24
1 Convex Risk Parity, Fat-tails Management, and practical CVA/XVA Mihail Turlakov Global Derivatives conference Budapest, May 2016

Global Derivatives - Turlakov - final

Embed Size (px)

Citation preview

Page 1: Global Derivatives - Turlakov - final

1

Convex Risk Parity, Fat-tails Management, and practical CVA/XVA

Mihail Turlakov

Global Derivatives conference

Budapest, May 2016

Page 2: Global Derivatives - Turlakov - final

2

Convex Risk Parity, Fat-tails Management, and practical CVA/XVA

CVA/XVA portfolio management – part 1XVA desks are portfolio managers, not perfect hedgers - what is the reality of

CVA/XVA management?

The major tension at the heart of CVA management

CVA/XVA are tail-risks

Convex Risk Parity and Tail Management – part 2

Cross-asset and ‘systemic carry’ portfolios – what are the practical and “efficient frontier” strategies to manage these portfolios?

The modern portfolio theory and the drawdown/loss aversion

Tail management

Convex Risk Parity

Page 3: Global Derivatives - Turlakov - final

3

Why not to hedge perfectly?

Single-name protection is not available for large part of the portfolio CDS market has limited coverage and liquidity. CDS market became less liquid (a side effect of

regulations), three times less volume from 2008 to 2015

Many names (i.e. project finance, etc.) never had publicly traded credit instruments

Shorting bonds is also a precarious effort

Systemic hedges by indices (ITRX, etc.) or sovereign CDSs is a proxy/model hedge Jump-to-Default and recovery are not hedged

MtM model volatility is hedged, but basis risk (between index and single-name) is increased

Although CVA was intended by regulators to be hedged fully, and the banks obeyed as

best as they can. But the direct hedging of the credit risk of CVA cannot be done in

practice - the major tension at the heart of CVA reality.

Reality is quite different than it actually is-Antoine de Saint-Exupery-

Page 4: Global Derivatives - Turlakov - final

4

XVAs (CVA, FVA, MVA and KVA) are systemic risks of corporate and sovereign

hedging which explode in the market crises

Systemic concentration risks transform into counterparty and funding crises, although rarely yet

quite abruptly

XVAs (CVA, FVA, MVA and KVA) are illiquidity risks

Not easily tradable/novated OTC, not repo-able and only CCP-convertible at full cost

CVA/XVAs are a measure of Tail-Risks (1)

Tail risk is a very rare but strong-impact event

Page 5: Global Derivatives - Turlakov - final

5

Coupled shocks in credit, FX, and IR/funding markets cause even more non-linear

changes of CVA and FVA

“Good, bad, and ugly” feedback loops and liquidity

Wrong-way risk is important

Defaults (jump-to-default in CVA) and liquidity squeezes (jump to insolvency in FVA)

are “last stage” tail-risk events

CVA/XVAs are a measure of Tail-Risks (2)

Page 6: Global Derivatives - Turlakov - final

6

Bank’s credit risk – a complex picture

Credit risk on a loan – accrual accounting in the banking book

Credit risk on a loan is not hedged typically, because credit risk is a main business of a

bank

Credit risk on a derivative – CVA MtM (mark-to-market) in the trading book

Credit risk is expected to be hedged (Tier-1s approach), because the management and

shareholders do not like revenue volatility

The exposure is in principle unlimited

CVA and Loan management on the bank’s level - mixed accrual and MtM risks

Credit exposures of the loans are determined by longer economic cycles and “rebalanced

by real economy”

Derivatives exposures are determined by shorter market cycles and ”hedged”/transformed

Different capital treatment and costs in the Trading and Banking books - FRTB

Credit risks on derivatives and loans are better managed on a portfolio basis, but

the reality and practice are complicated.

