Dimensional Analysis, Scaling, and Zero-Intelligence Modeling For Financial Markets
Dimensional Analysis, Scaling, and Zero-Intelligence Modeling For Financial Markets
Zero-intelligence modeling
Two principles
Two consequences
l
frequency
Market (v)
Limit (p,v)
LO at single price
clear in order of
arrival
buy
market
orders
sell
market
orders
price
s
h
a
r
e
s
bids
offers
spread
Observables of the CDA
Midprice:
m = (ask + bid) / 2
Diffusion constant of
the random walk is
called the volatility
<
[
m
(
t
+
)
-
m
(
t
)
]
2
>
0
2.0
4.0 6.0 8.0
time (events)
0
1E-4
Midprice motion is a good
approximation to diffusion
Courtesy Paolo Patelli
Market Impact
Yet it doesnt
price
original
ask
average price
you pay
Farmer, J. D., P. Patelli, and I. I. Zovko. "The Predicitive Power of Zero
Intelligence in Financial Markets" PNAS USA 102(11) (2005): 2254-2259
Observed regularities of the market impact
Corresponds to a power-law
distribution of limit orders
p
m
0
.
2
0
.
4
0
.
6
0
.
8
l0
-l
l0
-2
l0
0
Order size
Sell orders
0.23
p
m
0
.
2
0
.
4
0
.
6
0
.
8
l0
-l
l0
-2
l0
0
Order size
price
n ~ p
3/4
~ p
l/4
Courtesy Paolo Patelli
Liquidity
original
ask
average price
you pay
price
original
ask
average price
you pay
s
h
a
r
e
d
e
n
s
i
t
y
buy
market
orders
sell
market
orders
price
s
h
a
r
e
s
bids
offers
spread
Price (coordinate)
Spread (observable)
Diffusivity (observable)
How well does classical scaling work?
Try
collapsing
the market
impact four
different
ways
Farmer, J. D., P. Patelli,
and I. I. Zovko. "The
Predicitive Power of
Zero Intelligence in
Financial Markets"
PNAS USA 102(11)
(2005): 2254-2259
Classical scaling predictions for the spread
Tick size
Comparisons to data
are remarkably good
0.00 0.05 0.10 0.15 0.20 0.25
0
.
0
0
.
2
0
.
4
0
.
6
0
.
8
simulation f()
s: London Stock Lxchange (vOD)
n
o
n
d
i
m
e
n
s
i
o
n
a
l
i
z
e
d
s
p
r
e
a
d
The volatility
)
-
m
(
t
)
]
2
>
/
p
C
2
Candidate for a
regularity of behavior
From Mike, S., and J. D. Farmer. "An Empirical
Behavior Model of Price Formation." Santa Fe
Institute Working Paper 05-10-039
V. Some compelling problems in
income and wealth distribution
Income in wage
range is lognormal
or exponential
Income of wealthy
is power-law
Both distributions
are forms of
maximum-entropy
subject to different
constraints
Farmer, J. D., D. E. Smith, and M. Shubik. "Is Economics the Next Physical Science?" Physics Today 58
(9) (2005): 37-42.
Farmer, J. D., P. Patelli, and I. I. Zovko. "The Predicitive Power of Zero Intelligence in Financial Markets"
PNAS USA 102(11) (2005): 2254-2259.
Lillo, F., S. Mike, and J. D. Farmer. "Theory for Long Memory in Supply and Demand." Phys. Rev. E 7106
(6 pt 2) (2005): 287-297.
Lillo, F., and J. D. Farmer. "The Key Role of Liquidity Fluctuations in Determining Large Price
Fluctuations." Fluctuations and Noise Lett. 5 (2005): L209-L216.
Challet, D., Marsili, M., and Zhang, Y.-C. Minority Games, Oxford U. Press, New York (2005)
Farmer, J. D., L. Gillemot, F. Lillo, S. Mike, and A. Sen. "What Really Causes Large Price Changes?"
Quant. Fin. 4(4) (2004): 383-397.
Smith, E., J. D. Farmer, L. Gillemot, and S. Krishnamurthy. "Statistical Theory of the Continuous Double
Auction." Quant. Fin. 3(6) (2003): 481-514.
Lillo, F., J. D. Farmer, and R. Mantegna. "Master curve for price impact function." Nature 421 (2003): 129.
Bouchaud, J.-P., and Potters, M. Theory of Financial Risk: From Statistical Physics to Risk Management,
Cambridge U. Press, New York (2000)