![Page 1: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/1.jpg)
Introduction to Algorithmic Trading StrategiesLecture 2
Hidden Markov Trading Model
Haksun [email protected]
www.numericalmethod.com
![Page 2: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/2.jpg)
2
Outline Carry trade Momentum Valuation CAPM Markov chain Hidden Markov model
![Page 3: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/3.jpg)
3
References Algorithmic Trading: Hidden Markov Models
on Foreign Exchange Data. Patrik Idvall, Conny Jonsson. University essay from Linköpings universitet/Matematiska institutionen; Linköpings universitet/Matematiska institutionen. 2008.
A tutorial on hidden Markov models and selected applications in speech recognition. Rabiner, L.R. Proceedings of the IEEE, vol 77 Issue 2, Feb 1989.
![Page 4: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/4.jpg)
4
FX Market FX is the largest and most liquid of all
financial markets – multiple trillions a day. FX is an OTC market, no central exchanges. The major players are:
Central banks Investment and commercial banks Non-bank financial institutions Commercial companies Retails
![Page 5: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/5.jpg)
5
Electronic Markets Reuters EBS (Electronic Broking Service) Currenex FXCM FXall Hotspot Lava FX
![Page 6: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/6.jpg)
6
Fees Brokerage Transaction, e.g., bid-ask
![Page 7: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/7.jpg)
7
Basic Strategies Carry trade Momentum Valuation
![Page 8: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/8.jpg)
8
Carry Trade Capture the difference between the rates of
two currencies. Borrow a currency with low interest rate. Buy another currency with higher interest
rate. Take leverage, e.g., 10:1. Risk: FX exchange rate goes against the
trade. Popular trades: JPY vs. USD, USD vs. AUD Worked until 2008.
![Page 9: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/9.jpg)
9
Momentum FX tends to trend.
Long when it goes up. Short when it goes down.
Irrational traders Slow digestion of information among
disparate participants
![Page 10: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/10.jpg)
10
Purchasing Power Parity McDonald’s hamburger as a currency. The price of a burger in the USA = the price
of a burger in Europe E.g., USD1.25/burger = EUR1/burger
EURUSD = 1.25
![Page 11: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/11.jpg)
11
FX Index Deutsche Bank Currency Return (DBCR)
Index A combination of
Carry trade Momentum Valuation
![Page 12: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/12.jpg)
12
CAPM Individual expected excess return is
proportional to the market expected excess return.
, are geometric returns is an arithmetic return
Sensitivity
![Page 13: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/13.jpg)
13
Alpha Alpha is the excess return above the expected
excess return.
For FX, we usually assume .
![Page 14: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/14.jpg)
14
Bayes Theorem Bayes theorem computes the posterior
probability of a hypothesis H after evidence E is observed in terms of the prior probability, the prior probability of E, the conditional probability of
![Page 15: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/15.jpg)
15
Markov Chain
s2: MEAN-
REVERTING
s1: UPs3:
DOWN
a22 = 0.2
a11 = 0.4 a33 = 0.5
a12 = 0.2
a21 = 0.3
a23 = 0.5
a32 = 0.25
a13 = 0.4
a31 = 0.25
![Page 16: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/16.jpg)
16
Example: State Probability What is the probability of observing
0.250.40.4
![Page 17: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/17.jpg)
17
Markov Property Given the current information available at
time , the history, e.g., path, is irrelevant.
Consistent with the weak form of the efficient market hypothesis.
![Page 18: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/18.jpg)
18
Hidden Markov Chain Only observations are observable (duh). World states may not be known (hidden).
We want to model the hidden states as a Markov Chain.
Two assumptions: Markov property
![Page 19: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/19.jpg)
19
Markov Chain
s2: MEAN-
REVERTING
s1: UPs3:
DOWN
a22 = ?
a11 = ? a33 = ?
a12 = ?
a21 = ?
a23 = ?
a32 = ?
a13 = ?
a31 = ?
![Page 20: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/20.jpg)
20
Problems Likelihood
Given the parameters, λ, and an observation sequence, Ω, compute .
Decoding Given the parameters, λ, and an observation
sequence, Ω, determine the best hidden sequence Q.
Learning Given an observation sequence, Ω, and HMM
structure, learn λ.
![Page 21: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/21.jpg)
21
Likelihood Solutions
![Page 22: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/22.jpg)
22
Likelihood By Enumeration
But… this is not computationally feasible.
![Page 23: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/23.jpg)
23
Forward Procedure
the probability of the partial observation sequence until time t and the system in state at time t.
Initialization
: the conditional distribution of Induction
Termination , the likelihood
![Page 24: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/24.jpg)
24
Backward Procedure
the probability of the system in state at time t, and the partial observations from then onward till time t
Initialization 1
Induction
![Page 25: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/25.jpg)
25
Decoding Solutions
![Page 26: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/26.jpg)
26
Decoding Solutions Given the observations and model, the
probability of the system in state is:
![Page 27: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/27.jpg)
27
Maximizing The Expected Number Of States
This determines the most likely state at every instant, t, without regard to the probability of occurrence of sequences of states.
![Page 28: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/28.jpg)
28
Viterbi Algorithm The maximal probability of the system
travelling these states and generating these observations:
![Page 29: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/29.jpg)
29
Viterbi Algorithm Initialization
Recursion
the probability of the most probable state sequence for the first t observations, ending in state j
= the state chosen at t
Termination
=
![Page 30: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/30.jpg)
30
Learning Solutions
![Page 31: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/31.jpg)
31
As A Maximization Problem Our objective is to find λ that maximizes . For any given λ, we can compute . Then solve a maximization problem. Algorithm: Nelder-Mead.
![Page 32: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/32.jpg)
32
Baum-Welch the probability of being in state at time , and
state at time , given the model and the observation sequence
![Page 33: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/33.jpg)
33
Xi
![Page 34: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/34.jpg)
34
Estimation Equation By summing up over time, ~ the number of times is visited ~ the number of times the system goes from
state to state Thus, the parameters λ are:
, initial state probabilities , transition probabilities , conditional probabilities
![Page 35: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/35.jpg)
35
Estimation Procedure Guess what λ is. Compute using the estimation equations. Practically, we can estimate the initial λ by
Nelder-Mead to get “closer” to the solution.
![Page 36: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/36.jpg)
36
Conditional Probabilities Our formulation so far assumes discrete
conditional probabilities. The formulations that take other probability
density functions are similar. But the computations are more complicated, and
the solutions may not even be analytical, e.g., t-distribution.
![Page 37: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/37.jpg)
37
Heavy Tail Distributions t-distribution Gaussian Mixture Model
a weighted sum of Normal distributions
![Page 38: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/38.jpg)
38
Trading Ideas Compute the next state. Compute the expected return. Long (short) when expected return > (<) 0. Long (short) when expected return > (<) c.
c = the transaction costs Any other ideas?
![Page 39: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/39.jpg)
39
Experiment Setup EURUSD daily prices from 2003 to 2006. 6 unknown factors. Λ is estimated on a rolling basis. Evaluations:
Hypothesis testing Sharpe ratio VaR Max drawdown alpha
![Page 40: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/40.jpg)
40
Best Discrete Case
![Page 41: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/41.jpg)
41
Best Continuous Case
![Page 42: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/42.jpg)
42
Results More data (the 6 factors) do not always help
(esp. for the discrete case). Parameters unstable.
![Page 43: Introduction to Algorithmic Trading Strategies Lecture 2](https://reader036.vdocument.in/reader036/viewer/2022081422/5681617a550346895dd10a21/html5/thumbnails/43.jpg)
43
TODOs How can we improve the HMM model(s)?
Ideas?