quantitative risk management: var and othersΒ Β· quantitative risk management: var and others roy e....
TRANSCRIPT
![Page 1: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/1.jpg)
Quantitative Risk Management: VaR and Others
Roy E. DeMeo, Jr.
Wells Fargo
![Page 2: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/2.jpg)
What is Value At Risk?
β’ Value at Risk, or VaR, is roughly speaking, a measure of how much money a bank or other financial firm can lose on its positions in a fixed period, such as 1 day, 10 days, or 1 year in a βworst caseβ (bottom 1 percent) scenario.
β’ Losses can be due to diffusive moves (βgeneralβ VaR) or defaults or credit migrations (βincremental risk chargeβ), or so-called βspecial eventsβ (βspecific riskβ)
![Page 3: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/3.jpg)
What is VaR Used For?
β’ VaR is used as a tool to decide how large a trading book to have. An equities or FX desk will often have an overall VaR limit enforced by upper management. A βreal-timeβ VaR calculation can determine whether a trade is possible.
β’ VaR is used at the firm level to determine the amount of capital the Feds will require the firm to have. VaR capital is combined with capital requirements from Specific Risk, Stress Scenarios and other risk measures mentioned here.
![Page 4: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/4.jpg)
Basic Calculation Methods
β’ Historical Simulation β last 251 1 or 10-day interval market changes are applied to current conditions, and we take the second worst profit or (more likely) loss.
β’ Monte Carlo Model β Create a parametric model for the next periodβs moves based on current model and last yearβs historical data, simulate N times, take top of bottom 1% of P&Ls
![Page 5: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/5.jpg)
General vs Stress VaR
β’ General VaR estimates bad case 1 or 10 day loss based on the previous yearβs history
β’ Stress VaR estimates bad case 1 or 10 day loss based on the market moves that occurred during a particularly βstressfulβ year, such as Sept 2008 to Sept 2009.
![Page 6: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/6.jpg)
Historical General VaR, More Detail
β’ Bankβs positions defined as ππ , π = 1,β¦ , π.
β’ Each position has risk factors π1,π , π2,π , β¦ , πππ,π chosen primarily from market observable inputs, not from underlying calibrated parameters.
β’ Each risk factor has historical time series ππ,π(π‘) and 1-day shifts βππ,π,π‘ = ππ,π π‘ β ππ,π π‘ β 1 (absolute) or βππ,π,π‘ = ππ,π π‘ β ππ,π π‘ β 1 ππ,π(0)/ππ,π(π‘ β 1) (relative).
β’ For each t, each position indexed by j, define βππ(π‘) =π(π1,π 0 + βπ1,π π‘ , β¦ , πππ,π 0 + βπππ,π π‘ ) βP(π1,π 0 ,β¦, πππ,π 0 ).
![Page 7: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/7.jpg)
Historical General VaR, contβd
β’ Total P&L across all positions βπ π‘ = βππ(π‘)ππ=1
β’ Sort all 251 βπ π‘ β²π one for each business day over the last year, from low to high, and take the second one in the sorted list.
β’ Not really the first percentile, really closer to the 0.8 percentile, but the Feds do not allow interpolating between the 2nd and 3rd worst P&L.
![Page 8: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/8.jpg)
One Day versus Ten Days
β’ 10-day shifts βππ,π,π‘ = ππ,π π‘ β ππ,π π‘ β 10
(absolute) or
βππ,π,π‘ = ππ,π π‘ β ππ,π π‘ β 10 ππ,π(0)/
ππ,π(π‘ β 10) (relative).
β’ Shifts overlap and cover 260 business days total.
![Page 9: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/9.jpg)
Delta/Gamma Approximation
β’ Historical simulation with full revaluation requires 252 pricings for each position, current price plus effect of 251 different shifts.
