Holmes, I similarly have a system that was great (on US and European indexes) and although it hasnt gone into a serious drawdown has reached the point where the risk of one outweighs the rewards. The profit driver is now too small for the risk of catastrophe and the last two months have given minor drawdown. In my case I understand the two elements very well (now) and have ceased trading it. But I also now know what it takes to move back into a good risk/reward position and will monitor that weekly. Perhaps you could do the same thing with your system. Is that possible?
Kiwi Trader, Thank you for the suggestion. After I came to understand why the system went south I undertook a long period of investigation on how to avoid it. I have created a new system which is far superior and subsequently there is no need to go back and try to resurrect the old system. Sherlock
Yes agree. You need to understand that each moment in the market (in life) is unique. New systems evolve that work now that wouldn't have worked at any other point in time. To say that there are a finite amount of systems and we will run out of them is a sophomoric notion of scarcity. I just listened to an interview with the great trader Marty Schwartz. His trading philosophy was you 1. find and edge. 2. exploit (trade) that edge until it stops working. 3. goto step one. If you understand that there is always an edge, you will find a new one much more quickly as your old ones fail to work.
AMT for SWA was my old job when I first started posting here...I left SWA at the start of the year (the airline industry is a mess....good riddins!). The systems I was talking about that are continuously trading week after week (separate systems for the ES, ER, EC, ZB, and ZN) are always performance reviewed/measured and adjusted if needed. What seems to work is to find a way to measure the current volatility of the market and make sure that value is reflected in how your profit targets and stops are set up for whatever instrument you are trading. With an "automated" trading system, the per trade probability of trade goal success must be linked to some acknowledgement of the current markets volatility. The goal of each trade and then the probability of achieving that profit goal (target) is directly tied to a proper representation of the markets current value (measurement) of volatility....this seems to make a very big difference in a continuation of robustness of said system. Also the placement of the protective stop and the use of any trailing stops must also be elastic in relation to a current value of measured volatility. I guess the reason that this seems to work so well is that the "volatility" is what is always changing in the market.....for periods of time it is high and then for periods of time it is low.....but it just keeps on changing in cycles.
If your data for backtesting includes the entire history of the specific market and only add new data without eliminating old data then yes, the amt of systems is finite. But if you constantly eliminate old data and add new data, then yes, there are an infinite amount of systems.
I think we have different definitions of what exactly a system is. Systems have nothing to do with old data or new data. Systems are methods of trading. Methods of trading are any set of behaviors one uses or tests in order to operate in the market. If you take 100 systems, that each have only two differences amongst them, when you adjust those differences (say you adjust average risk to a different number and some other number in a system (there are an infinite amount of numbers, I HOPE you agree with this statement), you now have (100X100-100)/2 numbers of systems. If you have three different characteristics you can adjust, now you have (100X100X100-300)/3 number of systems. As you increase the number of changes you have to any small set of systems, while the number of changes increases linearly, the number of new systems that result increases exponentially. Have you ever seen the graph of a hyperbola? Given the number of combinations and adjustments one could make to even a small number of systems, its not very useful to conclude that there are a limited or finite number of systems. And that mathematical fact is evident from starting with only a small number. This idea does not take into account systems with more differences than just 2 or 3 minor changes. Also, given that there are an infinite amount of numbers (YES< even between 34 and 35 there exist an infinite number of numbers), its false to assert that there are a finite amount of systems. You can BELIEVE there are a finite amount of systems, but that belief is not really in line with reality.
I was refering to systems that are profitable. That is finite. It has everything to do with the amount of data. If say u keep the data for years '95-'00 constant in ur backtesting, there are a finite amount of system you can find which would work during that period, no matter how much new data u add in.