ATR Value

[ES]

we risk 2% on any given trade. period. richard and c.f. have always stresed the importance, exits over entries. period. at the end of the day the one with more money wins. period. i keep things simple. complication equals difficulty and stress. one must feel comfortable. if you're not emotionally stable about what you're doing it's not going to work. there;s that fine line between your comfort level and pressing the emotional envelope. please help me find a 3.00 stock with a 5 ATR. good luck
things are proportionate. $2.00 stock .89 ATR 50.00 stock 5 ATR
the quintessential element is the refining of the original turtle system but ensuring consistency is equally important. the analysis of trends ?
you see a chart making new highs or lows, if the ascent is greater than 30degrees get ready for a ride. the longer the trend time frame the longer you'll be kept in the trade. thus tighter atr's for stocks below 5.00 works well. if they make new highs / lows then buy them again and risk 1% not 2%. stocks between 8-20, 2% with a looser atr, higher priced items like AMGN, (pull it up) even looser atrs. i cannot stress the importance of the 1% risk rule for stock trading less than 5.00. i'd rather make little amounts of money in compressed timeframes than get killed rapidly with a gunslinging risk attitude.
all the best

[MCT]

The forum, as c.f. has pointed out, is for all types of traders, long-term, short-term, mechanical, discretionary, turtles, tribes, and what have you …

Equalizing risk to volatility is not an exclusive domain of the turtles; many other successful traders have been doing it in various forms for decades. Their performance is undeniably correlated to the turtles.

I presented the idea in it’s simplest form, a $10stock and $50stock with an atr of 5$ does not possess equal risk. According to my definition of risk “risk is relative to volatility, value, and exposureâ€

[Nathan]

[quote]I presented the idea in it’s simplest form, a $10stock and $50stock with an atr of 5$ does not possess equal risk. According to my definition of risk “risk is relative to volatility, value, and exposureâ€

[MCT]

Nathan,

My goal wasn’t to perplex but rather to make the reader “thinkâ€

[Kiwi]

Ok. Got that.

Without actually agreeing with you ( ), how you would apply it to futures?

John

[Nathan]

It is true; my original example was quite arbitrary. I was trying to demonstrate a concept, rather than make any statement concerning the merits of the particular approach, which by the way is not "my approach."

I am not sure "arbitrary" is a dirty word. Even the most well thought out and researched risk management system is based on the subjective value preferences and assumptions of the trader. So from a certain perspective, all risk control methods will appear arbitrary, as will most anything outside of hard science.

I think I understand your fundamental idea much better now, and I agree . (my confusion was with regards to application) Normalizing atr for the share price can in some applications create a more useful and relevant statistic. It appears that you are using this statistic in a way that makes sense in the context of your trading style/risk method. In that sense, I tip my hat to you. I am certainly not interested in debating someone's personal methodology, which i view as none of my business. However, you are making a few claims with regards to the standard atr futures model that may not be accurate (at least according to my frequently dysfunctional math )

You are correct the positions in my example *appear to be* proportional to your example – but you’re going to have to look much deeper!

I said "appear to beâ€

[CRM114]

Menelik,

I'd like to better understand your reasoning. You have talked about having your own definition of risk. Please give an example or two of the kind of scenarios that you feel are better handled by your method than by the fixed stop. Also, when do you get out? If you don't use a fixed stop, how do you decide when enough is enough?

[MCT]

Nathan

I have to say I enjoy reading your posts; it forces me to think in ways I normally don’t consider.

I agree with you one hundred percent, you can’t remove subjective preferences from trading and even science- Heisenberg’s uncertainty principle proves that point. There were two points I meant to present in this thread, mainly that 1) stops and exits skew the exposure profile of such models. 2) maintaining constant exposure to true-volatility irregardless of exits or entries is critical. As you said in your last paragraph, I was attempting to point out holes in such systems that I have personally experienced. Yes, risk is highly personal. Your CONCEPT of risk is the heart of your system and I feel every element of your portfolio should have equal exposure to that risk, and yes your example is equally exposed to the definition of risk as you understand it. My focus was on the CONCEPT of risk and other alternatives for defining risk. Some will say it is highly valuable, others will say it is naïve, everyone is right. It would be interesting to hear other definitions of risk from other forum members.

