Nematrian
Software
Research
Consulting
Log in
/
Views on non-Normal markets
This presentation explores what we mean by 'extreme events' in the context of investment markets, what might cause these sorts of events, how they can be best analysed and catered for in portfolio construction and how to avoid being overconfident concerning them.
Slides
1
Views on non-Normal markets
2
Views on non-Normal markets
3
Agenda
4
Agenda
5
Extreme events and fat-tails
6
Examples of extreme events
7
How really extreme has 2007-09 been
8
Agenda
9
Visualising fat-tailed behaviour in individual return series
10
Example derivation of a quantile-quantile plot
11
Fat-tailed behaviour depends partly on timescale
12
More periods give more scope for extreme events
13
Skew(ness) and kurtosis
14
Interpretation via Cornish-Fisher asymptotic expansion
15
Flaws in Cornish Fisher (and hence in skew/kurtosis)
16
A better approach?
17
Agenda
18
What causes fat-tailed behavior?
19
Time-varying volatility
20
Explains some market index fat tails, particularly on upside
21
Not just a developed market phenomenon
22
A longer term phenomenon too
23
Time-varying volatility
24
Regime switching
25
Regime switching (continued)
26
Crowded trades
27
Leverage
28
Other examples of leverage
29
Impact that leverage can have on portfolio behaviour
30
Agenda
31
Fat tails and portfolio construction
32
Fat tails - in joint return series
33
Visualisation of joint fat-tailed behaviour
34
Quantile-quantile box plots
35
Distributional mixtures - again!
36
Principal components analysis (PCA)
37
Applying PCA to sector relatives
38
Adjusting for time-varying volatility in joint return series
39
Longitudinal time-varying volatility adjustment
40
Cross-sectional time-varying volatility adjustment
41
Back-testing time-varying volatility adjustments
42
Cross-sectional dispersion of pair-wise combinations
43
Impact of time-varying volatility adjustment
44
Agenda
45
Possible selection effects arising from active management
46
Implications for modelling
47
PCA vs. ICA
48
Including 'meaning' as well as 'noise'
49
Selection effects are potentially very important
50
Selection effects - Summary
51
Agenda
52
Portfolio construction
53
Portfolio construction - sensitivities
54
Portfolio construction - impact of fat tails (1)
55
Portfolio construction - impact of fat tails (2)
56
Other approaches - (1) distributional mixtures
57
Other approaches - (2) lower partial moments
58
Estimating lower partial moments
59
Views on non-Normal markets
60
Agenda
61
Appendix A: PCA vs. ICA - Noise vs. Explanation?
62
Independent components analysis
63
E.g. Non-Normality and Projection pursuit
64
Projection pursuit algorithm (1)
65
Projection pursuit algorithm (2)
66
Independent components analysis (ICA)
67
Blending together PCA and ICA
68
Identifying Principal Components one at a time
69
Blending together PCA and ICA
70
Appendix B: System-wide leverage effects
71
Does marking-to-market exacerbate leverage effects?
72
Systemic risk
73
How best to regulate firms?
74
Appendix C: Model risk and back-testing
75
Model assessment
76
Link with calibration
77
Ideally model should also fit well 'period by period'
78
Look-back bias and in-sample testing
79
Look-back bias and out-of-sample testing
80
Market consistent risk measurement
81
Back-testing in other regulatory frameworks
82
Appendix D: Stress Testing
83
Stress Testing - Additional drivers
84
Different meanings given to the term 'Stress Testing'
85
Reverse stress testing
86
Stress testing - other observations
87
Stress testing - statistical techniques
88
References
89
Important Information
NAVIGATION LINKS
Contents
|
Next
|
Library
Desktop view |
Switch to Mobile