In my first experiences of backtesting systems I have noticed an interesting phenomenon in both cases. First of all, these were "idea first" systems which were created with the conscious intention of capturing certain classic behavior in intraday trading. What I mean to say, is that these were not "random" systems that just seem to work through trial and error. What I have noticed is that both systems tended to be more robust over the past couple years. They would both work well with a variety of parameters and the performance was regular and consistant. However, the further I go back, the more flippant both of the the systems would become. Ideal parameters would change often. There was more blowups. The performance was much less consistant and would range from great to terrible and back again (on a monthly basis). I am finding it difficult to set any parameters that would make the system work consistently back then even though the same kind of market behavior was in action. It seems like the intraday trading was implemented in a less consistant manner. It is as if the market is becoming more organized as more and more algorithmic trading comes into the market. For those of you who have been doing this a while, have you noticed the market becoming more and more condusive to systems trading? (These were both systems for oil futures, so perhaps the phenomenon is specific to that market)
In the futures market it has changed in both day trading and long term. Much of it I believe has to do with closure of the pits and other is so much more instruments which EFTs, options and foreign markets has caused much less volume. But also lack of knowledge. Those in the pits used certain methods, much more of either intraday Point and Figure or Pivot Point analysis, so many for years got use to ebb and flow of Indexes more controlled arb, either between New York/Chicago and S&P500/ES, also when price got to long term trend line, it almost bounced off by the tick or clearly broke through. There was much more new traders before as markets were not as watered down by developing "E-mini's", exchanges made the everyday guy be able to trade with much much less money. The pits are gone except for big S&P500 which never went to electronic, but the ebb and flow, seems much less New York/Chicago arbing replaced with HFTs, price seldom adheres to trendlines or previous pivots, and I believe the reasons are many are doing programming but never really learned to read a chart, they know generally just numbers. In past several years just on this forum alone many have written to me who either made some money by nuances during high volatility, spreading stocks or playing earnings reports ask after doing what they were doing up to 5 years, what to read/learn charts as what they were doing only loses now. Case in point, you are having problems with parameters changing, but you knew how to read charts well? Do you know what is changing and how to adapt much quicker? We trade people's emotions which is a form of risk, people don't stay the same, much based on what they watch or read on the news they become happy/unhappy, their emotions change, when people get scared, markets tend to go down with much volatility, at highs many are still buying but those who can read charts are feeding these traders as much as they want. Another area not spoke of yet is other systems are learning what the majority of smaller trader automated traders are doing, one of my thoughts are geared to why HFTs do as they do as they might be picking up enough clues of the automated trader getting in and force market to go other way by HFT. Volume for different markets come and go, one day some other market be king and Crude Oil and Indexes will one day be gone like Pork Bellies and T-Bills, these were at one time huge day trading markets and they been delisted. The more you can read charts, the better your automated systems will be fine tuned. But regardless, you have to learn to adapt when markets change, just like the seasons, summer people go to vacations, less volume, volume drives markets and when markets goes up on little volume, one can expect hard falls on huge volume. The more one studies the markets, more education one gets, and what was once many questions and saying market has never done something before, you change and have the answers and be able to say that current market conditions have done something eight years ago.
I have thought about what you both had to say very carefully. I am thinking my examination of action in recent times compared to years ago may be too simplistic. On the surface it looks the same. There are similar peak hours and you get one of three kinds of days..... flat consolidation, single trend, dual trend. My system works best on 2 of those kinds of days and not so well on one of them. I was theorizing that the kind of day that is toxic may have been more prevalent back then. However, I was not necessarily finding evidence of that. It is like the market was just slightly out of sync with my system. After what the two of you say, I am probably oversimplifying the whole analysis. The answer probably has more to do with what the dominant technical indicators that were being traded off of to form one of those 3 kinds of days. I'm thinking whatever indicators prevalent now end up, on average, forming a more "clean" market structure for me to trade off of.