securities

markets
.

This

important

field

of

study

has

grown

substantially

since

the

Stock

Market

Crash

of

1987
.

Market

microstructure

analyses

security

trading

and

pricing

at

the

institutional/market

level
.

It

focuses

on

modelling

the

influence

of

transactions

costs

on

security

pricing
.

With

a

wider

view

of

the

financial

market

structure,

this

field

of

study

attempts

to

improve

on

the

black

box

modelling

of

equilibrium

asset

pricing,

by

accounting

for

strategic

behaviour

among

trading

parties”
.




L
.

Harris,

Trading

and

Exchanges,

Draft

textbook
:

December

4
,

1998
,

University

of

Southern

California,

Marshall

School

of

Business
.


MOTIVATION / ISSUES (5)

Earlier research in

financial market microstructure


Computer simulations (Harris, 1998)


Trading games (Nasdaq Head Trader, etc..)


3D Virtual Market Simulations (NYSE market Trac)


Workflow modelling of the impact of electronic trading on market efficiency
(Konana et al, 2000)


Financial research:


analysing

the

behaviour

and

profitability

of

online

investors

(Oedan

et

al,

2000
)
;



analysing

the

impact

of

the

introduction

of

electronic

brokers

on

the

structure

of

foreign

exchange

market

(FX)

(Rime

et

al,

2000
)



comparing

floor

trading

in

NYSE

to

over

the

counter

trading

in

NASDAQ

with

reference

to

best

execution

prices

(Jennings

et

al,

2000
)



analysing

the

information

content

of

the

ambient

noise

level

in

the

Chicago

Board

of

Trade's

30
-
year

Treasury

Bond

futures

trading

pit

(Coval

et

al,

2000
)

OUR AIM


Proposing new principles for the
development of environments for
virtual trading to deliver VR using a
new approach to computer
-
based
modelling (Empirical Modelling)


Modelling interaction in a trading
environment while combining real
-
world knowledge with real
-
time
interpretation of abstract numerical
data and indicators.


Opening up VR technology to
embrace complex social environments

Image capturing
device/tool

Geometric abstraction/
computer aretefact

Real physical
objects

Construed mental
model

converter

VR environment

VR system

Financial
data
repository

Networking
architecture

Performance
evaluator

User immersion
and

experience


EM

VR MERGE

Case Study

The Monopoly Dealer Simulation


http://lharris.usc.edu/trading/DealerGame
\
Default.htm


The

program

simulates

trading

in

a

dealer

market

in

which

there

is

only

one

dealer

(the

user

of

the

simulation

model)
.



The

user’s

task

(the

sole

dealer)

is

to

set

and

adjust

bid

and

ask

quotes

(raise,

lower

quotes,

or

narrow

and

widen

the

spread)

to

maximize

his

trading

profits
.



The

computer

model

simulates

traders

arriving

at

random

times

to

trade

with

the

dealer

(user)

at

his

quoted

prices
.



The

aim

of

the

simulation

is

to

raise

the

awareness

of

its

user

to

the

trading

behavior

of

different

types

of

investors

(informed/uniformed),

and

the

true

value

of

the

security

.


The

role

of

the

user

(dealer)

is

to

estimate

the

true

security

value

by

examining

the

order

flow
.



While

the

simulation

is

running,

the

computer

estimates

the

user

(dealer)

position
.


When

quitting

the

simulation,

the

computer

shows

the

true

security

value

and

the

true

profits

of

the

dealer

(user)
.



The

dealer

(user)

should

know

how

to

attract

traders

by

adjusting

his

bid/ask

quotes

and

spread
.



The

simulation

will

end

if

the

dealer’s

inventory

goes

above

10
,
000

or

below

-
10
,
000
.



Evolution in State Visibility

Textual

2D

3D/VR



Caputring the Mental Model

in abstract textual and

numeric description



Keyboard press for user action



Tabulated format for visualization

of numeric data






Capturing mental model in

abstract 2D geometric description



Event triggered user action

(mouse click, textual input)



2D and tabulated format for

visualization of numeric data



navigation using scroll bar





user action visibility

& state visibility


Abstract Mental
Model in text

user action visibility

& state exploration




Combining knowledge of real
physical object with abstract 2D
and 3D visualization



Event triggered user action

(mouse click, textual input)



User walkthrough and flying
for state exploration




immersive 3D visualization of
numeric data



navigation using scroll bar





Case Study

(Textual Simulation)


The

simulation

is

abstracted

by
:



randomly

generated

data

for

buy/sell

orders,

true

price,

type

of

investor

(informed/uninformed)


keyboard

press

for

dealer

(user)

actions
;



mathematical

computation

of

true/realised

profit
.



random

generation

of

key

variables

that

can’t

be

determined

by

a

mathematical

formulation
:


the

true

price

of

the

security


The role of the dealer in the determination of the true price, and
transaction price


