To me, the idea of placing too much emphasis on benchmarking in the short term is ill founded and adds to informational noise which in turn causes irrational behaviour. Trying to magically outperform something that is uncontrollable over the short term kind of signals to me that there are some serious misunderstandings in what is trying to be achieved. In the short term it is inevitable that we may underperform the market and it should be expected. Until we can accept this can we then, and only then start to understand how to beat the market. As I have written before, an investment process is one which works over the long term by the fact that the process is weighing the probabilities in the user’s favour. As with the coin flipping example in the “Process vs Outcome” post, it is only once we give the process a chance can we start to realise the probabilities associated with the process - over the long term.
Thursday, 27 December 2007
Indexes
When thinking about performance, it’s interesting to note what we are trying to compete against. In most cases our benchmark is some kind of index or basket of indexes. So how are the constituents of an index chosen? For example the FTSE All Share index is basically a market capitalisation weighted index of the largest 350 stocks trading on the London Stock Exchange. There is low turnover and no human decision. The S&P 500 is determined based on several non human factors such as Liquidity, Four Quarters of Positive Net Income, Market Cap, Sector Representation and Lack of Representation. However, despite their simplicity the majority of fund managers have trouble beating these benchmarks. Which leads me t ask, are we doing something wrong and if so what is it?
To me, the idea of placing too much emphasis on benchmarking in the short term is ill founded and adds to informational noise which in turn causes irrational behaviour. Trying to magically outperform something that is uncontrollable over the short term kind of signals to me that there are some serious misunderstandings in what is trying to be achieved. In the short term it is inevitable that we may underperform the market and it should be expected. Until we can accept this can we then, and only then start to understand how to beat the market. As I have written before, an investment process is one which works over the long term by the fact that the process is weighing the probabilities in the user’s favour. As with the coin flipping example in the “Process vs Outcome” post, it is only once we give the process a chance can we start to realise the probabilities associated with the process - over the long term.
To me, the idea of placing too much emphasis on benchmarking in the short term is ill founded and adds to informational noise which in turn causes irrational behaviour. Trying to magically outperform something that is uncontrollable over the short term kind of signals to me that there are some serious misunderstandings in what is trying to be achieved. In the short term it is inevitable that we may underperform the market and it should be expected. Until we can accept this can we then, and only then start to understand how to beat the market. As I have written before, an investment process is one which works over the long term by the fact that the process is weighing the probabilities in the user’s favour. As with the coin flipping example in the “Process vs Outcome” post, it is only once we give the process a chance can we start to realise the probabilities associated with the process - over the long term.
Tuesday, 25 December 2007
Process vs. Outcome
Enterprise software development is a discipline in which process is often a deciding factor between success and failure. In my experience, more often than not projects run over time and over budget which I would put down primarily to one thing, focussing on the outcome of the project rather than focussing on the process of achieving the outcome. Through desperation, most managers falsely believe that adding more developers will help. Instead deadlines slip further. Getting developers to work longer. Instead deadlines slip further. Emergency measures whatever they may be. Yet deadlines still slip further. In other words, most software projects are conquered with brute force rather than careful thought and finesse. Instead of concentrating on eating the carrot, why not first work out how to get the carrot.
As anyone who has thought about software development for what it is can tell you, the complex relationships between understanding client requirements and technical complexity are difficult. Only when the actual process of the software development process is addressed are development teams enabled to start building malleable and flexible software that meets business and technical requirements. What needs to be understood is how software that is malleable and flexible in its design is cheaper to maintain, easier to build and less problematic overall. Only with a disciplined process can this result be achieved. Unfortunately, the brute force approach nearly always creates software that is late, over budget, barely satisfies the business requirements. Moreover, a long term legacy of difficult and expensive maintenance is born. This approach of focussing on the outcome rather than the process behind the final outcome comes down to our need for instant gratification regardless of long term results. It’s the feel good factor that drives us. When projects are faced with challenges a well defined process does not instantly lead to better results but in aggregate it does and this is why most people continue to carry on with the brute force approach. It’s the average speed, not maximum speed that counts.
To me, it is pretty clear that investing is not much different from software development in its need for process. Process brings to the table a set of disciplines that are not available with a “gut feeling” approach. To consistently repeat logical decisions a well defined process must be in place. Sustainability of execution is the all important. However, it is important to realise that even with a “superior” process we will not be guaranteed against failure. What we will see is a lower chance of failure. We are just moving the odds further into our favour. Over the long term, we will see an stronger aggregate result.