β’ We may save time by using the estimate
βππ(π‘) β πππ
πππ,πβππ,π,π‘ +
1
2
π2ππ
πππ,π2 βππ,π,π‘
2 + π2ππ
πππ,ππππ,πβππ,π,π‘βππ,π,π‘
π<π
ππ
π=1
It is common to leave out the cross terms since they are often very small. Cross terms are a form of Risk Not in VaR (RNiV), to be covered in a later slide.
![Page 10: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/10.jpg)
Delta/Gamma Contβd
Greeks estimated as
πππ
πππ,π=ππ π1,π , β¦ , ππ,π + β,β¦ , πππ,π β ππ π1,π , β¦ , ππ,π β β,β¦ , πππ,π
2β
π2ππ
πππ,π2
=ππ π1,π , β¦ , ππ,π + β,β¦ , πππ,π β 2ππ + ππ π1,π , β¦ , ππ,π β β,β¦ , πππ,π
β2
For relative bump sizes replace h with βππ,π .
![Page 11: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/11.jpg)
Most Accurate Bump Sizes
β’ Absolute: βπ,π β π . π. (βππ,π)
β’ Relative: βπ,π β πππ ππππππ π£ππ(ππ,π)
![Page 12: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/12.jpg)
Calibrated Parameters vs Risk Drivers
β’ The risk drivers for VaR, the Xβs, are typically market inputs, e.g. spot, ATM volatility, some sort of volatility skew measure, and interest rates, rather than underlying model parameters.
β’ Full revaluation would require re-calibration for every historical shift of the inputs.
β’ Delta Gamma approximation requires inverting a Jacobian matrix (example next slide)
![Page 13: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/13.jpg)
Calibrated Parameters vs Risk Drivers
Let π1,π , β¦ , πππ,π = πΊ(π1,π, β¦ , πππ,π). Then
πππ
ππ1,π, β¦ ,πππ
ππππ,π= π½β1
πππ
ππ1,π, β¦ ,πππ
ππππ,π
π
Example would be if we priced an option using a Heston model with spot, instantaneous vol, correlation and vol of vol, but risk drivers are spot, ATM vol, risk reversal and butterfly. Second derivatives are messier but can be worked out.
![Page 14: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/14.jpg)
Grid Approximation
β’ Usually done on underlyer price risk factor only, so consider the case of just this risk factor.
β’ Compute ππ π β ββπ , β¦ , ππ π ,β¦ , ππ π β βπ .
β’ Suppose that βππ,π‘ = 1 β π βπ + πβπ+1
β’ βππ = 1 β π ππ π + βπ + πππ π + βπ+1 β
ππ(π)
β’ For additional drivers, add the delta/gamma terms.
![Page 15: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/15.jpg)
Expected Shortfall
β’ VaR takes the second worst of 251 P&Ls for the total position.
β’ Expected shortfall computes the average of the 6 or 7 worst P&Ls for the total position.
β’ Some banks use this in lieu of or in addition to VaR because it appears to be more stable.
β’ Regulators will eventually require it.
![Page 16: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/16.jpg)
Common risk drivers
β’ Equities β stock price, discount rate, ATM volatilities by tenor, volatility skew and convexity.
β’ FX β spot, ATM volatility, foreign and domestic rates, risk-reversals and butterflies
β’ Commodities β spot, ATM volatility, various measures of skew
β’ Interest Rates β forward swap rates, ATM volatilities, volatility skew for each expiry and tenor.
![Page 17: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/17.jpg)
Monte Carlo Method
β’ In this case we postulate a stochastic process for the inputs to the price of the position.
β’ May be similar in structure to front office pricing model, except that model must reflect past 1-year history rather than strictly calibrate to market data.
β’ Instead of 251 P&Ls for each position, we can simulate 5000, or 50000 or more scenarios.
β’ VaR is set to 1st percentile, i.e. 500th worst out of 50000.
