Tungsten Risk Management

Tungsten Risk Blog

Nov 25 2011

Portfolio Slice and Dice

by Jens

Every die hard Tradar user knows that the three core components of a trade that describes a transaction in a portfolio consists of three simple data points : fund, strat, id.

This may seem very obvious to many, but bear with me, there is a lot more we can do with just those three components.

Fund = The Fund structure your trade belong to

Strat = The strategy your trade is allocated to - Example: If you are trading an Equity Pair strategy, say Carnival UK vs Carnival US you would naturally put the two trades into a strategy called something obvious such as Carnival

ID = The security ID, i.e. in our above example it could be CCL.US and CCL.CFD - the UK leg would probably be trading as a CFD.

The two legs of the pair, CCL.US and CCL.CFD would both booked to example: Fund = Lodestar and Strat = Carnival.

With these three core fields, we can define more information to make the slicing and dicing of risk and performance attribution clearer and more intuitive.

Here are two simple steps you should take for each trade you do:

1) Make sure you assign an appropriate risk matrix to each strategy you setup in Tradar. In our above example we could create a risk matrix with name=Holding Companies and riskSubCat=Pairs. When you create the Carnival strategy make sure it is assigned to the RiskMatrix you have defined for these type of pair trades.

2) Define a Portfolio and Subportfolio for your strategy - Example: if your firm has several trading teams you could call set Portfolio=RelativeValue and the subPortfolio=Mean Reversion to all strategies that are traded by the RelativeValue desk.

3) Assign a trader to each strategy - this is mandatory by the system, but normally you would assign the certified/head trader for each strategy. 

Be as creative as possible with the RiskMatrix and SubPortfolio/Portfolio structure. Tradar already comes pre-loaded with a rich set of categories for you to use, but you can always create your own. A well planned structure is going to reward you and your clients with a richer structure for risk and PNL attribution.

Once you have set it up properly, you will have access to these in your Tungsten risk system - allowing you to slice and dice your data in all imaginable ways.

Let us follow a drill down on one of our Tungsten demo portfolios. 

The image above shows the top fund level of the Tungsten fund (fictional). We have opted to display the Range Total (YTD PL), MTD Total, Day Change (all in fund base ccy which happens to be USD in this case). Act. StdDev is the actual or realised standard deviation given the YTD PL time series. Est. StdDev is the forecast standard deviation on current position universe. Sigma Event is Day Change / Forecast Risk (T-1), Max DD (maximum drawdown in the YTD period), Curr DD and Correlation against the risk buckets. In this case as grouping only contain one bucket, it is obviously 100% correlated to itself. We will see this change when we drill further down into the portfolio structure.

The image above show second level which we have selected to be Portfolio level. Here we can see the Correlation number is reflecting the correlation between the Portfolio's in the Tungsten fund. The EQ portfolio is contributing to the biggest loss both YTD and MTD.

A closer inspection of the details, we note that the actual standard deviation is almost double the estimated standard deviation in the EQ portfolio - the main reason being we have reduced the long equity beta exposure due to very tough markets conditions. The risk forecast picks this up so going forward we can expect our EQ portfolio to be less volatile than in the past (82,406 vs 149,232).

Likewise in the FX portfolio, we can see that the FX exposure has doubled (now 70,295 vs before 31,027), i.e. we can expect more volatility from our currency team going forward. In the Metals portfolio we have no positions hence the 0 Estimated Risk.

Next level down is the details in the FX portfolio, and as we can see we have only defined one RiskName: Currency Tactical - We could have defined several risk name categories here depending on the circumstances. Say your FX desk trade several core themes: Macro, Tactical, Hedge, you would then define and assign your strategies to these RiskNames. 

Further drilldown into Currency Tactical gives us the next level which we have chosen to be Strategy level in this example. This is the closest level before position level. On this quite detailed level we can see how the correlation and drawdown's are calculated for each bucket. So we can glean so much more when strategies are setup on fine detail such as this instead of throwing it into one strategy called FX (example).

The last level is position level and here we can see the performance and risk attributes of the EUR.CHF pair.

Note: Each level of drilldown is also showing a time series chart and return distribution.

In conclusion, spend some time and plan carefully your portfolio structure and you will get much more out of your systems. This will benefit your insight to your portfolio risk and P&L dynamics and also your clients when you send those weekly and monthly risk reports.

Good luck out there!

The Lodestar Team

Tags: , , , , , , ,


Comments are closed

Month List

QR Code

Tungsten Risk QR Code