Genetic Programming in Computational Finance

Economics:
Rethink or Sink
Edward Tsang
Acknowledgement:
Much of the idea here came from
Richard Olsen
Classical Economics





To model economy and prices mathematically
Start with assumptions
Results follow
Robust…
… as long as the assumptions hold…
“If I were the
queen of
France, I shall
give you 1
million Euro”
“If you give me
a fish, I shall
sing you a
song”
A Shaky Castle
Classical economics built on critical assumptions
Everybody is perfectly rational
Homogeneity
Market has changed!
Computation is ignored
How RATIONAL
are we?
Which Option Will You Take?
Which Option Will You Take?
£100 now
£10 per month for 12 months
…
What Is Your Move?





What is the optimal
move?
Rules are clearly defined
No hidden information
Shouldn’t a rational
player pick the optimal
move?
Problem: too much to
compute!
What is Rationality?
Some decisions require computation
 Some computation methods are better
than others
 Does it mean that some are more
rational than others?
 If so, the statement “investors are
rational” is ambiguous

Dubious Assumptions
 “Investors
are rational”?
 “Investors
are homogeneous”
 Computation
matters but ignored
… but are they close enough?
Changes In The Market
New
Instruments
e.g. Options,
CFDs
Stock Exchange
The Amazon Butterfly Effect

Movements amplified


By increased leverage
By algorithmic trading

Chain reactions

One reaction causes
more reactions
Markets have changed
Now what?
Can we study market as a hard science?
The “Biology” of Markets
(Richard Olsen)

How was biology studied?





Observe
Copy
Measure
Generalize
…
How Long is a Coastline?
How much weight can we put on various positions?
Market
Physics
We don’t know where people might add weights
But we can study the consequences of possible actions!
A Wiki-style repository of software for studying finance
Data
Users
upload /
retrieve
modules
Interaction
between
modules
(FX rates, stock prices, interest rates, etc)
modules
implementing
models and
operations
Web-based
Open-source
Possibly through
machine learning
Exchanges
Economist
Computer
Scientist
Concluding Summary

Classical economics build castles on sand


The ground has moved


Due to unrealistic assumptions
The castle is sinking
New approaches needed:
Remove unrealistic assumptions
 Treat market as hard science
 Infrastructure required

“The Economic Crisis is a
Crisis for Economic Theory”
Alan Kirman
CESifo Economic Studies, Vol.56,
Issue 4, 2010, pp.498-535
Want more?
This video: http://youtu.be/zIkHjo1ubpg
More info:
http://www.bracil.net/finance/HFF/brief_intro.html
Edward Tsang: http://edward.bracil.net/