![Page 18: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/18.jpg)
Stress VaR
β’ Similar to General VaR, except that historical shifts in risk factors are taken in a particular βbadβ 1-year period in the past, such as Sept 2008 to Sept 2009, instead of the most recent 1-year period.
β’ Stress period does not change frequently.
β’ t ranges over business days π, π + 1,β¦ , π + 251, for some T at the beginning of the stress period, rather than β252,β251,β¦ ,β1 (yesterdayβs close).
![Page 19: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/19.jpg)
Stress Scenarios
β’ Reflects a particularly bad interval in the past, just two times π‘1, π‘2 .
β’ Examples include 9/11, Lehman bankruptcy. β’ Usually computed separately for several different
scenarios.
β’ Compute βππ = ππ πβ²1,π, β¦ , π
β²ππ,π β
ππ π1,π, β¦ , πππ,π
β’ πβ²π,π = ππ,π + ππ,π π‘2 β ππ,π π‘1 absolute
β’ πβ²π,π = ππ,π β ππ,π π‘2 /ππ,π π‘1 relative
![Page 20: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/20.jpg)
Risk not in VaR (RNiV)
β’ Models for exotics usually depend on trader-marked parameters which are estimated and donβt change frequently
β’ Examples include correlation or correlation skew for basket options, mean-reversion in commodity markets, and correlation between spot and volatility
β’ Cross terms in delta-gamma approximation
![Page 21: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/21.jpg)
RNiV, continued
β’ Situations where shifting one driver causes another to move (extra terms in the chain rule, double-counting).
β’ Mismatch between driver sensitivity and actual historical time series.
β’ The regulators require us to keep track of sources of Risk not in VaR.
![Page 22: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/22.jpg)
Specific Risk (SR)
β’ Specific risk βpicks up where VaR leaves offβ, meaning that it attempts to capture rare events not likely in a 1-year period.
β’ Examples include huge moves in bond price, stock price or volatility caused by credit migrations, scandals, wars, drug non-approvals, etc.
β’ Specific risk is made up of Debt Specific Risk (DSR) and Equity Specific Risk (ESR)
![Page 23: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/23.jpg)
Debt Specific Risk
β’ VaR for sovereign, agency, municipal or corporate bonds and credit default swaps captures the risk from moves in the generic credit spread for that instrumentβs rating.
β’ DSR attempts to capture risk from the βidiosyncraticβ part of the credit spread, i.e. the difference between the credit spread for that CDS (or bond) and the contribution from the generic spread from that instrumentβs rating.
![Page 24: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/24.jpg)
DSR, Continued
β’ Run regression on each bond between overall spread and generic spread:
βπ π = π½πβππ + βπ πππππ
πΈ βπ πππππ = ππ
π . π. βπ πππππ = ππ
The index i is for the bond or CDS.
![Page 25: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/25.jpg)
DSR, continued
Then we average over all the bonds and over all the tenors to get a average volatility and mean for a single idiosyncratic spread. Then we simulate it by Monte Carlo and compute
500,000 P&Ls of the form ππ
ππ βπ ππππ. Then we
sort the total P&Ls and take number 5000.
![Page 26: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/26.jpg)
Equity Specific Risk (ESR)
β’ ESR attempts to capture events that may occur over a 10-year period, rather than a 1-year period.
β’ For each name, compute the 1-year historical vol and 10 year kurtosis.
β’ Create a random variable with that volatility and kurtosis β’ Use principal component analysis to correlate all the
different namesβ random variables. β’ Simulate 500,000 times for each name, and zero P&L
contributions out any returns less than 2.6 standard deviations
β’ Add all the P&Ls over all the positions, sort and take Number 5000 as the specific risk.
![Page 27: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/27.jpg)
ESR Model β A Simple Example
β’ There are N stock indices, and each index has a return of the form
ππ = π ππ,1π1 + ππ,2π2 +β―+ ππ,πππ
ππ~π 0,1 , all independent
πΈ π = 1, πΈ π2 > 1, independent from Zβs πΎπ’ππ‘ ππ = 3(πΈ π
2 β 1)
The Zβs represent the principal components of the market.
