Saturday, June 20, 2015

Risk Decomposition Tool[P&L Attribution]

Portfolio Manager's Wish:-

        Every portfolio manager wishes to construct predictable & manageable portfolios that are less risky, profitable & with less investment. This is going to be a uphill task given that many types & number of instruments are present in the portfolio. In this post we have demonstrated tools which helps in improving predictability.

Anecdote:- In real life Best Friend[Portfolio Manager] knows the possible reasons for your[Portfolio]Mood Swings.What would have made you feel down :).

Decompose Portfolio Risk Why ? Rationale behind this!!!

      Keeping track of portfolio risk becomes cumbersome when number of instruments increases. With the help of P&L Attribution or P&L Explain Portfolio Risk can be managed easily as it attributes or explains daily fluctuation in value of a portfolio of trades to the root cause.

            P&L Attribution in One Line!!

                        Portfolio P&L = P&LExplainedByRiskFactors + UnExplainedP&L


            How? P&L Attribution!![Choose=>Map=> Aggregate]

                      Instead of trying to determine Risk of each instrument, decompose these instruments into 
               primitive instruments which is used to calculate Regulatory Capital also.  Overview:-  
            
               1) Choose     : Set of elementary risk factors & estimate their probability distributions.
                2) Map          : Find common RiskFactors among positions in a given portfolio. Map 
                                      financial instruments into exposures on these risk factors. 
                            [i.e.Each Position is marked to market,value of each instrument is allocated to risk factors.]

                3) Aggregate  : Aggregate the exposure for all positions & build distribution of P/L on portfolio. 

            RiskFactor should be a principal determinant of change in value of a transaction used for quantification               of risk.Usually, Factors that are deemed relevant for pricing should be included as Risk Factors.            

           Factors to be considered:-
              Based on type of Instrument Asset Class, you can decide which RiskFactors to be used. Guidelines                   below(not exhaustive though):-




Asset ClassRisk Factors

Interest Rates1) RiskFactor corresponding to each maturity segment of yield curve.

Forex Rates1) RiskFactors corresponding to individual foreign currency in which bank's
   currency are denominated.
2) RiskFactors corresponding to exchange rate risk.


Equity Prices1) RiskFactors capturing market—wide movements in equity prices
   (e.g S&P 500,..etc).
2) RiskFactors capturing sector-wide movements in equity prices          (e.g.Manufacturing,Pharma.etc).
3) RiskFactors corresponding to volatility of individual security.


Commodity Price1) RiskFactors corresponding to each of commodity markets in which banks hold significant position.

2) For banks with limited exposure following are acceptable:-
           a) One RiskFactor for each commodity price.
                                    (or)
           b) For small position, it might be acceptable to use single 
RiskFactor for a                      relatively broad sub-category of commodities.

3) RiskFactors should account variation in “”convenience yield” between derivative positions.


Credit1) RiskFactor to incorporate spread risk(eg between bonds & swaps).

         Example:-
         See below an example of how instruments are mapped to RiskFactors and their exposure measures.(More detailed in the next post).







RiskFactors
Instrument
Change in Interest Rate Change in Equity Index Price
Fixed Income Security Modified Duration
Convexity


Equities
Beta=1.5





           Change in Equity Index Price has no impact on Fixed Income Security.
           Increase in Equity Index Price has corresponding positive impact on Equities price.(+1.5)

          Effectiveness Metrics:-


             To determine whether risk factors included in captures material drivers of bank's actual P&L                              effectively,use:-

                  1) Average  P&L left UnExplained as a unit of Dispersion of Actual P&L.  

                                       Mean of UnExplained P&L[Predicted P&L - Actual P&L ]
                                       ----------------------------------------------------------------------------------
                                                      StandardDeviation of Actual P&L.
                                           

                  2)Variance of P&L left UnExplained as a unit of Variance of Actual P&L. 
                                       Variance of UnExplained P&L
                                       ---------------------------------------------
                                        Variance of Actual P&L

             Both measures quantify the impact of Actual P&L dispersion on P&L Attribution Accuracy.              

        Advantages:-

                  1) Assist the risk manager in evaluating positions whose characters may change over time.
                  2) Assist the risk manager in evaluating positions even in absence of long history.
                  3) Assist the risk manager in  evaluating positions by reducing number of variables.

        Conclusion:-
                RiskFactor mapping is an effective way in managing & measuring portfolio risk. This effectiveness            can be felt when the portfolio size is bigger, portfolio contains many types of instruments, instruments                 having unique behaviors...etc.

        To Be Continued....:-             
               In the next post we will delve deep into factor exposures and see how to use those factor exposures           in quantifying risk,predicting returns,,..etc. How to predict the portfolio risk in a better way using the                   exposures, riskfactors,.etc.

