Forum

Replies

• 

  Mike Benishek October 27, 2011 12:48

  Models are over-rated, especially "weather related risk" models.

  If they were accurate, no one would live in Florida......... :-)
  Mike Benishek, Pd.T.F.

  
  Lindsay Skardoon said:

  As a manager of portfolio's in particular fixed income I can say that I have some sympathy with the idea that risk management is insurance. In fact I cannot see how one can argue otherwise. Its a necessary insurance and any trader /PFM should undertsand its the essence that keeps him trading and his business alive. However I do believe that Risk Management has become unnecesarily complicated in a manner to give substance to Central Bankers and Academics.

  Risk systems work well as long as the market is not overly volatile as long as any deviation are within 3 SD's on the curve as long as those looking at risk understand the mathematics and how the model works and what the model CANNOT DO!!!!! Too often in the past I have heard that some outlier is responsible for an exaggerated loss. I have seen FI portfolios suffer loss because of gamma effects, and where the risk controllers took time to understand how that happened, because everything still fell within the bell curve... I have seen people marketing risk systems where the data is normalised and have them try to convince me that it remains a normal distribution curve. The problem with Risk is that most people can see it, can calculate as long as the risk is well behaved , but most cannot cope when we get movements outside 3 SD's. At this point every year some tradeable sector moves by more than 3 sd's.

  So in my view its much easier to keep it simple. Forget about the fancy maths, the  arbitrages that supposedly offset losses and lets just concentrate on have I made money or lost money. For me the simplest way to understand risk is if a security moves 20 bp adversely how much do I lose? Most models that I have seen never contemplate that something could move adversely as a pair. When things go wrong thats what usually happens. example.. Buy corporate sell Govvie. Corporate spread widens Government rallies. Which does our hero cut to fix his position.

  To me we have made our models way too complicated, its far easier to understand and control risk with a tight limit structure adherence and understand how much the trader loses if his exposure moves adversely. That way Banks/ Fund immediately know how much is on the table how much they stand to lose rather than calculate 2 nd derivative effects and allow those effects to take place before anything is done.

  Its unusual for a first order effect to blow up a company rather its the second order and the reason being is simply that those second order effects are expected to happen Normally when in essence they never do when a market is stressed.

  Those are my observations what do other people think?

   

   
• 

  Lindsay Skardoon October 27, 2011 9:11

  Quoting Eistein , and simple are two often diametrically opposed view. Yes his language was rather beautiful his analogies were quite delightful but given the complex nature of his work, that is probably the only way that most people could understand what he was talking about.

  Having spent my earlier life studying physics doing honours in laser physics and having to wrestle with a range of Quantum Mechanical conundrums I can assure you there was nothing simple in the maths nor the explanations, however that was his mark of genius, taking a complex problem and reducing it to something even a simpleton like myself could understand. Relatively easy .... not! which is why most people dont do physics....
• 

  Lindsay Skardoon October 27, 2011 8:54

  Andre the focus on P/L is all anyone cares about . An investor in my experience or for that matter a Global Head of trading never cared how money was made within reason rather their focus was how money was lost . Losing money kills a business. The other stuff at the end of the day is nice fuzzy stuff, like how did I make money, I am sure anyone who trades for a living can tell you immediately how they made the trade work. , really they care how you lost money nothing else and any explanation is an excuse why you lost. If you run money you would understand this sentiment.

  Of course one looks at how a trading view can be replicated, but you know what, once an opportunity has passed it has passed. On a trading viewpoint jobbing securities thats different, no matter how much analysis one does does not make you a better jobber.

  The reality for the market will be one day when the correlation traders have a bad day just like LTCM, where most correlation traders are like LTCM on roids. Yes one can replicate and see and demonstrate that a security behaves in a certain except when a large trade comes through and then its all over folks. If you have ever traded Kiwi bonds you would understand. That market can be doing everything that it needs to do when a large hedge fund or a bank decides it wants some fun and shunts the market one way or another and then comes back to pick off the carcass. A lot of trading is intuitive which is what the correlation traders attempt to remove, but you still require someone who knows what they are doing to prescribe a probabilty of an event.

