Position Sizing Confusion - Help Please!

[Jetson]

Hi,

I am currently looking to design a trading system and am reading various literature on all aspects of this. One concept that has confused me a little is the % volatility model of position sizing (ps). My current understanding is based on the Van Tharp explanation and my issue is this.....

My ps approach is currently the % risk model and in my case 1R = 2%. This model makes sense to me because my ps is based on tech/analysis to set a stop-loss which using my 1R determines my maximum allowable position.

Van Tharp when describing the % volatility model shows how to determine the ps but there is no mention of a stop (other than the system's 20 day low stop). If for example the ATR is $1 and 1R = 2% and assuming capital of $100,000 I can allocate $2,000. $2,000 / $1 = 2000 shares. Let's say these shares are bought at $30 and at the time the 20 day low is $26. From what I understand (which I am not sure of) the stop is set at $26. Assuming that the stop is hit and I lose 2000 shares X $4 my loss and therefore risk when I entered the trade is $8,000 resulting in a risk of 8% rather than 2%.

Any confirmation of or correction in my understanding is much appreciated.

 

[cynic]

Re: Position Sizing Confusion Help Please.

Hi,

I am currently looking to design a trading system and am reading various literature on all aspects of this. One concept that has confused me a little is the % volatility model of position sizing (ps). My current understanding is based on the Van Tharp explanation and my issue is this.....

My ps approach is currently the % risk model and in my case 1R = 2%. This model makes sense to me because my ps is based on tech/analysis to set a stop-loss which using my 1R determines my maximum allowable position.

Van Tharp when describing the % volatility model shows how to determine the ps but there is no mention of a stop (other than the system's 20 day low stop). If for example the ATR is $1 and 1R = 2% and assuming capital of $100,000 I can allocate $2,000. $2,000 / $1 = 2000 shares. Let's say these shares are bought at $30 and at the time the 20 day low is $26. From what I understand (which I am not sure of) the stop is set at $26. Assuming that the stop is hit and I lose 2000 shares X $4 my loss and therefore risk when I entered the trade is $8,000 resulting in a risk of 8% rather than 2%.

Any confirmation of or correction in my understanding is much appreciated.

If the stop is to be set at $4 below the purchase price, then surely one would calculate the purchase quantity as (($2,000/$4) = 500) shares thereby resulting in the appropriate risk level of 2%.

Alternatively, if you believe ATR of $1 to be a suitable determinant for stop placement, then you might consider a stop at $29 thereby allowing the purchase of 2000 shares at $30. Bear in mind that many traders base their stop level on ATR multiples greater than 1 and have sound reasons for doing so.

Whichever stop level is considered (whether ATR multiple or historical low), backtesting of your strategy might give a better indication of the more suitable stop level methodology.

I hope this helps.

 

[Jetson]

Re: Position Sizing Confusion Help Please.

If the stop is to be set at $4 below the purchase price, then surely one would calculate the purchase quantity as (($2,000/$4) = 500) shares thereby resulting in the appropriate risk of 2%

Hi Cynic,

Thanks for the reply.

The quote above is precisely my question. I would use (($2,000/$4) = 500) under the ps system I currently use (I.e % risk model) but I am trying to understand How the % volatility model works. The ATR/historical low was just using the basis of van tharps explanation. Upon reflection I suppose my question really is......

Using a % risk model of PS my initial stop forms part of the process. I.e purchase price - stop price is my per share risk and I apply that to my 1R to determine position size. When using a % volatility model how is the stop determined?

Hopefully that is a bit clearer.

 

[cynic]

Re: Position Sizing Confusion Help Please.

Apologies for the misunderstanding.

I haven't examined Van Tharp's models nearly as extensively as yourself.

The only examples I've witnessed to date seem to indicate that the ATR is simply divided into the total amount risked (i.e. 2%) to determine number of shares/contracts.

So now I'm confused - are you trying to integrate 2 different models?

And if so, what has led you to believe that there is compatibility between these models?

I may still be off track here, so, hopefully other ASFers will jump in with some insightful answers.

 

[howardbandy]

Hi Jetson --

Dr Tharp has some very good ideas, and I recommend his books. But he, along with Ralph Vince, over simplify position sizing. They use single-valued results derived from the set of trades being used to establish position sizing. And they assume the system continues to be healthy for the entire period.

I recommend analysis of the distribution of results rather than the single value. This provides:
1. A much better estimate of the future performance of a system.
2. A metric for determining whether a system is healthy or is broken.
3. A method for deciding position size in relation to the traders personal risk tolerance.

My latest book, "Modeling Trading System Performance," describes my thoughts and provides free tools so you can do the analysis yourself.
http://www.modelingtradingsystemperformance.com/

Before committing to buy the book, you can read several articles I have written and posted on my website that will introduce the ideas, methods, and results.
http://www.blueowlpress.com/WordPress/

Also, the book's website has several chapters of the book you can download and read.

--------------------

Personal risk tolerance and position sizing are intimately related; and they depend on the distribution of trades and on ongoing system health.

The correct position size for a system that is broken is zero.