The type of investor (informed/uninformed)


The flow of buyer and sellers


The hidden intentions of the investors to buy or sell


Bid

Ask

True price

Normal
buy/sell

Bid

Ask

True price

More
sellers

Bid

Ask

True price

More
buyers

Bid

Ask

Wide spread

Few
uniformed
trades

Bid

Ask

Dealer

Buys at
his bid
price

Sells at
his ask
price

Impact of

bid /ask /true price

on buy/sell flow

Different Construals

Trading behavioural impact
Dealer actions

Trading patterns
External events

True price

Trading patterns
Trading patterns

dealer actions

trading patterns
Trading patterns

dealer actions

true price

trading patterns
Market efficiency impact
Dealer actions

trading patterns

true price
Dealer action & external events

trading patterns

true price
Time duration impact
Duration for trading intention to
materialize in actual trading action

true price

dealer actions
Market transparency impact
True price is transparent to dealer

(true price

dealer
behavior)
True price is transparent to trader

(true price

trading behaviour)
Adopted convention: x

y indicates that x has a direct impact on y
Different situations can be construed in different ways

EM principles, techniques,

tools and notations

Empirical Modelling
principles


observation


agency


dependency


definitive representations of state



definitive scripts


agent-oriented analysis


representation of state-transitions.
Empirical Modelling
techniques


construe a situation


construct
interactive situation
models (ISM)


metaphorical representation
through ISM
Empirical Modelling
notation

LSD account
for
observables
{

oracles
handles
states

derivates

protocol
}
Empirical Modelling
tools


EDEN interpreter


distributed variant of EDEN
2D Simulation

LSD Description


Agent Oriented Analysis:


observables :


dealer (bid, ask, spread, inventory, true profit, actual profit)


buyer/seller (informed/ uninformed, quantity bought/sold,
transaction price)


simulation clock (simulation time and speed)


transaction details (side, quantity, actual price, true price)


security (thinly traded, actively traded)


market (monopoly dealer market, single security)


warning message to dealer


LSD description

Agent

Dealer{

state

inventory, bid , ask , spread, true profit

Oracles

flow of orders, order side (buy/sell) (+1/
-
1), order quantity, inventory level, actual
profit, his estimated true value of security, his knowledge of informed
trader

Handles


Bid, ask, spread

Protocols

if (estimated true price > ask) || (informed trader rush to buy) ) => raise ask

if (estimated true price < bid) || (informed trader rush to sell)) => raise bid

if (spread between true and trade price is wide) || (informed traders are trading
more often) => adjust quotes according to flow of orders

if (spread is wide) || (few uninformed traders are trading) => narrow the spread

if (inventory approaching the limit of +/
-
10,000 ) => adjust quotes to attract buy
and sell orders appropriately

}




State and state change



Visualization of
trading patterns

Textual info on
current transaction

Visualization of bid, ask,
true price positions

Visualization of the dealer
true profit, actual profit, and

inventory position

The Tkeden interpreter
input window

Time

Dealer’s actions

Transaction history on
dos prompt window of
Tkeden interpreter

Figure 2.
Snapshot of the ISM (Interactive Situation Model) of the Monopoly Dealer




GUI

ž
buttons: user (dealer actions),
report generation

ž
dpi visual metaphor:
investors flow

ž

visual metaphor for
true
price, bid, ask, spread, dealer
position

ž
Text window: current
transactions per flow time
interval, messages to dealer



Tkeden in
put window:

Stopping/ starting,
agents manipulation


Eden script



Agents:

DEALER


BID


ASK


SPREAD

TRUE PRICE

FLOW OF BUYER & SELLER
{NORMAL,
RUSH}

FLOW TIME INTERVAL

REAL TIME CLOCK



Actions:

DEALER ACTION OVER BID ASK SPREAD

CLOCK TIME TICK

REPORT REQUEST



Formulated scenario:

Flow of buyers & sellers
is

Func(BID, ASK, TRUE PRICE
, FLOW TIME
INTERVAL, INVESTOR IN RUSH, INVESTOR
IN NORMAL MODE

)













Dealer po
sition is Func(TRADE GENERATION)



EDEN constructors:

Dependencies, switch, if, while, todo







Donald script

Visualize true price, bid, ask, dealer position


Scout Script

Buttons, text area, pictures to enable user action
and display state









Bi d

Ask

True price

Normal
buy/sell

Bi d

Ask

True price

More
sellers

Bi d

Ask

True price

More
buyers

Bi d

Ask

Wi de spr ead

Few
uniformed
trades

Bi d

Ask

Dealer

Buys at
his bid
price

Sells at
hi s ask
price



2D Simulation

State and state
representation

2D Simulation

Distributed Views

Distributed views is to frame

Agent’s Oracles (what an agent can see)

Agent’s Handles (what an agent can manipulate)


Two views are identified:

Dealer view :

Oracles
-

the dealer can see:


1) flow of buyers/seller but not the type of buyer/sellers


2) clock


3) bid, ask, spread, actual position


4) warning messages


5) current & history of transaction

Handles
-

the dealer can manipulate bid/ask spread and request report

Global Market view:


this view include dealer view plus oracles to true price, type of trader
(informed/uninformed), true position of dealer and it has control (handles)
over time.