My hands have a hard time keeping up with the speed of my thought process … I’m currently assisting my wife with a stock system…thus I fumbled in referring to the risk in my example as being denominated in shares. You’re right in saying it is ALWAYS the dollar risked that matters not the amount of shares. What I should have said refered to was “the relative true market volatility exposureâ€

[Dan G]

You could try Parkinson or Garman-Klass volatility to see if these measures better fit your needs. They are similar volatility in that they express what the price movement has been in percentage terms, so to get them in a comparable form to ATR, it is necessary to multiply by the current price, or an average of the price.

I am not sure I am following this disussion very well.

Whatever measure one uses for volatility, you are assuming it is capturing important information - namely, how much the price can be expected to move over a period of time. All volatility measures have this in common.

So if you are using ATR as your measure over a constant period of days, a 1 ATR movement in one stock will be directly comparable to other stocks. If a stock has a traditional volatility of 35%, another stock with a voltility of 35% is exactly the same. Why? Because you are using a measuring tool - like a ruler or a scale.

Now, it is entirely reasonable to wonder if ATR measures volatility well. I think it is a little misguided to put some adjusment factor based on the stock/commodity price. Why? Because you will have to take out that stock/commodity price at some point to get to a dollar risk.

Did I misunderstand everything?

[MCT]

Hi mickslam,

So if you are using ATR as your measure over a constant period of days, a 1 ATR movement in one stock will be directly comparable to other stocks. If a stock has a traditional volatility of 35%, another stock with a voltility of 35% is exactly the same. Why? Because you are using a measuring tool - like a ruler or a scale.

In the tradition of the falsificationist philosopher Karl Popper, I would say, don’t be so certain. For me personally, the exposure will forever be uneven. But if you believe exposure is equal with such methods, for your definition of risk, you are right. We all have our own definitions for what constitutes risk and return. Neither of us is either wrong or right. What you choose to believe is optional. Our theories will remain to be true only until they are falsified. i.e Quantum theory gave way to the theory of relativity, and as of recently, relativity theory gave way to unified theory or the theory of everything. Attempt falsifying your own concepts, you just might come by greater insight.

My point, in general, has been there is an implicit structure in volatility current methods don’t account for.

Good Trading,
MT

[Dan G]

I absolutely agree that ATR doesn't capture all of the information that we experience as 'volatility'. That is why I suggested other potential volatility measures, like Parkinson or Garman-Klass. They use more data, and are theoretically better than ATR. If they are practically better than ATR depends on the use.

I will state that if you are using a metric, you should use it in a consistent manner - otherwise it loses its meaning. Using subjective correction factors seems dangerous to me.

Your method of dividing ATR by the avg. will give you an average percentage movement of the stock over the last period of time. Not all instruments are the same -for example, just compare the U.S. bonds to something like QLogic. You'll get radically different values. This is what you pointed out. Correcting ATR for the current price of the instrument really doesn't tell you anything more than just the ATR - it just presents the same information in a different manner.

I just thought you might have lost your way a little - it seemed the position sizing method you were describing would be more focused on trading equal amounts of volatility per dollar, rather than dollars per volatility, or something like that.

Garman-Klass volatility is a good measure, simply because it uses more information than ATR. GK is not perfect. It is not hard to calculate or understand. I can dig up the exact formula if you are interested.

[MCT]

I will state that if you are using a metric, you should use it in a consistent manner - otherwise it loses its meaning. Using subjective correction factors seems dangerous to me.