Appendix:-
      Risk-Theoretical P&L is defined as the daily desk-level (hypothetical) P&L that is predicted by the risk management model conditional on a realisation of all relevant risk factors that enter the model.

Friday, June 12, 2015

Not Only Risk Optimization Tool [Instrument Score Card]

Portfolio Manager's Wish:-

        Every portfolio manager wishes to construct optimum portfolios that are less risky, profitable & with less investment. This is going to be a uphill task, In this post we have demonstrated Portfolio Risk Optimization processes & tools.


Optimize Portfolio Risk How to? Tools Used

Portfolio Manager can optimize[Reduce Risk, Improve Returns] portfolio by doing:-

               1) Conscious allocation of economic capital.
               2) Trade off risk and return.

Some of the tools available are ComponentVar , MarginalVar, IncrementalVar, and other VaRs for that matter. But one drawback of VaR is it is a single number focusing on risk only and even worse it is is not even a coherent risk measure.

Optimize Portfolio Risk How to? Instrument ScoreCard[Not Very New!!!]
                When trying to optimize a portfolio, by either including or excluding an instrument portfolio manager can make use of below table for taking an informed decision. This table can be used as a framework for arriving at a score for given instruments(taking into consideration trends,risks,exposures,funding,...etc).

Given multiple instruments, manager can derive the score for those instruments and then choose the one with high score[Usage explained after the table].







Category SubCategory Effect Weightage

Funding





Available Capital + w1


Interest Rates # w2


Transaction Cost - w3

Protection





Protection Cost - w4


Hedging Cost - w5


Hedging Benefits + w6

Trend





Seasonality Trend # w7


Market Trend # w8

Exposure





Soverign Exposure # w9


Counterparty Exposure # w10

Organization Goal





Alignment + w11







Situation





Demand for Instrument + w12


Demand for Sector + w13


Predicted Future Demand for Instrument + w12


Predicted Future Demand for Sector + w13

Strength





WorstCaseScenarioPerformance + w14


BestCaseScenarioPerformance + :


AvgCaseScenarioPerformance + :


NormalCaseScnearioPerformance + wi

Usage:-






       1) Assign Weights to each factor with restriction of (w1+...+wi=1).
            [Note:- This is extensible framework,i.e. you can have as many factors as you want ]
            [Guidelines:- Don't have so many, then it will be maintenance nightmare].

       2) Calculate Instrument Score by
                 Effect * Weightage * Instrument's Score
             [Note:- Keep Instruments Score on a scale 1 to 5 for each factor]
     
       3) Rank the instruments using the score into categories[HighlyRequired,Required, NotRequired].
             [Note:- This is extensible framework,i.e you can have as many categories as you want.]
             [Guidelines:- Don't have so many, then it will be maintenance nightmare].

       4) Make investment decisions using the investment Category.
             [eg:- If an instrument is in NotRequired Category, then see whether it is going to become less                       worthy for sure, then buy a Put Option :)]

Conclusion:-

Using the Instrument Score Card Approach we can get holistic view of an instrument's impact on the portfolio rather than focusing only on Risk. But can we improve on this still, yes i think so. Watch out this space for new insights into this.

Thanks & Regards,
Srinivasan      

Sunday, June 7, 2015

Risk Optimization tools

Portfolio Manager's Wish:-

        Every portfolio manager wishes to construct optimum portfolios that are less risky, profitable & with less investment. This is going to be a uphill task, In this post we have demonstrated Portfolio Risk Optimization processes & tools.

Optimize Portfolio Risk How to? Process

Portfolio Manager can optimize[Reduce Risk, Improve Returns] portfolio by doing:-

               1) Conscious allocation of economic capital.
               2) Trade off risk and return.


Optimize Portfolio Risk How to? Tools are:-






Tool Measures Helps in deciding

Incremental  VaR Effective Portfolio Risk, after adding a new instrument. Whether to add a new instrument to portfolio?

Marginal      VaR Effective Portfolio Risk, after increasing unit investment in existing instrument. Whether to invest more in existing instrument?

Component VaR Effective Portfolio Risk, decomposed between instruments. Whether to retain/drop an instrument?




Optimization Tools: In Detail

Marginal VaR


Measures change in portfolio VaR resulting from adding additional dollar to an instrument(ith).     

Marginal VaR of ith component is  ∆VaRi  =   α Ïƒip/σp  =(VaR/P * Bi)

Component VaR

Measures VaR contribution of every instrument to Portfolio VaR.

Component VaR of ith component is  ∆VaRi  =  (VaR * Bi * wi)

Incremental VaR

Measures VaR contribution of a new instrument to Portfolio VaR.