GLOBAL MARKET VIEW

Oracles to

true price

Oracle to type


of investor

Oracles to

true position

of dealer

DEALER’S VIEW

Flow of

investors

Reporting

history of

transactions

Tkeden

input

Dealer’s

oracles

Dealer’s

actions

Current

transaction

VR Scene

Observables
-

State
-

State Change

Clock

Dealer

(1) Visualisation of Observables

that have real geometric representation

Flow of investors

(2) Visualisation of Observables that have

abstract geometric representation

prices

profit

inventory

Current transactions

History of transactions

(3) Visualisation of Actions

Dealer actions

Time advance

request for reporting



Type of investor

The Virtual Monopoly Dealer

Steps

for

constructing

a

VR

scene

for

the

monopoly

dealer

simulation
:



the

EM

analysis

was

imported
.



we

adopted

a

visualization

for

abstract

numeric

indicators,

agent

actions,

and

the

human

(user)

role

in

the

scene,

and

added

sound

support

to

produce

warning

messages

to

the

dealer
.



The

distributed

views

in

the

2
D

simulation

were

replaced

by

a

single

VR

scene

including

3

rooms
:

the

dealer

action

room,

the

transactions

history

room,

and

the

hidden

knowledge

room
.



Transactions

are

saved

in

a

file

and

are

visualized

in

the

transaction

history

room
.

The

set

up

for

the

experiment

is

developed

on

a

Silicon

Graphics

Machine

running

Irix
6
.
5
,

and

using

Parametric

Technology

Corporation’s

VR

modelling

tool

Dvise,

and

the

peripheral

includes

a

3
D

mouse

as

an

input

device,

CrystalEYES

glasses

for

the

Stereographic

image

and

3
D

auditory

feedback
.


The VR Scene


Dealer Action

Performance Metrics

Quantitative Metrics

User (Dealer profitability)

Sum of profit for 5 users

Qualitative Metrics

(Cognitive Dimension)

Viscosity
-

Visibility
-

Textual

2D

VR scene

High

Low

Medium

Medium

Medium

High

1000

2000

1000

Virtual Reality


Immersive Systems


devices such as headsets, gloves, 3D
pointing devices are used to immerse the
user into the virtual reality world


Telepresence


linking remote sensors in the real world
with the sensors of a human operator via
robot. The real world in this case is too
hazardous or impossible for a human to
enter.


Window on World (WoW) systems


using standard computer monitor, 3D
spatialised sound, and a stereoscopic display
using LCD shutter glasses, to display the
virtual world

Systems


Human/Technology

integration



Enabling

the

user



immersion


walkthrough


flying

through

the

VR

scene


Convey

reality

using

3
D

geometric

representation,

visual

metaphors,

and

enhanced

multimedia

support

Features

VR For Financial Trading

Vs VR for assembly

Financial Trading

Assembly

Considerations

Steps for the

creation of

VR scene

User Role

Objective

Enhanced cognition of

financial markets phenomena

Minimise the requirements

to build physical prototypes

3D capturing/conversion/

added behavior

Exploration of possible construals/

geometric abstraction of concepts/

integration with databases

Defining objects/visualizing objects/

adding behavior

Construing a situation/

Identifying observables/ state/state change/

geometric abstraction of observables/

visualizing state of observables in a scene/

inducing and visualizing state change

Direct manipulation of virtual objects

Exploring a state/

taking actions to change state

Broad, weakly structured,

many situations

admits different construals,

has an infinite development space

VR For Financial Trading

Vs VR for assembly

Financial Trading

Assembly

Concise

well structured

has a finite development space,

solution, technologies, and tools

Conclusions


The use of VR to model a social context (financial trading)


Different approaches should be adopted for the
use of VR in different contexts


VR is suitable for the exploration of a state
drawn from a social context


EM Analysis supports VR implementation


Qualitative (Cognitive Dimensions) and
Quantitative metrics (empirically recorded in
interaction with the VR scene and in the real
world) should be adopted to assess the VR
added value


Future Research in VR


Scaling up

VR technology to
allow multiple state exploration
(the design of distributed VR
system to model a social context)



Opening up

VR technology to
model complex social environment


Integrating
VR technology with
existing programming paradigm
(databases, definitive programming,
…)


EM Script
Definitions
Re-definition
Agent Actions
state 1
state 2
state 3
state 4
state
n-1
state n
pattern of state
evolution
observer /
participant
oracles
handles
oracles
handles
:
VR scene
state
exploration
VR as a front
-
end to


definitive EM tools