[Sharp2Be]

Originally Posted by jperl »

There actually is a much simpler analytic approach in determining what time frame you should use for trading. The idea is based on the premise that most traders are actually trading market noise: the volatility of the market as it oscillates up and down during the day. The time frame to chose then is one in which the range of each bar is much less than the market volatility. Once you know the market volatility, it's a simple matter to choose a time frame.

So quantitatively, what's the market volatility? The market volatility is one standard deviation of the price data with respect to the average price. So for example if the standard deviation is 20 ticks, choose a time frame in which the range of each bar is much less than 20 ticks.

If you are using a time frame for which the range of the bars is more than 20 ticks then you are using too long a time frame.

Jerry, sounds like a good approach. Question: Which timeframe would you then use to calculate the stddev? Thanks.

[JBWTrader]

Hi Suri
That is a very helful tool for traders. Thank you for posting that.
Your comments on Tick chart size vs decision making is very relevant.

What you have diplayed is a tool that can certainly help people form a more robust approach to the markets and it really asks the Question for people. How well does one know their strategy/Tactics/Systems.

Of course there will always be debate about wether stop losses or targets should be rigid etc however at the end of the day these tools forces one to think further on many aspects of trading including ones style and rules..

You have been around a number of years and have a lot of people who respect your knowledge. I am one also who does...

Thanks again
All the Best
John

[jperl]

Originally Posted by sevensa »

How do you calculate the average price? If you use x minute bars to calculate this average, wouldn't this mean that you always will need to use a time frame less than x to trade?

The average I am referring to is that of all trade data over the time frame which you are looking at. So, for example, if you are a day trader and your chart data starts at 9;30 this morning, you would average all the data from 9:30 till the present. For a complete discussion of averages, standard deviations, see the Trading with Market Statistics Threads

Originally Posted by sevensa »

Wouldn't the result be drastically different depending on the number of ticks you use?

and

Originally Posted by Sharp2Be »

Jerry, sounds like a good approach. Question: Which timeframe would you then use to calculate the stddev? Thanks.

No if you start every average computation at the same starting time. Yes if you start the computation at different starting times. For example, if you compute the average of the trade data starting at 9:30, it will look the same on every chart you use, 1 min, 2 min 3min etc. That's the beauty of this type of statistic. It's time frame independent. If however you start the computation including yesterdays data along with today's, then the average will have a different value.

Originally Posted by sevensa »

... wouldn't you want to use an interval with range larger, instead of smaller than the volatility/noise to filter out the noise?

No. If you filter out the noise you essentially filter out any trade. Noise is where the action is. In the words of Nihabaashi "To fear volatility is to fear profits". Again a complete discussion of this is in the Trading with Market Statistics threads.

[sevensa]

Originally Posted by jperl »

The average I am referring to is that of all trade data over the time frame which you are looking at. So, for example, if you are a day trader and your chart data starts at 9;30 this morning, you would average all the data from 9:30 till the present. For a complete discussion of averages, standard deviations, see the Trading with Market Statistics Threads

Sorry, but I still have a difficult time to understand how you will use this to decide the timeframe to trade on? Do you use the Standard Deviation from yesterday to decide what time frame to trade on today? But then the standard deviation at the end of the day will be drastically different than what it was after the first 30 minutes if you continue to include all the data from the beginning. Or do you run it from 9:30 and check it every hour to see what the current calculations are and change your timeframe after each check?

No. If you filter out the noise you essentially filter out any trade. Noise is where the action is. In the words of Nihabaashi "To fear volatility is to fear profits". Again a complete discussion of this is in the Trading with Market Statistics threads.

I don't think noise and volatility are the same thing. You want to filter out noise and trade volatility. This will be accomplished by trading a higher range than the results of the calculations, which in essence will filter out the noise and allow you to trade the volatile moves which rise above the noise. There is no rhyme or reason to noise. Why would someone want to tade that?

[jperl]

Originally Posted by sevensa »

Sorry, but I still have a difficult time to understand how you will use this to decide the timeframe to trade on? Do you use the Standard Deviation from yesterday to decide what time frame to trade on today? But then the standard deviation at the end of the day will be drastically different than what it was after the first 30 minutes if you continue to include all the data from the beginning. Or do you run it from 9:30 and check it every hour to see what the current calculations are and change your timeframe after each check?

What I personally do, is look at the the SD from the end of yesterday and see if there is a position trade that I could enter for today. If there isn't then I will start a computation of the data just for today and wait until the SD gets large enough compared to the bar range before initiating a trade for today. Again this has all been discussed in the Trading with Market Statistics Threads

Originally Posted by sevensa »

I don't think noise and volatility are the same thing. You want to filter out noise and trade volatility.

I gave you a quantitative definition of volatility. It is a measure of the deviation of the data from the mean. If I enter a trade at the mean, I expect the market to move 1 standard deviation. How it gets there is not of any concern. It could move there smoothly or jiggle its way there. If you wash out all the jiggles(which are just standard deviations of the data on a shorter time scale which I presume is what you mean by noise) then you miss the opportunity to define a good entry and a good exit point.

[sevensa]

Thanks. I think I have a vague idea of what you are saying, but it sounds more like a trading methodology than a way to pick a time frame to trade on. In fact, it sounds like the time frame is irrelevant.

With all due respect and I am not trying to sound sarcastic, or launch some kind of attack as this is not my intend at all, but when you said in your initial post "There actually is a much simpler analytic approach in determining what time frame you should use for trading.", I didn't expect someone will need to read 12 threads with 1000's of post to understand this simpler way of determining the time frame. After all, Suri outlined his approach in one post.

In the end, I think this is just another case of "for each their own". I appreciate your insights and input on determining the trading time frame and thank you for it, but I think I will pass.

[jperl]

Originally Posted by sevensa »

Thanks. I think I have a vague idea of what you are saying, but it sounds more like a trading methodology than a way to pick a time frame to trade on. In fact, it sounds like the time frame is irrelevant.

Sorry for the delay in getting back to this.
What I sense here, is a need for a better understanding of market probabilities and how that helps your trading style. The market statistics threads will help you in this regard, which is why I wrote them. They are NOT a methodology(although I do present some methods along with them). It isn't necessary for you to read all the threads. In fact if you just read the first few and look at the videos you will have a good working knowledge of market statistics.
I think once you have a feel for the concept of standard deviation, you will understand my statements above about why you should not trade bars whose range is larger than 1 standard deviation

[BlowFish]

Originally Posted by sevensa »

Thanks. I think I have a vague idea of what you are saying, but it sounds more like a trading methodology than a way to pick a time frame to trade on. In fact, it sounds like the time frame is irrelevant.

With all due respect and I am not trying to sound sarcastic, or launch some kind of attack as this is not my intend at all, but when you said in your initial post "There actually is a much simpler analytic approach in determining what time frame you should use for trading.", I didn't expect someone will need to read 12 threads with 1000's of post to understand this simpler way of determining the time frame. After all, Suri outlined his approach in one post.

In the end, I think this is just another case of "for each their own". I appreciate your insights and input on determining the trading time frame and thank you for it, but I think I will pass.

You can read the first post in each thread and get the concept. The rest of the posts are discussions questions and all the other stuff that you get on forums. The first posts are clear concise and structured very well. It is not often you get a really unique perspective on trading but this is. Regardless of your chosen trading style this is an important piece of work (imo).

It is simple but requires some knowledge. Having said that choosing a time frame is simple if you have experience. It is important as its one of the few variables you have complete control over. Personally I just tweak things until it shows what I need to see to capture the swings I am interested in. Of course you need to know what you need to see.