Your non-response was a restatement of your original. The question remains, "How?" As in, "So, how does one "correlate" a common indicator with future price?" I'll go ahead and pre-suppose an observation that your "test" did not use the subject indicator as does your current-use algo, huh. Yeah yeah yeah, I'm prescient. And so then, "So for example after MDAC crossover you measure the change in the price in future X bars. Where X can be like 1 to 20. You test different values and time frames. Soon you will discover there's not much correlation or relationship. I'm not talking 100 datapoints like they do in books to show you it works... that can be done easy, it's just a curve fit. I'm talking about at least 1000 data measurements or more. " So, was this data 1-minute? 1-hour? Something else? 1-day? 1-week? Over the last year, I have conducted rudimentary tests on the MACD and found it reliable on S&P500-related, as well as FAANG-related. ("Reliable" means, 67% profit-to-loss, *and* profits being roughly twice the size of losses.) This was on 1-day candles, for 3, 6, 1yr, 2yr+. As a 'trend' oriented tool, I cannot imagine using it in periods for which trends would be hard even by eye to identify. Like using a screwdriver when what is needed is a chisel: do NOT expect good results. Further, while your test statement mentions checking series output 1-day out to 20-days, the actual test statement is only "MACD Crossover" -- no sense of degree in there, just a binary outcome? This was programmed, right? So it would not have been a big deal to write an elasticity-based test instead. (That would be, "the angles of the crossover" for those following along.) This avoids a common (and related) T/A curse of having your indicator lines blob together for many candles at a time: numerical results from easy discernment of usable triggers. I have little doubt that this is what you have in your current production system. Yo.
Back on topic... Has anyone made use of adaptive moving averages? As mentioned, I'm currently looking into MAMA by Ehlers and I find it interesting, but don't fully understand the math behind it which is rather complex. There are other adaptive moving averages with less complexity.
It's still just MAs, just layered, and labeled to look cool. In use, it's like the choice of using an SMA(20) or an EMA(10): you can fiddle and get the two to get very close, but what can your *theory* support? (Cuz when things get sticky, it's the theory to which we return...) Feedback loops -- whether by eye, whether by hand-calculated, or whether on-going in a multi-layered ML set-up, are still advisable. To expect a market regime to stay static because your favorite T/A is set at 11|25|4 or whatever -- that is a recipe for disaster. Test, re-test, tune. (I think that'll be my next tattoo....)
I thought this was a trading forum. not investing forum... I can take almost ANY indicator and curve fit it to ANY market on 1 day candle yearly charts by using parameter optimization, and show you a beautifully smooth equity curve beating the market. But the truth is it will fail in the future on unseen data, meaning it does not have predictive power. Anyways.... charts and indicators are nothing but Rorschach tests
No No No!!! This is the problem with losing traders, they view things ass about face and then go about blaming something else for their failures. Where did you get 'quant' from in your username? Anyhow, enough said, someone so sure of themselves as a trader sure is doing a fine job of helping other winning traders into a profit.
Mickey, please enlighten me, if an indicator does not have any kind of predictive power to help determine future direction of the price, what is the value of an indicator then? Why would anyone use it? To make the chart look pretty? For an emotional support? Maybe I should go back and re-read all the books on technical analysis coz I missed the whole point. I'm not saying you can't use indicators and trade profitably. All I'm saying is the people who are using indicators and make money are profitable for other reasons, but not the indicators. I'm not blaming anything or anyone here. I'm just expressing my opinions, I hope we can agree to disagree in a peaceful matter without personal attacks.
My suggestion. You have been registered on ET for only a matter of weeks. ET had thousands upon thousands of pages of posts of highly intelligent discussion about every aspect of trading. Instead of me or anyone on ET spoon feeding, bite your tongue and use the search feature prior to making 'informed observations' to ET within your own small trading world. But one leg up from me, if you think indicators predict, there is a major issue in bad thinking. Indicators indicate some clarity on what happened, no what will happen. Think probabilities not certainties.