Page 7: Global Derivatives - Turlakov - final

7

XVA management – the part 1 – key points

CVA/FVA/KVA models can vary somewhat, but CVA/XVA management is very

different in different banks due to multiple (organisational, top-management, different

markets, and business models) reasons

CVA/XVAs are tail-risks with non-hedgeable jump-to-default and recovery risks

Significant model risks – “basis risks by design”, wrong-way risk, forward-rating risk, ... appreciated

only over economic cycle (2-5 years)

CVA/XVA management is better to have some pro-active elements

Tier-1s and some Tier-2s banks have to hedge CVA PL volatility to manageable levels

Tier-2s and Tier-3s can, very sensibly, keep the credit risk and not hedge, yet “management-

sensitivity threshold” is likely to be breached under some stressed conditions

Page 8: Global Derivatives - Turlakov - final

8

Motivational questions and other talks at Global Derivatives 2016

XVA desks are portfolio managers, not perfect hedgers - what is the reality of

CVA/XVA management? Quantitative understanding and proposals

Good example – FVA debate - FVA can be hedged in theory but not in practice (Kjaer, Burgard), the

reality of the markets and internal banks operations are the main issues

Andrey Chirikhin – junior quant seminar – the overview of XVA and its capital requirements

Andrew Green – some real world problems of XVA risk managing

Mihail Turlakov – part 1 – brief review of the practice of CVA/XVA management

Chris Kenyon – the trade-off between MVA and KVA

Massimo Morini – innovating XVAs (including “perfecting imperfect hedging”)

Leif Andresen – bank’s balance sheet and FVA – the FVA debate continued

Cross-asset and ‘systemic carry’ portfolios – what are the practical and “efficient

frontier” strategies to manage these portfolios? Portfolio and Risk management

understanding and proposals

Buy-side Summit: QUANTITATIVE INVESTMENT & PORTFOLIO STRATEGIES Nick Baltas – exploiting cross-asset carry Kartik Sivaramakrishnan, Robert Stamicar - Optimising Multi-Aset Portfolios To Minimise

Downside Risk: A CVaR Based Approach

Emanuel Derman – the review of “good and not so good ideas”

Mihail Turlakov – part 2 – Tail risk management and Convex Risk Parity

Page 9: Global Derivatives - Turlakov - final

9

part 2 - Convex Risk Parity, Fat-tails Management, and practical CVA/XVA

Convex Risk Parity and Tail Management – part 2

A brief review of Risk Parity

The portfolio theory and a drawdown/loss aversion

Kelly portfolio leverage and fat tails

Fat-tails management

Convex Risk Parity

Page 10: Global Derivatives - Turlakov - final

10

Risk Parity (1) – a brief review

Risk Parity (RP) is a risk-adjusted portfolio management approach The assets in the portfolio are risk-balanced, i.e. weighted by the inverse of the risk, here volatility

The leverage is used for low-volatility, positive return, assets (G10 government bonds, typically).

More successful approach than the classic 60/40 mix of stocks/bonds throughout the long history

Avoid unintended skews and risks of the mean-variance portfolio theory

RP assumes equal risk-adjusted expected returns Returns are not predictable, no input of “exact expected” returns and correlations

Correlations can regime-change and are not stable

Exploiting the leverage by using derivatives - the drawdowns have to be controlled

A connection to part 1 - RP is well suited for carry portfolios (i.e. CVA portfolio) In practice, a combination of model hedging (the standard approach according to the internal

model) and Risk Parity can be applied in order to keep PL volatility within desired threshold.

Page 11: Global Derivatives - Turlakov - final

11

Risk Parity (2) – criticisms and developments

Risk Parity criticisms “Lucky” success due to the era of trending lower rates and the bonds leverage

Returns (risk premiums) have to be considered

Correlations between stocks and govt bonds has been consistently negative. What happens if

that's no longer the case?

Negative skew of positive return assets and tail/drawdown risks

Risk Parity family – the variety of approaches and performances “All Weather” (Bridgewater Associates), Tail Risk Parity (AllianceBernstein, GAM), “Two regimes”

Risk Parity (First Quadrant), Active Risk Parity (AQR), Risk Parity with the inclusion of momentum

(Salient), Factor-based Risk Parity (JP Morgan AM), Invesco, PIMCO, etc.