I just thought you might have lost your way a little - it seemed the position sizing method you were describing would be more focused on trading equal amounts of volatility per dollar, rather than dollars per volatility, or something like that.

Exactly my point. Trading only in terms of dollars-risked-per-volatility is subjective because it forces you to subjectively size, equalize and fix dollars to volatility using an arbitrarily selected multiple of atr, which in itself presents various exposure conundrums. Such methods do not reflect the inherent realities of the market. I agree using inappropriate subjective correction factors are dangerous. I also agree volatility metrics should be used in a consistent manner –but we might have different definitions as to what consistency is.

I’d like my portfolio’s risk exposure, which is tied to the dynamic nature of market action, to maintain a consistent exposure in terms of volatility-per-dollar risked, as well as dollars-risked per-volatility. One has to determine the nature of volatility-per-dollar risked before answering how many dollars to risk per volatility. It is my belief the dynamics of market action should be an integral part of anyone’s system design. The two should exhibit strong cohesion. There should be a two-way feedback mechanism in which market realities not only shape the system’s responses but also is apart of the system. Thus, the collective whole cohesive system shapes reality in an unending process in which our creation (the trading system) and reality may never become one, but become identical. That to me my friend is being one with the markets, nature and life. (Refer to “The Problems of Inductionâ€

[MCT]

check out, the following quote. It presents the massive problems of induction in economics and our personal lives. Take a look at both my example and SirG’s follow up example. The problem is everywhere …


[quote="Menelik C. Theodros"]Western philosophy regarding event causality seems to have inverted causal relationships. Social mood, reflected by the aggregate record of stock market prices, “leads eventsâ€

[Dan G]

MT - many nice points.

I think volatility only has meaning relative to the time it was measured over. That is what volatility measures, is the movement of an instrument over a period of time.

In your two instrument example, the first is without a doubt more volatile than the second. However, the measurement was taken over a period of time. Using an ATR, we are assuming that the near future will be like the recent past. It's not a bad assumption, but not a great one either. Volatility is notoriously anti-persistent, in that it switches direction more than a random walk would. This characteristic of volatility might be what you are looking to add to your mental image of what vol is.

Here is an example:

Instrument 1: $200 stock/ $2 ATR
Instrument 2: $20 stock/$2 ATR

We get a buy signal in both.

Over the next 2 days, both of the instruments move up by 4ATR.

The second is 10 times more volatile than the second. Let's use your volatility position sizing strategy as I understand it. The first instrument will get 10 times as much allocated to it as the second.

Instrument 1 will get $1000 and buy 500 shares
Instrument 2 will get $100 and buy 50 shares.

Profit 1: $4000
Profit 2: $400

Lets use a standard % equity;

Instrument 1 will get $1000 and buy 500 shares
Instrument 2 will get $1000 and buy 500 shares

Profit 1: $4000
Profit 2: $4000

You are saying, that because 1 has a lower volatility, it has a higher likelyhood of having a big move that is many times the recent ATR. I would test this as some sort of entry technique/filter - is it appropriate to use as a position sizing technique? What if you are wrong in your direction. Then, that large move works against you - you are exposed to that large move.


I think if you test your position sizing stragegy over many instruments, you're going to find the dollar returns completely dominated by the lowest volatility instuments. I don't know if this is true or not, and if I am wrong, I would appreciate hearing about it. I've been wrong before!

[Sir G]

The Problems of Induction

I don’t see this so much as a problem, but more of an issue of our perceptions.

In my book, Ed Seykota was right when he said, Everyone gets what they want.

From scientists, to teachers, to traders and everything and everyone else in between, we all have biases that draw us to the conclusions that we want.

Ed Seykota has a bias to succeed in his theories and in the markets, so he, it appears, keeps looking and searching for the gems.

Others, may have a bias to feel and appear that they are succeeding, so they do just enough to get what they want.

