If you think about your trading system as a production machine, you can apply the principles of quality management and statistical process control in order to improve the quality of the output.
The work of Deming, Juran, Ishikawa come to mind.
Lean 6 Sigma is the most prominent current example of this line of thinking.
It's the reason why Toyota is dominating and will continue to dominate the auto industry.
Their principles of system design and process management are industrial grade, and worth understanding and also are applicable to wide variety of efforts where you care about repeatability and predictability.
Their work shows that you cannot make meaningful predictive change to a system until that system is under control.
Being under control means that you have reduced variation from systemic causes to a manageable level.
When you have a system that is under control you can then begin to make changes to see what the effect of those changes may be, and through study and analysis you may then move towards improved performance.
The use of averages, statistically significant sample sizes, and standard deviations are essential tools in bringing that kind of understanding and control to a production system (and by extension, to a trading system).
The difference between a trading system and a production system is that you don't mind having variability on the upside end (your gains) and so you really are striving in this mental model to gain control over your losses.
In a production system you want tight performance around the mean.
In a trading system you want tight performance around the mean for losses, with unbounded upside gains.
Consider that an engineer's way of saying "control your losses, let your profits run".
Why would you want a system that is under control, by this definition? Why would you put this kind of engineering effort into examining your trading system? Because you have a belief that your system is robust, and therefore will have a lot of opportunity going forward, and that through effort you can gain predictability in your performance which does a number of things for you: 1) you can make a sensible business decision on how much capital to apply to that system based on an expected outcome 2) you can make a sensible business decision on how much risk is appropriate given the amount of variation in losses you have experienced 3) 1 and 2 give you a range of risk parameters that you can use to achieve your objectives 4) You have a benchmark of expected performances that also serve to inform you on whether or not your system is still in control or if you have experienced a set of results that tell you its time to stop trading.
On the production floor, if you have a series of errors (bad parts) that exceed the performance standards you stop the line and conduct causative research to find the reason.
If your system is not under control though, you have no idea if it was your system or an outside environmental change.
Therefore the first thing to do is Reduce Variability.
Averages and standard deviations are the means by which you determine if this is happening.
These are among the simplest reasons why someone might put the effort into understanding and using these tools and procedures.
If you are operating a trading system with the exclusive performance measure of bottom line, you cannot establish the reason for a change in performance other than the same gut-feel that is operating the system.
For an engineer, this would be considered "out of control' in the precise sense that this is a system that performs in ways we cannot predict and therefore cannot engineer the appropriate amount of risk.
In other words, it would be a novelty, but not a subject for serious consideration
The work of Deming, Juran, Ishikawa come to mind.
Lean 6 Sigma is the most prominent current example of this line of thinking.
It's the reason why Toyota is dominating and will continue to dominate the auto industry.
Their principles of system design and process management are industrial grade, and worth understanding and also are applicable to wide variety of efforts where you care about repeatability and predictability.
Their work shows that you cannot make meaningful predictive change to a system until that system is under control.
Being under control means that you have reduced variation from systemic causes to a manageable level.
When you have a system that is under control you can then begin to make changes to see what the effect of those changes may be, and through study and analysis you may then move towards improved performance.
The use of averages, statistically significant sample sizes, and standard deviations are essential tools in bringing that kind of understanding and control to a production system (and by extension, to a trading system).
The difference between a trading system and a production system is that you don't mind having variability on the upside end (your gains) and so you really are striving in this mental model to gain control over your losses.
In a production system you want tight performance around the mean.
In a trading system you want tight performance around the mean for losses, with unbounded upside gains.
Consider that an engineer's way of saying "control your losses, let your profits run".
Why would you want a system that is under control, by this definition? Why would you put this kind of engineering effort into examining your trading system? Because you have a belief that your system is robust, and therefore will have a lot of opportunity going forward, and that through effort you can gain predictability in your performance which does a number of things for you: 1) you can make a sensible business decision on how much capital to apply to that system based on an expected outcome 2) you can make a sensible business decision on how much risk is appropriate given the amount of variation in losses you have experienced 3) 1 and 2 give you a range of risk parameters that you can use to achieve your objectives 4) You have a benchmark of expected performances that also serve to inform you on whether or not your system is still in control or if you have experienced a set of results that tell you its time to stop trading.
On the production floor, if you have a series of errors (bad parts) that exceed the performance standards you stop the line and conduct causative research to find the reason.
If your system is not under control though, you have no idea if it was your system or an outside environmental change.
Therefore the first thing to do is Reduce Variability.
Averages and standard deviations are the means by which you determine if this is happening.
These are among the simplest reasons why someone might put the effort into understanding and using these tools and procedures.
If you are operating a trading system with the exclusive performance measure of bottom line, you cannot establish the reason for a change in performance other than the same gut-feel that is operating the system.
For an engineer, this would be considered "out of control' in the precise sense that this is a system that performs in ways we cannot predict and therefore cannot engineer the appropriate amount of risk.
In other words, it would be a novelty, but not a subject for serious consideration
SHARE
