Let me see if I can enlighten somewhat. A moving average trails price and frequently used as a buy sell signal. For this exercise, lets call it a stoploss line. A moving average lags price and splits it, ie a 10Ma is the sum of prices from today all the way back 10 days divided by 10. The result is a smoothed line. To my way of thinking its illogical technically because the longer the MA, eg 200MA, the more irrelevant price is summed over 200 days. Whereas an alternatve way of looking at a trailing stop, just look at price 10 days ago, end of story. Now trading chart bars occurs over calendar periods, eg, 1 day, 5 days, 22 days (month), 65 days (1 quarter), 130 days (6 months) So using the alternatve method of calendar weeks and months, you can use stops of 5,10,22, 65 etc, they become rolling stop loss periods. If for example you wish to draw the indicator, you use a 1 period MA, offset 5, 10, 22, 65 etc. You can offset positively or negatively, I prefer positively as it unclutters the chart more. Price below the line, is your exit signal, ie price is below a trailing/rolling 22 days. It's also is easy to program into an algo. There is no lag as such, all you're doing is saying, "if price falls below price 22 days ago, I'm out!" (No lag such as averaging a series of days) You'll notice reactions in particular, 10, 22, 65. Gold seems to operate on 15.
If you wanna make this more interesting, if not more elaborate, start utilizing nested conditional statements (eg. if bullish(if crossover...then reverse)...then exit) and assign weights to different scenerios (eg. trend=5, range=1, etc).
Thanks for the detailed explanation. It makes sense, but I prefer simply looking at key levels. To me it wouldn't matter what price did 5 or 10 days ago unless there was an obvious reaction there. But I'm sure that if you look at this enough, your brain starts picking up on the patterns, and hence why I'm sure you're making it work for you!
I may not be the guy you are talking about - "candle by candle" - but that is basically what my Algos do. While I can set it to "look back" - say 30 days - 100 days - or whatever - there really is no need to do that. I do NOT use Indicators - and there are NO Inputs that can be backtested or changed. And - I only look back - One Bar - because that is the Default in MultiCharts - "Maximum number of bars study will reference" - and because I am only interested in what Price is doing - NOW - since the previous Bar. And - I don't need to Backtest - as there are NO Values to Backtest. That may sound - NOT possible - but it works. If you want to see sometime - I can bring you into a Zoom Room. I am sure most traders will think that is not possible - but the proof is seeing.
Yup, the brain has definite biases, very easy to see what are coincidences. I tried using it, that was how my last Journal was based on but it turned out an utter flop. That's why I'm sharing it, like MA's it works in theory but not much in practise, prolly about 50/50 coin toss.
IMO if you are trading without EW, then its like driving without GPS. EW encompasses everything about TA using statistical data. If you aren't using it to its full extent then you are only getting snap shots. I believe that renaissance technologies were essentially using EW. For example, I omitted the pullback expectation to (iv) from my original chart but I have had it sitting there all week, waiting for (iii) wave to complete. I was planning on selling a put spread at EOD on the 27, since the (iii) wave was extended past the 261.8% level, the retrace to (iv) .382 level as was a certainty statistically, (.382 to .50) is pretty typical of a (iv) wave. I however didn't pull the trigger because my current platform doesn't allow conditional orders to take profits based on the underlying price. As you can see, I could have sold the spread at the eod on the sep27 and woken up to profits, and then given some profits back by eod. Had I sold the spread at the eod on sep28, I would have woken up to profits, but then be losing money at the eod sep29...so with options especially, timing is crucial...but order execution is paramount. Anyway I am poised for a trade on tuesday EOD with IKBR. I'm just getting used to the platforms conditional orders.
Support and resistance work well with many futures, don't know about stocks, as I don't follow them. It's extremely rare that a support line or resistance line is broken in futures and I'm surprised it just keeps moving in the same direction. I'd guess around 5% of the time I'm surprised. The issue isn't whether there are bounces at support or pull backs at resistance, the issue is, can you profit off it? Knowing whether it will not break support and miss it by 2 ticks, break support by 2, 5, 10, 25 ticks before reversing and then if it will bounce 2 ticks, 5, 10, or 100 in terms of profitably is the hard part. The hardest of course is when to Add to your position (e.g. when it keeps falling and hits a larger support level or breaks resistance but is an even better shorting opportunity) and when to stop out if it doesn't go your away. For example, I buy at 80.03 as the support level is 80, it hits 79.98 and bounces, but moves up only to 80.15 and I had a profit target at 80.25 so eventually my stop is hit and I lose money. The support level worked you just didn't know how impactful it would be. That's the hard part of trading. The profit target and stop loss portion, not whether it will bounce or pull back. To some degree it almost always does, at least in futures.
This is exactly what keeps coming up to me over and over, year after year. There is simply nothing on its own that is either really good, or really bad. I'd love an indicator that never worked. If every time it told me to buy something and the price dropped, fuck, I'd just learn to always short and collect my income stream. But when you really look at things on their own, its right around that 40-60% level. What most traders seem to rely on is a confluence of things lining up, but now you're talking way less signals.
The myth of trading is that everything is statistics and nothing is 100% right/wrong or good/bad. Any indicator or support/resistance can work as well as fail. So most arguments are meaningless. A random action is 50/50 probability. That’s why trading attracts so many uninitiated because anyone can do anything and potentially win. The Market Wizards book says it all. No untrained person will ever try to perform a heart surgery but everyone wants to trade. In fact, trading is far harder since if you’re trained you can do heart surgeries. However, even well trained traders may not succeed and winning traders will inevitably lose sometimes.