We all have the ability to seek our truths and I believe it has much to do with our perceptions. We can spend our time perceiving things in such a manner that we feel safe, secure, smart… what have you…In our minds, this comfort zone seems to have a very high rate of winners and very few losers… because the meaningful purpose of this zone is to reinforce our comforts, not to challenge us by pointing out our mistakes and errors.

You can look beyond the comfort zone where the ratio of winners to losers are now titled in the side of losers… or that 4 letter word…failure. This is where we test things out, we run stress tests on our ideas or the ideas of others to see if they truly hold up. And to what degree they hold up. Few things will actually succeed in this environment of seeking the truth.

It’s the difference of reading a book (or a post on a forum) about something and simply assuming that everything is factual and real vs. reading a book and finding out for yourself how it holds up, what/where are the strengths and weaknesses. And if the book in whole or part, should be accepted or rejected.

Trading is very much the same way, there are so many opportunities to lead ourselves down the wrong road.

My hopes for this forum is for it to be a place where conventional wisdom is checked at the door, stripped down to find out what it is made of, prior to it being accepted for what it appears to be at first or second glance.

It is great to see the volleys of thought bouncing around in this thread!

The membership of this forum almost 500 strong, has created a very nice collective brain trust!

Gordon

[Nathan]

I have noticed that the "2atr" thing keeps being mentioned. As I said in my second post, my example was arbitrary. However even given this, I made a mistake by adding an unnecessarily arbitrary element, which was the stop. My point did not require mentioning anything about a stop.

The problem here is the method has invested an equal amount of dollars in the two instruments with differing volatility profiles as I define it, thus being over exposed in the $60stock. Yes there is proportionality in the theoretical commitment BUT it remains theoretical, it hasn’t been risked yet

I am trying to figure out if there is a real issue here or just semantics:

There is NOT an equal amount invested in both stocks. that was my point in my last post. However, the intended risk (what i call stop out point) is the same. The model has attempted to equalize the volatility impact on the portfolio through Position sizing, not stop placement or intended risk. Even with no stops, I believe the ratio of one position ($$ cost of buying each stock) to the other is the same using both methods. Stops, ect have no effect on this.

In my example, market risk is always 1ATR/Close, period-that will always denote true market volatility to me.

ATR is just a number, as is your %adjusted ATR number. Both numbers have nothing to say about risk, or much of anything else, without a context or application. Could you explain how you take this number and apply it so that It is meaningful in the context of risk? Perhaps using it to size a few futures positions in context of a portfolio would clarify things, showing how your method differs with regards to dollar commitment, contract #, ect from the conventional model. I know you did this with stocks, but I think another example would help.

I agree with your induction post. However, I am not sure that is so much a problem here. I am not (is anyone else?) trying to assemble a system from "little parts". My "little parts" were just real world examples of the abstraction I was trying to demonstrate.

This subject is interesting to me because it is right on the edge of something I have been considering with regards to position sizing for short term trading. Because stops are closer, I think a qualitative assessment of potential risks, as well as the total face value dollar commitment (rather than just historical testing, ect) May be something I am more comfortable with (as opposed to just historical volatility based position sizing).

best, Nathan


[/u]

[Sir G]

In my example, market risk is always 1ATR/Close, period-that will always denote true market volatility to me.

Just a friendly reminder to everyone who wants to test something like 1 ATR/Close... Don't use continuous contracts when you do it! Use actual contracts, as that is the only way you will get a true close price.

Even with stocks... I have a hard time thinking of the split adjusted contracts as "real" prices. It is simply a continuous contract for the stock market.

IMHO, I'm sure the fella that thought of continuous contracts was a software designer who didn't want to put the added overhead into his program to properly manage the various contracts.

Gordon

[MCT]

The second is 10 times more volatile than the second. Let's use your volatility position sizing strategy as I understand it. The first instrument will get 10 times as much allocated to it as the second.

What if you are wrong in your direction. Then, that large move works against you - you are exposed to that large move.