Incremental VaR of NewInstrumenta = VaRp+a - VaRp


Optimization Tools: Implementation

Following Steps are followed:-

         1) Calculate Instrument Statistics[Return,StdDev]
         2) Calculate Portfolio    Statistics[Weighing,Return,StdDev]
         3) Calculate Related      Statistics[Betas].
         4) Calculate VaR           [Marginal VaR, Componenet VaR, Incremental VaR]

Instrument Return:-


InstrumentsValue(past value)
I1I2I3I4I5I6I7I8I9I10
01-Jan-20012.32323.31213323.31213
02-Jan-20011.45.12.61.832.121.62.61.832.121
03-Jan-200121.62.13.34.12.23.411.52.212.41.5
04-Jan-200121.1121321.1223121.1223119
05-Jan-20012.152.123.2112.512.213.122.152.2113.12.34
06-Jan-200111.9521.2213.1111.9512.2231.1112.9521.213.211.23
                                      From the historical returns obtain returns for every date.
                                       
InstrReturns
I1I2I3I4I5I6I7I8I9I10
01-Jan-2001..........
02-Jan-2001-0.39131.55-0.13333-0.922741.6750.661538-0.13333-0.922741.6750.61538
03-Jan-200114.428-0.588230.2692301.277777-0.93146-0.842593.4230760.222222-0.61370-0.9285
04-Jan-2001-0.0234.7142852.9393934.146341598.1176470.83478291.511.6666
05-Jan-2001-0.898-0.82333-0.753077-0.407109-0.899545-0.89935-0.89810-0.89954-0.57741-0.8768
06-Jan-20014.558139.009433.084112-0.0447644.5294118.9711535.0232558.597285-0.75496-0.4743
            do i = 1 to 10;
                 CurrValue     = InstrumentsHistory [,i];
                 PreValue       = Lag(InstrumentsHistory [,i],1);
                 Returns         = ((CurrValue-PreValue)/PreValue);
                 InstrReturns  = InstrReturns||Returns;
           end;
                 Portfolio Returns:-
PortfolioWeight
W1W2W3W4W5W6W7W8W9W10
P100.10.20.40.300000
P20.2000.40.20.10000.1
P3000.20.40.300.1000
P40.30.10.20.4000000
                         PortfolioReturn= Return(PortfolioValue=(PortfolioWeight*Instrument))
                                                     
PortReturns
P1P2P3P4
01-Jan-2001....
02-Jan-2001-0.170554-0.209743-0.194645-0.79926
03-Jan-2001-0.721441-0.410103-0.6307283.1428571
04-Jan-20014.94321772.20174173.80535281.0656285
05-Jan-2001-0.653875-0.708885-0.669671-0.649812
06-Jan-20011.02269591.00062290.89500311.0158388
                  do i = 1 to 4;
                       PortfolioValue    = (PortfolioWeight[i,]#InstrumentsHistory)[,+];
                       PrePortfolioValue = Lag(PortfolioValue,1);
                      CurrPortReturns   = ((PortfolioValue-PrePortfolioValue)/PrePortfolioValue);
                      PortReturns       = PortReturns||CurrPortReturns  ;
                 end;
                               
                              PortStdev=Sqrt(Var(PortfolioReturn))
PortStdev
P1P2P3P4
2.37439831.21052731.87823711.6012189
Portfolio Beta Calculations:-

                              PortfolioBeta= Corr p,i * (InstrStdDev/PortStdev)

PortBeta
I1I2I3I4I5I6I7I8I9I10
P1-0.2724921.39315480.45491350.25711371.76470821.65986380.20999071.29739670.1034512.0616687
P21.60344213.07419111.18972480.84052953.11526063.66272181.02971972.96892030.10145743.5945565
P31.03342141.98132190.76677980.54172282.00779122.3606310.66365631.91347460.06538952.3166983
P43.23255490.25823330.56214771.1437981-0.2616830.30767010.93416622.2445145-0.383511-1.019062

Measure of  relationship between Portfolio loss & Instrument loss.


Let's Calculate VaR:-

MVar:- 


PortMVar
I1I2I3I4I5I6I7I8I9I10
P1.0.25078880.08189130.04628430.317674.....
P20.1510749..0.0791940.29351710.3450984...0.338676
P3..0.11649230.08230070.3050318.0.1008254...
P40.39484540.03154230.06866440.139711......
            For Portfolio P1 investing in I4 is less risky.

ComponentVar:-

PortCVar
I1I2I3I4I5I6I7I8I9I10
P1.0.33079040.21602920.24419621.257036.....
P20.3882021..0.40699360.75422160.4433825...0.4351309
P3..0.28803890.40699361.1313324.0.1246504...
P41.55280840.04134880.18002430.7325885......

            For Portfolio P1 more risk contribution is done by I5 and less contribution is done by I3.

Incremental VaR:-

          Full Revaluation should be used to calculate this (Varp+a - Vara)

Conclustion:-

Using these three VaR measures portfolio manager can decide on optimizing portfolio by reducing risk. But this is only one end of the coin :). Follow this space to do it more effectively.