Page 12: Global Derivatives - Turlakov - final

12

Risk Parity (3) – addressing the drawdown/tail risk

The markets and the world are fat-tailed, therefore we have to deal with tail events

Tail-Risk Parity (TRP) adopts Risk Parity approach with the focus on the drawdown

protection The drawdown (tail loss), rather than volatility, is considered as more important measure of risk

Risk Parity focuses on volatility, Tail-Risk Parity [Alankar] defines risk as expected tail loss (ETL)

Markets can sustain the notion of risk as drawdown risk (permanent loss)

Empirical results - “Risk premium is strongly correlated with tail-risk skewness but very little

with volatility” (a clear exception is trend following) [Lemperiere]

Trade-off between skewness and excess returns – basic reasons and mechanisms

Crowding into carry trades – decreasing returns and increasing downside (“unwind tail risk”)

Liquidity shocks and “super-shocks” - fast market snap reactions and re-pricing

Behavioural balance (prospect theory) of the “skew value” for buyers(hedgers)and sellers

Page 13: Global Derivatives - Turlakov - final

13

Portfolio theory and the drawdown aversion (1)

Drawdown/loss[Kahneman]/leverage[Asness]/ETL/VaR aversion “What the investors fear is the possibility of permanent loss”… not the volatility

More important risk measure

Drawdown aversion is essentially a more general and quantitative version of the

leverage aversion Leverage aversion changes the predictions of the portfolio theory [Asness]

Leverage is a tool and a capability, not a danger by itself, while a permanent loss is

The different costs and types of the leverage/drawdown aversion The cost of OTM insurance - “hard drawdown” aversion - overall portfolio leverage

Liquidity and rebalancing costs – margin cash and leverage of some particular assets

Gap (il)liquidity costs – unwinding of the portfolio

Outflows risk

Page 14: Global Derivatives - Turlakov - final

14

Portfolio theory and the drawdown aversion (2)

Concave and lower efficient frontier due to the drawdown aversion Different Capital Lines for different levels of drawdown aversion

Differentiation between investors due to their different drawdown aversion

Page 15: Global Derivatives - Turlakov - final

15

Portfolio theory and the drawdown aversion (3)

The investors are sensitive to mark-to-market and price of liquidity– [Illinski] The stop-loss cost makes efficient frontier strongly concave and even downward sloping

Unstable leveraged returns due to “selling of tails "and optionality

Superiority of option portfolios in “Extremistan” (world with fat tails) – [Thorp] Option portfolios can reach beyond geometric efficient frontier of the underlying – buy

attractively priced “lottery tickets”

Options, being on the efficient frontier, can have better MDD (Maximum Drawdown Distribution),

even in “Mediocristan” (Gaussian world)

return

risk

𝑅𝑓𝑟𝑒𝑒

CAPM

Reality

Page 16: Global Derivatives - Turlakov - final

16

Kelly allocation for the long run – the stinging tail

The effect of the drawdown on Kelly ratio [Kelly] Kelly allocation is very sensitive to vol skew and tail drawdowns

Kelly skew – asymmetric affect of left and right fat tails*

Sharpe ratio is not a good guide in the presence of the tails

The growth becomes negative

* from extended Kelly calculations Sharpe= Kelly * volatility

Page 17: Global Derivatives - Turlakov - final

17

Kelly Risk Parity and skew – open questions

Kelly Risk Parity (KRP) If the unknown excess returns are the same, RP is equivalent to Sharpe maximization (i.e. MVA).

Under the same condition, maximizing long-run growth is 1/variance parity – KRP, the

general leveraged risk parity, is the variance parity (VP) therefore.

Positive skew of KRP due to 1/variance weighting versus negative volatility and Kelly skews

The dynamics and conditional probabilities are very important Volatility skew is similar to wrong-way risk (WWR)

Vol skew - left-tail risk-off spot move is accompanied by higher vol

WWR - left-tail credit widening is accompanied by risk-off moves in spots/rates

See a scenario model of wrong-way risk [Turlakov] in CVA/XVA context

We need to free ourselves from “average” thinking-P. W. Anderson (Nobel laureate in Physics)-

Page 18: Global Derivatives - Turlakov - final

18

Fat-tails management – general aspects

Tail dangers-opportunities[Brown] is much more about risk management than

about the core strategy Be prepared not to make stupid mistakes under stress conditions

(All) Tail events cannot be hedged by definition

Risk Management is about human decisions Market takes a single path unlike Monte Carlo simulations

A strong behavioural uncertainty about the consequences and the level of the drawdown

Portfolio management – a mix of active and passive elements Diversification is a risk reduction, not risk management

Actively managed fat left and right tails To protect against major drawdowns and be flat during “normal periods”.