I think if you test your position sizing stragegy over many instruments, you're going to find the dollar returns completely dominated by the lowest volatility instuments.

Using an ATR, we are assuming that the near future will be like the recent past. It's not a bad assumption, but not a great one either. Volatility is notoriously anti-persistent, in that it switches direction more than a random walk would. This characteristic of volatility might be what you are looking to add to your mental image of what vol is!!!!!

You are saying, that because 1 has a lower volatility, it has a higher likelyhood of having a big move that is many times the recent ATR. I would test this as some sort of entry technique/filter - is it appropriate to use as a position sizing technique? What if you are wrong in your direction. Then, that large move works against you - you are exposed to that large move.

I don’t think it is wise to calculate position-sizes by comparing different elements in a portfolio, that will surely skew the exposure equation. The standard is the structural volatility of the market. You’re right, less dollars “happens to beâ€

[Howard Brazzil]

You said, "even with no stop." How do you plan to size your positions? ...You see after all you’re stops matter;
the size of your position depends on the stop ‘you’ pick.

Hi Menelik,

I believe you've hit on a common misconception, which is that "position sizing depends on the stop you pick."
That's not always true. Allowing stop placement to define the risk on a trade for the purpose of sizing positions
is only one way of viewing the world. (Granted, it gets confusing when the stops are based on the same measure
of volatility often used to size positions.)

There are many systems that don't use stops. For instance, how would you size positions for a pure moving average
crossover system? You'd probably size by some measure of volatility.

At this juncture, a lot of traders traders will start screaming, "But that's not the same as actual risk!" True. They're not
the same. And that's exactly the point.

For another example, look at the default parameters for the Turtle System Two:

Percent Account for N = 1%
Entry Breakout = 55
Exit Breakout = 20
Stop in N = 2.0

Here, "N" is a measure of volatility. It just so happens that "N" is used both to size positions, and to place stops.
In this case, a single position (or "Unit," in Turtle terminology), represents 1% of account equity in terms of dollar
volatility.

The actual risk on the trade is (approximately) defined by the distance from the entry price to the Stop in N, or the
exit breakout, whichever is closest to price.

Neither has an impact on how positions are sized.

Stop in N can be changed from 2.0 to 4.0, or even 8.0, with no impact on position sizing. The two are separate.

[MCT]

Hi, Howard

I believe you've hit on a common misconception, which is that "position sizing depends on the stop you pick." That's not always true.

Couldn’t agree with you more. That should be clear to all, not everyone trades the same. There is no misconception on my part, just different measuring sticks.

The actual risk on the trade is (approximately) defined by the distance from the entry price to the Stop in N, or the exit breakout, whichever is closest to price.

I personally find a problem with that. It might have worked for the Turtles needs, but not for me. My concept requires my system to be fractally robust. As time frames decrease (from weeks to days to hours) entropy increases. Try using the turtle system on hourly bars … the turtle system's risk exposure increases as time frames decrease, excatly the opposite of what I want; the system degrades. In my opinion, the purpose of a fractally robust trendfollowing system should be to smooth out and ignore market noise.

It just so happens that "N" is used both to size positions, and to place stops. Neither has an impact on how positions are sized. Stop in N can be changed from 2.0 to 4.0, or even 8.0, with no impact on position sizing. The two are separate.

Let’s try your example, to see if what you say is true.

Percent Account for N = 1%
Entry Breakout = 55
Exit Breakout = 20
Stop in N = 2.0

100,000 portfolio. 1% risk, or $1000.

A) $25tradable. $2.48atr N = 2.0
B) $82tradable. $4.80atr N = 8.0

stock A.
$25 stock. @ 2.48atr = $5 stop. 200 shares

Stock B.
$82 stock @ 4.80atr =$38 stop. 26 shares

A) N factor of .10
B) N factor of .06

(A) is thus riskier than (B)!

Yes I said,â€