Is there a simple way to determine a trend and calculate it programatically? For example I am looking at 2min candles on ES, I want to know if the overall trend is going up or down, so I plot a 20 day SMA. Now I can see with my eye that it seems to slope down today, so its in a down trend. But the sharp reversal that happened at Mondays open on Sunday evening how do you calculate that? I was thinking take the last SMA and take a SMA 5 periods before and 10 periods before. If the 10 > 5 > 1 then we are in clear downtrend. if 10 < 5 > 1 we had a reversal? Etc. Im sure there is an indicator for this, or a better way to do it? Perhaps something that considers volume too, if we drop in price on low volume weight that very highly as a downtrend. Vs a drop in price on large volume weight that low as that can easily rebound.
There are lots of ways discussed at the forum for "determining if its a trend" but it does not forecast that the trend will continue as such nor does it imply you have good trade management to stay in the trade if the trend decides to continue. I believe that if someone can not visually look at a chart to see if its a trend or not a trend...they aren't going to trust any math formula or program codes. In contrast, if they designed something themselves...they'll trust their codes more than codes from someone else. You should use the search function here at ET because I've seen some decent trend discussions involving coding and indicators although I personally do not care for codes...I'm a visual guy. Good luck with your trend code design.
There's nothing inherently wrong with the components or the method you described but both of them can be improved upon. Setting that issue aside until you're able to acomplish those 2 tasks, which will occur naturally as you continue on the path you're already on...then the important task is what you're already focused on. IOW, the gist of what you will need to do, regardless of the components or method ...is to figure out their constraints. Learn to distinguish the limits of your tools and methods. When u can do this, which as you've already stated you're already doing by cataloging the precise details of when they are not working ...then you will have found your bliss and you will be able to create the components and tools which enable trading against the direction of your tools and components. ...because in essence, you will know their weaknesses because you created them and live with them daily. Don't change anything until it makes good sense to change it.
At it's core, successful trading is one of the easiest things an individual can do. ---Most people lose money trading.
I think a good Trend identifier should work whether it is one minute charts or monthly. I like to use as John Hill's "Thrust" bar and way to program it is a bar that is completely outside of 18-21 SMA, longer period SMA will keep you in longer period of time but it will also take longer for trend change. So until entire range of a bar is completely on other side of SMA, trend remains the same. Another good programmable trend identifier is a close beyond most recent Pivot. I also besides the two, use slope of SMA, but that is much tougher to program and would have to be optimized on weekly or bi weekly basis. First close beyond an envelope, Kelter Channel or Bollinger bands. Downsloping trendline, if one, two, three closes beyond the trend line, trend is up. 7>21 or 7<21 ema's which are fast and offer many more losses cause of the speed and can offer much and many signals when in chop. You can even use Monthly charts if last months low is above previous low, trend is up. Can also use yesterday's, weekly, monthly close, if current price above one of those close's=trend is up. Others that I have used and or in automation systems using is RSI, above 50 is trend up, below trend down, same with MACD, Stochastics, just about any oscillator. But I prefer to stay away from indicators flipping over middle lines and let price itself show the trend. Good luck.