Left-Tail-Risk options hedging with monetisation rules and momentum[Bhansali]

Right-Tail (risk-on) options (beyond base efficient frontier [Thorp])

Page 19: Global Derivatives - Turlakov - final

19

Tail Dangers-Opportunities (1a) – the insights

Volatility is the only real asset class [Artemis]… for liquid financial assets Carry strategies, risk parity, etc. are effectively volatility strategies

Page 20: Global Derivatives - Turlakov - final

20

Tail Dangers-Opportunities (1b) – the insights

Inconvenient truth about stocks-bonds anti-correlation Shadow Convexity of major Central Banks [Artemis] - exploit or protect against?

Page 21: Global Derivatives - Turlakov - final

21

Convex Risk Parity – portfolio management framework

Convex Risk Parity A strategic approach/portfolio is Risk Parity – passive core (in the sense of conventional core-

satellite approach of the portfolio management). Naturally, short volatility and convexity. Risk Parity

creates a systemic framework with low turnover and liquidity costs

A positive convexity (long volatility) for selected drawdown risks – active satellite. Active

management is about major tensions/tail risks – buying insurance-lottery tickets

Convex Risk Parity – positive carry, negative convexity, core Risk Parity diversification – a portfolio of pairs of (possibly illiquid) anti-correlated assets

Kelly Risk Parity - optimal leverage and positive skew

Macroeconomics with the view on major structural imbalances a) carry and major trends b) risk-

off/safe-heaven assets (currently, G10 govy bonds)

Convex Risk Parity – active, positive tail convexity, satellite Actively managed fat left and right tails. These strategies aim to protect against major drawdowns

(the main objective) and at a flat return during “normal periods”.

Avoid illiquid (usually credit/counterparty related?) and model-dependent risks and assets

Page 22: Global Derivatives - Turlakov - final

22

Summary

CVA/XVA management is the portfolio management “Going the full circle” – from introducing the illiquidity adjustments (CVA/XVA), enforced by

regulators, to realising the limits of CVA/XVA hedging

Practical CVA/XVA management must have some pro-active elements

Convex Risk Parity and Tail Management

Convex Risk Parity is an attractive portfolio approach for ‘systemic carry’ portfolios

combine dynamic Positive Convexity and Kelly Risk Parity

The modern portfolio theory must consider the drawdown aversion

Risk Management of tail events is better to be dynamic (option-based, momentum, etc.)

Page 23: Global Derivatives - Turlakov - final

23

References

1. [Asness] C. Asness, A. Frazzini, L. H. Pedersen, “Leverage aversion and Risk Parity” (2012)

2. [Illinski] K. Illinski, A. Pokrovski, “Lies of Capital Lines” (2010)

3. [Thorp] E. Thorp, S. Misuzawa, “Maximising capital growth with Black-Swan protection” (2013)

4. [Alankar] A. Alankar, M. DePalma, M. Scholes, “The introduction to Tail Risk Parity” (2012)

5. [Lemperiere] Y. Lemperiere,… J-P. Bouchard, “Tail risk premiums versus pure alpha” (2015)

6. [Bridgewater] Bridgewater Associates, Daily Observations (August 18, 2004)

7. [Kelly] J. L. Kelly, “A new interpretation of information rate” (1956)

8. [Artemis] Vega Fund, C. Cole, “Shadow Convexity and Prisoners Dilemma” (2015)

9. [Kahneman] D. Kahneman, A. Tversky, “Advances in prospect theory……” (1992)

10. [Brown] A. Brown, “Red Blooded Risk” (2012)

11. [Bhansali] V. Bhansali, “Tail Risk Hedging” (2014)

12. [Booth] J. Booth, “Emerging markets in an Upside Down World” (2014)

13. [Turlakov] M. Turlakov, “Wrong-way risk, credit and funding” (2013)

Page 24: Global Derivatives - Turlakov - final

24

The opinions presented here is a personal opinion of the author. They

do not represent the opinions of Sberbank CIB. Neither author nor his

employer are responsible for any use of the presented material. None

of the ideas in this presentation are claimed to be used or will be used.

Disclaimer