How do I accurately choose the period for indicators?

[maxinger]

many indicator sellers, trading coaches, newbies talk about this.

Simply delete ALL indicators.

WHY?
market move in various forms and sizes and shapes and manners.
market can move certain distance with very small volume or very huge volume.
market can move slower than snail, or faster than rocket.
market can move at 1 degree gradient, or 89 degree gradient.

There is no ONE SIZE FITS ALL period for indicators and people simply cannot accept this fact.

 

[tomorton]

Changing the time period of an indicator will have no effect on its "truthfulness". For one thing none of them are essential in determining what price has been doing - you can see that - and none of them can predict the future.

Indicators are not forensic identifiers of what the market is doing - they are not like fingerprints or DNA or universal scientific proofs: they can be helpful as guides to what you do next.

 

[Orbiter]

Hi guys! I'm new to trading, been educating myself on indicators, chart patterns, etc,. They always teach you how to use them, their history, etc,. but they never seem to tell you how to choose the period for an indicator, for example: Stochastic Oscillator has the period K and period D set to 13 and 3 by default (default settings for the trading platform that I use) Why is that? Why is it 13 and 3? Does changing the period make any effect on the accuracy of the indicator? Does it depend on the time period of the chart? If the time period of the chart is in minutes, does that mean 13 period is 13 minutes and 3 period is 3 minutes? Also another thing that I've noticed is that some traders can predict the exact time the market is going to be bearish or bullish in the future, for example: at X time in the future, the market will be bullish. How do they do that? Is there an indicator for that or is it just an experience related thing? I've tried looking all this stuff up on YouTube and Google but nothing's been helpful so far. Could you guys please help me?

Technical indicators are derivatives of price action (and sometimes volume) so they give you condensed information about the evolution of the price for the specified period. As someone mentioned in a prior post, the default values are just historical artifacts with no special predictive power.

A good way to assess the effectiveness of different indicators and their values is to run a statistical analysis on them and identify intervals with better mean returns. But that requires pretty deep math and programming skills.

The next best way is to construct portfolios of stocks and run backtests with various inputs for each indicator. I hear that there are a few platforms that don’t require coding but most do require a certain level of proficiency in writing code.

If coding a backtest is not your cup of tea you could look up internet for quant blogs and fellow traders that publish research ( I.e. their own take) on the subject. Here’s a sample https://axelarati.github.io/short-term-stock-analysis/2019/09/04/Overcoming-Noise.html that I quote often.

One more note: I personally find that visually assessing price action can be misleading due to the myriad of biases that hunt us. A lot of people use that method and it can be profitable for the chosen few but if you can quantify your rules and test your beliefs why not do it ?

 

[tommcginnis]

If I were you I would forget the indicators and just learn how to read a bar chart. The indicators are only there to confirm what is on the chart, not the other way around.

Errrr -- the chart IS an indicator. Japanese Candlesticks, price "bars", etc, ARE indicators. Even a chart of closing-price against a given time period still takes all the market info for that period and waits until whenever the last trade is printed, and then prints whatever that last value is -- painting that value as being indicative of the market for the entire period. And that process is multiplied by four, for bars and Japanese Candlesticks.

"...Price as indicative of the market for that period..."

The distinction you wish to draw is common, but mistaken.

 

[Orbiter]

many indicator sellers, trading coaches, newbies talk about this.

Simply delete ALL indicators.

WHY?
market move in various forms and sizes and shapes and manners.
market can move certain distance with very small volume or very huge volume.
market can move slower than snail, or faster than rocket.
market can move at 1 degree gradient, or 89 degree gradient.

There is no ONE SIZE FITS ALL period for indicators and people simply cannot accept this fact.

Just curious, what are you left with after you delete all indicators ?

I agree that using indicators just by looking at charts without understanding their meaning can be dangerous but seriously why discard indicators if you believe in price action ?

And even if you don’t believe in price action and TA why would you advice everyone to get rid of them? I suspect there are people making money using them one way or another. Need some help here. Anyone ? Raise your hand, please.

It’s not a rhetorical question. I bet a lot of the contributors here would be interested to hear your answer.

 

[maxinger]

Just curious, what are you left with after you delete all indicators ?