![Page 28: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/28.jpg)
Individual Stock Treatment
ππ = πππΌ + ππ ππ ππ π, π, ππΌ πππ πππππππππππ‘
Let ππΌ2 = π£ππ ππΌ
Then π£ππ ππ = πππΌ2 + ππ
2 πΎπ’ππ‘ ππ
= 3π4ππΌ4 πΈ ππΌ
2 β 1 + ππ 4 πΈ ππ
2 β 1
π4ππΌ4 + 2π2ππΌ
2ππ 2 + ππ
4
![Page 29: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/29.jpg)
Changes in Volatility
We also compute the average implied volatility change, and model it as
Ξ(ππ£) = π ππ
ππ+ 1 β π2 πππ£πππ£
where Ο is the correlation between implied volatility and equity index return. Once again, we choose random variable πππ£ to have
πΈ πππ£ = 1, πΈ πππ£2 > 1.
![Page 30: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/30.jpg)
P&L Contributions
β’ We only keep βlarge enoughβ P&L contributions from either price or volatility
β’ Price: βπ = π π 1 + ππ , β¦ β π(π,β¦ ) if ππ > 2.6 β π . π. (π), 0 otherwise. k is the path
index. Calculated by grid approximation.
β’ Implied Volatility: βπ = π β¦ , πΌπ + β πΌπ ,β¦ βπ(β¦ , πΌπ, β¦ ) if β πΌπ > 2.6 β π . π. (β πΌπ ), 0 otherwise.
β’ There are both 1-day and 10-day versions of ESR.
![Page 31: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/31.jpg)
ESR Shortcomings
β’ Difficult to determine the best threshold for defining βeventβ risk (we use 2.6 standard deviations, but this is hard to calculate).
β’ Returns and changes in implied volatility follow a distribution which has kurtosis, but not skew. Difficult to allow for skew in this general framework.
![Page 32: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/32.jpg)
Incremental Risk Charge (IRC)
β’ Aims to capture impact on the credit book (bonds and CDSs) from migrations and defaults, which are not included in VaR or DSR.
β’ Create up to 3 N(0,1) random variables based on market, regional and sector returns.
β’ Use these to create N(0,1) random variables for each name, with the desired correlation matrix.
β’ Default or migration for each name can occur in 1 year if the variable breaches any of a set of thresholds.
![Page 33: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/33.jpg)
IRC, continued
β’ Recovery is a function of each nameβs random variable
β’ Each simulation is a single 1-year drawing for the underlying variables
β’ For each position, compute a P&L only if a migration or default occurs (most are 0).
β’ Add up total P&L, sort, and choose number 500 out of 500,000 simulations (99.9% confidence level, rather than 99%)
![Page 34: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/34.jpg)
Some recent and on going regulatory requirements and modeling projects
β’ Incorporate changes in volatility skew and convexity as a function of strike as risk factors in Equity, Interest Rate, FX and Commodity VaR.
β’ Improve internal pricing models used for VaR for bonds.
β’ Eventually move toward full revaluation rather than delta/gamma and grid approximations.
![Page 35: Quantitative Risk Management: VaR and OthersΒ Β· Quantitative Risk Management: VaR and Others Roy E. DeMeo, Jr. ... β’Total P&L across all positions βπ = ... forward swap rates,](https://reader035.vdocument.in/reader035/viewer/2022062600/5ae7c70e7f8b9a3d3b8efccc/html5/thumbnails/35.jpg)
Ongoing projects, contβd
β’ Incorporate dividend risk in Equity VaR.
β’ Compute CVA in a manner consistent with front office models and then compute CVA VaR (this is extremely difficult).
β’ Compute risk for private equity positions
β’ Estimate impact of Risk Not in VaR by varying marked parameters.