We are seriously limiting our chance of success by focussing on the short term. As investing is a probabilistic endeavour it can not reasonably be expected that a process will lead to a desired result every time. Coin tossing is a useful analogy of how long term results revert back to a process mean. A toss of a fair coin obviously has a 50% chance of heads and a 50% chance of tails. If coin is flipped twice is this going to be an indicative of these probabilities? The chances of flipping two heads or two tail are pretty high (25%). However the more times the coin is flipped the closer our results will be 50% heads, 50% tails. This is the mean of our process and over an extended set of outcomes it can logically be assumed. In other words, this is “the law of large numbers”.
An investment process is exactly the same. If you only give it a limited chance to work then the chances that you will be disappointed are high. You are not moving the odds into your favour because you are forgetting about the law of larger numbers. Accepting that there will be periods of under performance and that the aggregate results are what count. There is no such thing as “the law of small numbers”.
As anyone who has thought about software development for what it is can tell you, the complex relationships between understanding client requirements and technical complexity are difficult. Only when the actual process of the software development process is addressed are development teams enabled to start building malleable and flexible software that meets business and technical requirements. What needs to be understood is how software that is malleable and flexible in its design is cheaper to maintain, easier to build and less problematic overall. Only with a disciplined process can this result be achieved. Unfortunately, the brute force approach nearly always creates software that is late, over budget, barely satisfies the business requirements. Moreover, a long term legacy of difficult and expensive maintenance is born. This approach of focussing on the outcome rather than the process behind the final outcome comes down to our need for instant gratification regardless of long term results. It’s the feel good factor that drives us. When projects are faced with challenges a well defined process does not instantly lead to better results but in aggregate it does and this is why most people continue to carry on with the brute force approach. It’s the average speed, not maximum speed that counts.
To me, it is pretty clear that investing is not much different from software development in its need for process. Process brings to the table a set of disciplines that are not available with a “gut feeling” approach. To consistently repeat logical decisions a well defined process must be in place. Sustainability of execution is the all important. However, it is important to realise that even with a “superior” process we will not be guaranteed against failure. What we will see is a lower chance of failure. We are just moving the odds further into our favour. Over the long term, we will see an stronger aggregate result.
We are seriously limiting our chance of success by focussing on the short term. As investing is a probabilistic endeavour it can not reasonably be expected that a process will lead to a desired result every time. Coin tossing is a useful analogy of how long term results revert back to a process mean. A toss of a fair coin obviously has a 50% chance of heads and a 50% chance of tails. If coin is flipped twice is this going to be an indicative of these probabilities? The chances of flipping two heads or two tail are pretty high (25%). However the more times the coin is flipped the closer our results will be 50% heads, 50% tails. This is the mean of our process and over an extended set of outcomes it can logically be assumed. In other words, this is “the law of large numbers”.
An investment process is exactly the same. If you only give it a limited chance to work then the chances that you will be disappointed are high. You are not moving the odds into your favour because you are forgetting about the law of larger numbers. Accepting that there will be periods of under performance and that the aggregate results are what count. There is no such thing as “the law of small numbers”.
Friday, 7 December 2007
Moving From Target Returns to Expected Returns
Investing is really just a probabilistic endeavour not unlike gambling. Lets think about that for a moment. Like gambling, investing has a set of expected outcomes and like gambling, the only certainty is that there is no certainty. However, despite this lack of certainty we need to act.
For precise prediction, we rarely have the required information and by adding more information we do not necessarily obtain more accurate results. To counter this we need to look at decision making as a set of probabilistic outcomes. By calculating a probability distribution we can arrive at an expected value which takes into account probable chances of loss. In other words the chance of an adverse result. In total this will enable us to calculate an expected return rather than a target return. This is no different from producing a decision tree of probabilistic outcomes.
By expressing opinions in expected value terms we are admitting that there may be a chance of a negative result. By reviewing all of the scenarios (good, bad and neutral, etc) we can start to think of the payoffs and the probabilities of those payoffs. It removes the risk of focusing too much on particular scenarios (often the positive ones). In behavioural finance this is known as "anchoring"- if we start to think of the target return for of a certain stock we start to look for evidence that will support that target whilst dismissing contrary information.
By considering multiple scenarios, we are enabled psychologically to consider all the available information. This allows us to enter into the investment position with the idea that there may be an unfavourable result. In other words, we can be wrong without a fear of failure.