I agree that using indicators just by looking at charts without understanding their meaning can be dangerous but seriously why discard indicators if you believe in price action ?

And even if you don’t believe in price action and TA why would you advice everyone to get rid of them? I suspect there are people making money using them one way or another. Need some help here. Anyone ? Raise your hand, please.

It’s not a rhetorical question. I bet a lot of the contributors here would be interested to hear your answer.

Decades ago, I started with thousands of indicators. As time goes by, I deleted those indicators till there is none left.

My trading is mainly technical. My chart is just volume based candlesticks with perhaps zig zag line.
Trading based on technical doesn't mean you must use indicators.

 

[tommcginnis]

A good way to assess the effectiveness of different indicators and their values is to run a statistical analysis on them and identify intervals with better mean returns. But that requires pretty deep math and programming skills.

...Or a pair of eyeballs.

The next best way is to construct portfolios of stocks and run backtests with various inputs for each indicator. I hear that there are a few platforms that don’t require coding but most do require a certain level of proficiency in writing code.

...Or, using your eyeballs, you could mark off on a graph, the entries and exits of that contract per your pre-determined rules, compute gains and losses, and go from there. Labor intensive, but a year of EOD data will take you no more than an hour or two to chunk through -- not the end of the world. And if your rules can't survive for a year's run? They're not worth evaluating for longer. As the song goes, "What have you done for me lately??" is a pretty fair scientific standard for time series.

If coding a backtest is not your cup of tea you could look up internet for quant blogs and fellow traders that publish research ( I.e. their own take) on the subject. Here’s a sample https://axelarati.github.io/short-term-stock-analysis/2019/09/04/Overcoming-Noise.html that I quote often.

Whattttt? More sciencey stuff? You're going to kill 'em! Blow up their brains! (Or, perhaps, ground them on prior data, efforts, standards, and lessons-already-learned. Heh! This Science stuff....!)

One more note: I personally find that visually assessing price action can be misleading due to the myriad of biases that hunt us. A lot of people use that method and it can be profitable for the chosen few but if you can quantify your rules and test your beliefs why not do it ?

1,000% AGREED. The magic word is Pareidolia --
https://en.wikipedia.org/wiki/Pareidolia
It's a thing that has helped us as species, but that very experiential bias is what keeps us from a totally objective -- evidence-driven -- viewpoint. The ONLY solution is as Orbiter observes: HAVE RULES. No cheating with rules -- especially *coded*, *immutable* rules.

 

[Orbiter]

Decades ago, I started with thousands of indicators. As time goes by, I deleted those indicators till there is none left.

My trading is mainly technical. My chart is just volume based candlesticks with perhaps zig zag line.
Trading based on technical doesn't mean you must use indicators.

That’s fair. Thanks for clarifying this.

I used to trade Elliott Waves (many zig-zag lines) for a couple of years. I had good success with it but I never felt confident enough that the results weren’t based mostly on luck (or maybe at best on my ability to not panic). Statistical analysis, technical indicators, trading many positions at the same time by using automation changed everything at least for me. But again it’s a matter of style.

Best of luck !

 

[Orbiter]

...Or a pair of eyeballs.


...Or, using your eyeballs, you could mark off on a graph, the entries and exits of that contract per your pre-determined rules, compute gains and losses, and go from there. Labor intensive, but a year of EOD data will take you no more than an hour or two to chunk through -- not the end of the world. And if your rules can't survive for a year's run? They're not worth evaluating for longer. As the song goes, "What have you done for me lately??" is a pretty fair scientific standard for time series.


Whattttt? More sciencey stuff? You're going to kill 'em! Blow up their brains! (Or, perhaps, ground them on prior data, efforts, standards, and lessons-already-learned. Heh! This Science stuff....!)


1,000% AGREED. The magic word is Pareidolia --
https://en.wikipedia.org/wiki/Pareidolia
It's a thing that has helped us as species, but that very experiential bias is what keeps us from a totally objective -- evidence-driven -- viewpoint. The ONLY solution is as Orbiter observes: HAVE RULES. No cheating with rules -- especially *coded*, *immutable* rules.

Love your post(s) and had a really good laugh. Needed that. Thanks.

 

[tommcginnis]

Love your post(s) and had a really good laugh. Needed that. Thanks.

!