  I agree getting the sectors correct, but I will say thats what gives one their p/l its not the other way around. I get p/l by understanding my market and making constructive bets that something is happening. For example not believing US economists and shorting the S&P , buy short bonds and sell $. Alternatively I look at my home market, look at what I thought was happening to Basel III wrt Banks and went long Tier 1 and bank sub debt / Insurance sub debt in names that I thought were undervalued. By doing this we managed to get returns for my fund yoy till june 30 at 23% and are still maintaining 16% yoy. That was done by taking a view, understanding my market and picking my sector.. any trader or fund manager does that, thats how they get their results, but everyone in my limited experience does some research and puts thought into the trade.

  Of course one thinks about how they lost money and how they avoid that in the future.The best loss is the first loss because thats the lesser. Remember traders are only as good as the last trade and their profitability. Otherwise why not evaluate traders on a different basis, or explain to the investor why he continually loses money. (By the same token , equally an investor who continuously sees outperformance or continous performance over long periods, should be extremely worried.Returns are Brownian in nature and never a one way bet, anyone who thinks otherwise is either an insider or dont invest. In fact as a prop trader in a large German Bank at the time it was always a pain when one made too much money because there was always the explanation how it occurred how one stayed within limits and how one did not overexpose the Bank.) Because no one can continuously generate performance.There are times when mistakes are made and when losses will be taken. We had our purple patch took a hit the last two months and will be positive (strongly ) this month assuming current closes. Its life, of course we try to minimise this but to say quant can do this is a ridiculous proposition. Look at trading theory .. remember you have the trader , the insider the fool. The insider only trades to his advantage, the trader wins sometimes and the fool never wins.. Markets are just like that, if a quant says they can do that then pay up because you have found the goose that can lay a golden egg.

  Correlation funds purport to do that but I have not sen too much information on what happens when things go amiss, I only hear about the good time results. Thinking maybe not so good currently given the returns in FX have been abysmal this year and that stuff is mostly correlation traded. QED.

  So maybe you can define what a controllable event is and how you made money for your clients.. I would be really interested, how do you prove that your assertion was correct, how did you back test that result that perhaps better result could be achieved, how was that optimised/

  I can think perhaps you run a matrix introduce some fast fourier transforms, overlay with a beta distributiuon or perhaps a bernoulli and then watch the results fall out ... so much more fun than monte simulations.. however at the end of the day investors care about their money , you dont grow rich losing money but you sure get wealthier making it.. I think sometimes we overcomplicate things and are rather too introspective. When trading  look for the next move, understand why your strategy will work, how much you lose if it does not, how to position, how to exit and when, and research the topic so that you can move quickly either way. So in a long winded explanation, my simple explanation had a lot more to it than meets the eye. Performance is difficult to continuously produce.
• 

  Andre Mirabelli October 27, 2011 7:46

  Lindsay,

  First, to quote Einstein, “Everything should be made as simple as possible, and no simpler.”

  Second, you say “lets just concentrate on have I made money or lost money.” (A) I would think that you might want to also include, as the rest of your comment implies, a version of this question that is concerned about what is best to do in the future, not only a phasing that focuses you only on the past. And (B), in regards to the past, it also might be helpful to learn HOW you “made money or lost money.”  That is why I am also dismayed by your total focus on evaluating the uncontrollable things that the market did to make or lose you money while your comments on their face thoroughly ignore any evaluation of the controllable things you did. Now maybe I did not understand all you intend, but that in itself can be a problem since I believe that a clear statement of the question is a necessary foundation for obtaining any useful insight.

  The work I do focuses upon answers to questions like “How did my decision to allocate sectors change my active return and IR?” Since I control my sector allocation and do not control sector returns, the answer to this question seems at least as useful to me as an answer to the question “How much did the returns of the sectors contribute to or bring about the return of my fund?”

  I do not think that ignoring such distinctions and the dimensionality of the issues involved will make things simple while remaining useful. Poorly founded complicated models are definitely a problem. But I do not think that implies that a simple one-dimensional approach (to a problem that itself is never even clearly stated) must be better.