So what does all of this mean? Lets take a look at an example.
Stock exceeds earnings target. 25% probability. Stock rises by 3%
Stock meets earnings target. 50% probability. Stock rises by 1%
Stock misses earnings target. 25% probability. Stock falls by 4%
(0.25 x 3% + 0.50 x 1% + 0.25 x -4%) = 0.25%
We can see here that our expected return is not really all that favourable, even though the probability is clearly in favour of a positive result. Lets look at an even more illustrative example.
Stock exceeds earnings target. 25% probability. Stock rises by 3%
Stock meets earnings target. 65% probability. Stock rises by 0.5%
Stock misses earnings target. 10% probability. Stock falls by 10%
(0.15 x 3% + 0.65 x 0.5% + 0.20 x -10%) = -1.225%
This is clearly a bullish outlook but the expected return is negative. So what should we do? Short sell?
Now lets look at a typical stock which is facing some problems. Each piece of bad information gives us a small reduction in the price but any positive information has a big impact (upwards) on the price. Lets look at an example:
Stock exceeds earnings target. 25% probability. Stock rises by 15%
Stock meets earnings target. 50% probability. Stock rises by 0.5%
Stock misses earnings target. 25% probability. Stock falls by 2%
(0.25 x 15% + 0.35 x -0.5% + 0.50 x -2%) = 2.575%
Again, the odds are not in favour of a positive outcome but the expected value is positive. Look past the frequency of success and start thinking about the expected value. It's not the frequency of being correct that matters; it's the magnitude of being correct that matters. These are simple examples just used to illustrate a point but remember that through investing, we are dealing in a probabilistic endeavour so it needs to be asked "are target returns really that relevant"?
For precise prediction, we rarely have the required information and by adding more information we do not necessarily obtain more accurate results. To counter this we need to look at decision making as a set of probabilistic outcomes. By calculating a probability distribution we can arrive at an expected value which takes into account probable chances of loss. In other words the chance of an adverse result. In total this will enable us to calculate an expected return rather than a target return. This is no different from producing a decision tree of probabilistic outcomes.
By expressing opinions in expected value terms we are admitting that there may be a chance of a negative result. By reviewing all of the scenarios (good, bad and neutral, etc) we can start to think of the payoffs and the probabilities of those payoffs. It removes the risk of focusing too much on particular scenarios (often the positive ones). In behavioural finance this is known as "anchoring"- if we start to think of the target return for of a certain stock we start to look for evidence that will support that target whilst dismissing contrary information.
By considering multiple scenarios, we are enabled psychologically to consider all the available information. This allows us to enter into the investment position with the idea that there may be an unfavourable result. In other words, we can be wrong without a fear of failure.
So what does all of this mean? Lets take a look at an example.
Stock exceeds earnings target. 25% probability. Stock rises by 3%
Stock meets earnings target. 50% probability. Stock rises by 1%
Stock misses earnings target. 25% probability. Stock falls by 4%
(0.25 x 3% + 0.50 x 1% + 0.25 x -4%) = 0.25%
We can see here that our expected return is not really all that favourable, even though the probability is clearly in favour of a positive result. Lets look at an even more illustrative example.
Stock exceeds earnings target. 25% probability. Stock rises by 3%
Stock meets earnings target. 65% probability. Stock rises by 0.5%
Stock misses earnings target. 10% probability. Stock falls by 10%
(0.15 x 3% + 0.65 x 0.5% + 0.20 x -10%) = -1.225%
This is clearly a bullish outlook but the expected return is negative. So what should we do? Short sell?
Now lets look at a typical stock which is facing some problems. Each piece of bad information gives us a small reduction in the price but any positive information has a big impact (upwards) on the price. Lets look at an example:
Stock exceeds earnings target. 25% probability. Stock rises by 15%
Stock meets earnings target. 50% probability. Stock rises by 0.5%
Stock misses earnings target. 25% probability. Stock falls by 2%
(0.25 x 15% + 0.35 x -0.5% + 0.50 x -2%) = 2.575%
Again, the odds are not in favour of a positive outcome but the expected value is positive. Look past the frequency of success and start thinking about the expected value. It's not the frequency of being correct that matters; it's the magnitude of being correct that matters. These are simple examples just used to illustrate a point but remember that through investing, we are dealing in a probabilistic endeavour so it needs to be asked "are target returns really that relevant"?
Subscribe to:
Posts (Atom)