I never backtest over more than 3 years because I think that in that period you will had almost all kind of markets. I also think that markets from years ago are not behaving like markets do now. Internet, lower margins, better possibilities for non professional to trade with decent software, HFT.... All these things have changed the behavior of the market. So to me it doesn't make sense to backtest data from years ago. But that's just my opinion and based on what I use to trade. For me it is more important to built a self adapting system. A system that adapts itself to the markets. I think it gives better returns and better control over losses. A least if the system is built well. About 9/11. People who stayed overnight are LT traders. If you watch the charts the trend was short, all indicators were short, so the drop only increased the profits. At 8/11 normally nobody should be LT long. The problem is not how big was the move, but what was your position at that moment according to your system? If markets crash 25% in the direction of my position, I have no problem with that. My experience is that most crashes happened with the trend, not against it. The red arrow is just BEFORE 9/11.
I think we are talking two different things: What I meant was for those people trading 100-200 contracts, they would normally risk only a few ticks, not 50 points, and they have a smaller profit targets like 1-3 points (not 20 to 100 points), so in this way they only stay in the market for a brief period of time to reduce the risk exposure, and they can also sleep well without worrying the overnight dramatic changes (could be profit or loss). You are talking about it is easier to pile up big shares/contracts if entering a trade that would last for a longer length (between entry and exit), and that is correct.
100 contracts in es is a drop in the bucket, wont even move the price by a tick. on some commodities the last 3 years have been a downtrend, should at least include an uptrend period in your tests.
If you were day trading Indexes and you were not watching TV and not checking email, like who does when day trading, and didn't have a broker calling you to say exchanges closing in fifteen minutes and you were long.....my broker called me and asked what I wanted to do, I went near totally flat bought Puts in Indexes. Yea, for long time I used three years, but I am opposite of your thinking the markets have changed, emotions don't change. I have Wheat charts from 1920s and they haven't changed in near 100 years. Much has to be defined as what is YOUR definition of trend, long term commodities I never open origin trade with trend, so trend doesn't matter to me, but what if someone using monthly charts or buying the lows of April at 1100. The problem is exchanges don't usually close down and if one took long position thinking they were going to risk couple hundred each on ten contracts is down $35k week later. Not only can he lose his account but lose his home and family cause he was greedy and using too much of his account for a position.
I only trade intraday ES, and there were uptrends and downtrends all the time. I am daytrader, so probably intraday I would see uptrends and downtrends in other commodities too. For LT trading it can be different. But I think you don't need both kind of trends to test. A downtrend is an uptrend but upside down. If you can manage an uptrend you can manage a downtrend, because trends are a result from market behavior. More important is to find and test the STRENGHT of trends, but that appears to be problematic for many traders. Because strength influences evolution of prices more than being up- or downtrend. It gives an indication how far the move will go.
I have to disaagree upmove and downmove have very different characteristics at least on the commodities that i follow. on daily bars and effects the intraday movement of it also.
That's very simple for me. I want to take every move from the bottom to the top and vice versa. By move I mean intraday 2 to 3 moves a day. So not every "move" on 1 minute charts. Moves bigger than 2 points. If the move is 2 points I want to take 2 points, if the move is 25 points I want to take 25 points. I don't say I take it, I say I WANT TO TAKE. I am satisfied if I take at least 75% of the full range of every move.
I think it depends on your "system". My system is divided in a (huge) amount of different scenario's. Continuously my computer monitors the situation and adapts my strategy according the development of the market. I traded for many years forex, which is completely different from the ES. I use identically the same system now in futures with good results. What you see as different might be incorporated in my system, that is self adapting, so that for me there is no difference, wereas for you there is. I will try to give you an example of what I mean: We both have a car. We both drive in summer and in winter. You say that winter is more difficult because of slippery road (ice and snow). I say that winter is not so bad for me, although we drive in the same conditions. The only difference is that you drive all year round with summertires and I use in winter wintertires. The changing of tires represent the self adapting part of my system. Markets change and a good system will adapt itself automatically to this chane in markets. So I would never use fixed MA's, or stochastics or MACD or RSI or anything else. Parameters in all parts of my system changes continuously in reaction to the market behavior. The try to find the optimal setup for that particular trade. After each trade the "ritual" starts over again from scratch automatically.
so your describing an online machine learning algo, ive experimented with those but still prefer to do manual adjustments on my systems.