   
• 

  Lindsay Skardoon October 27, 2011 2:31

  As a manager of portfolio's in particular fixed income I can say that I have some sympathy with the idea that risk management is insurance. In fact I cannot see how one can argue otherwise. Its a necessary insurance and any trader /PFM should undertsand its the essence that keeps him trading and his business alive. However I do believe that Risk Management has become unnecesarily complicated in a manner to give substance to Central Bankers and Academics.

  Risk systems work well as long as the market is not overly volatile as long as any deviation are within 3 SD's on the curve as long as those looking at risk understand the mathematics and how the model works and what the model CANNOT DO!!!!! Too often in the past I have heard that some outlier is responsible for an exaggerated loss. I have seen FI portfolios suffer loss because of gamma effects, and where the risk controllers took time to understand how that happened, because everything still fell within the bell curve... I have seen people marketing risk systems where the data is normalised and have them try to convince me that it remains a normal distribution curve. The problem with Risk is that most people can see it, can calculate as long as the risk is well behaved , but most cannot cope when we get movements outside 3 SD's. At this point every year some tradeable sector moves by more than 3 sd's.

  So in my view its much easier to keep it simple. Forget about the fancy maths, the  arbitrages that supposedly offset losses and lets just concentrate on have I made money or lost money. For me the simplest way to understand risk is if a security moves 20 bp adversely how much do I lose? Most models that I have seen never contemplate that something could move adversely as a pair. When things go wrong thats what usually happens. example.. Buy corporate sell Govvie. Corporate spread widens Government rallies. Which does our hero cut to fix his position.

  To me we have made our models way too complicated, its far easier to understand and control risk with a tight limit structure adherence and understand how much the trader loses if his exposure moves adversely. That way Banks/ Fund immediately know how much is on the table how much they stand to lose rather than calculate 2 nd derivative effects and allow those effects to take place before anything is done.

  Its unusual for a first order effect to blow up a company rather its the second order and the reason being is simply that those second order effects are expected to happen Normally when in essence they never do when a market is stressed.

  Those are my observations what do other people think?

   

   
• 

  Elmari van Deventer June 15, 2011 14:21

  I've heard:

  "Why do we need Risk Management, its not that important"

  and after arguing that point....

  "Yes, we know we have to manage our risk, but there is more important things to focus on"

  I gave up.  We will talk after their audit.

   

  and I've also heard from a very reliable source (from a member of board from an international company):

  "...this is an international company but their risk management is not up to scratch...actually.. I was shocked to see the minimal focus they have on risk management"  (Brian Henry- this is an opportunity for you :) will mail you.)  
• 

  Martin Davies June 3, 2011 7:08

  I fell off my chair laughing
  
  Trevor Levine (RISKCZAR) said:

  Two years after I was hired by the CFO to implement an ERM program, and after this same senior insurance company executive moved on to a new position, he told me, "Trevor, risk management is important; we just don't have time for it."
• 

  Martin Davies June 3, 2011 6:52

  Client : “Martin, we want to turn this risk reporting system off”

   

  Me : “Why, I have reviewed it and it seems quite spot on actually”

   

  Client : “The technical guru that set it up two years ago left the bank and no one understands what is on this screen”

   

  Me : “What?”

   

  Client : “Look we simply don’t know what these numbers mean and we have been doing fine so there is no point to any of this, please can you have it turned off so that we don’t have to sign off on it anymore”

   

  Me : “How have you been managing risk then?”

   

  Client : “We are low risk no measurement required”
• 

  Martin Davies June 3, 2011 6:45

  I was asked to review a hedging strategy of a firm a couple of years ago which was constructed using several swaps that were traded through back to backs …

   

  After reviewing the various cash flow legs of the trade I said “Gosh this back to back is wrong way and quite risky, we have a huge leg spread going on here and we are naked on the other side, we are exposed to be taken for a ride”

   

  The manager looked at me and said. “Wow leg spreads and being naked, I never knew risk could be so exciting how do I get onto this ride”

   

  I then realised what I had said outside the context of the domain of trading had been misunderstood entirely and as I removed my jacket to release the heat from the embarrassing misunderstanding a couple of the conservative people left the room.
• 

  Andre Mirabelli May 30, 2011 22:54

  "Even if the calculations are demonstrably incorrect, if the reports they produce do not appear patently absurd, we use them."