----------------------

A word of caution when reading Dr Tharp's "Definitive Guide to Position Sizing." All of the examples are over simplified and most of them are unrealistic. In the text, he talks about trading systems that have system quality numbers above 3, and suggests that people designing systems expect to be able to find systems with SQNs of 10. SQN is t-score.

t-score can be applied to any metric, but it is commonly applied to geometric profit per trade. A t-score associated with a set of trades is a measure of the degree of confidence that the mean of the gains of the set is higher than zero. A very high quality system will have a t-score (or SQN) of 2. Assume there is a set of 60 trades that have a t-score of 2.00. Referring to a table of t-scores, with 97.5% confidence the mean is greater than 0.

Try this. Use my tools to:
1. Generate a set of trades with whatever performance you want to test. Say one with a t-score of 2.00. Taylor the set to reflect your preferred trading style -- whether that is trend following with low percentage winning trades and high ratio of gain to loss, or short holding periods with higher percentage of winning trades and lower ratio of gain to loss. The tool works equally well with either or any other.
2. Analyze the distribution of profit and drawdown that can be expected over the next, say, 100 trades.
3. Decide on your personal risk tolerance.
4. Determine the position size appropriate for this system and your risk tolerance -- they go together.
5. Estimate the future performance, particularly the equity curve, final wealth, and maximum drawdown, for the next 100 trades.

If the system has a t-score of 2.0 throughout the 100 trade period, you will become unbelievably wealthy. Higher t-scores (or SQNs) are neither necessary nor achievable. Do not let perfect become the enemy of good.

There are several threads in ASF containing thoughts on position sizing, Dr Tharp's ideas, and my comments. Please read them before tilting at windmills.

-------------------

Thanks for listening,
Howard

 

[Jetson]

Re: Position Sizing Confusion Help Please.

Apologies for the misunderstanding.

I appreciate your time in answering and the misunderstanding is no doubt due to me being unclear.

I haven't examined Van Tharp's models nearly as extensively as yourself.

The only examples I've witnessed to date seem to indicate that the ATR is simply divided into the total amount risked (i.e. 2%) to determine number of shares/contracts.

From what I have read this is my understanding also however in the example I gave the ATR is $1 so that led to a position size of 2000 but didn't decide on a stop loss figure (the ATR was $1 but the stop loss ended up being $4). This is where I am confused as the stop loss is incorporated and controls the 1R in the model I use but in the % volatility method it doesn't seem to factor in a stop loss price (using the ATR). Am I missing something?

So now I'm confused - are you trying to integrate 2 different models?

And if so, what has led you to believe that there is compatibility between these models?

No, just trying to understand different methods

Thanks again for your help.

 

[Boggo]

Re: Position Sizing Confusion Help Please.

I appreciate your time in answering and the misunderstanding is no doubt due to me being unclear.

From what I have read this is my understanding also however in the example I gave the ATR is $1 so that led to a position size of 2000 but didn't decide on a stop loss figure (the ATR was $1 but the stop loss ended up being $4). This is where I am confused as the stop loss is incorporated and controls the 1R in the model I use but in the % volatility method it doesn't seem to factor in a stop loss price (using the ATR). Am I missing something?

I think that you are overcomplicating the whole process. The ATR varies so much from one stock to another that the only time it seems to be useful is to give you a picture of where the stock behaviour is in relation to its normal movement.
Even then you need to be looking at an ATR in excess of 2 x ATR otherwise you may find yourself getting stopped out unecessarily.

Develope the ability to "see" what applies to any individual chart.

Below are two charts of RSG, a trade that I went long on yesterday.
The top chart shows a potential target area and an "eyeball" exit should it all go wrong. Based on the entry and exit and using my $ at risk (< $1000 inc brokerage) I can plot a scenario with the R/R etc.

The bottom chart is my Metastock scan chart, the red line is a 2xATR and can be used as a barometer once in the trade.
In the case of this entry the 2xATR ($1.80) is too close to the action, I could buy a lot more shares but stand a greater chance of getting stopped out so I prefer to "eyeball" the exit trigger.

Don't spend too much time on complicated books etc, paper trade a few commonsense entry and exit scenarios.

Just my

(RSG - click to expand)

 

[Wysiwyg]

Re: Position Sizing Confusion Help Please.

From what I have read this is my understanding also however in the example I gave the ATR is $1 so that led to a position size of 2000 but didn't decide on a stop loss figure (the ATR was $1 but the stop loss ended up being $4). This is where I am confused as the stop loss is incorporated and controls the 1R in the model I use but in the % volatility method it doesn't seem to factor in a stop loss price (using the ATR). Am I missing something?

The ATR multiple is the stop loss! If you choose a multiple of the ATR of 3, as suggested by Tharp, then whatever that amount is your stop loss. You don't want to use the ATR because you will get stopped out often because it is the average range of the stock price so you use a multiple as per Tharps explanation.

 

[Boggo]

Below is an example of using a 2xATR on TLS, it just gives you an eyeball view of where a profit "lock in" may be warranted.

(click to expand)

 

[Wysiwyg]

Scroll down on this page and you will find a Position Sizing Spreadsheet Template for private use & copyrighted by Bulkowski.

http://www.thepatternsite.com/MoneyMgmt.html

 

[Jetson]

Re: Position Sizing Confusion Help Please.

Thanks for the response boggo,

I think that you are overcomplicating the whole process. The ATR varies so much from one stock to another that the only time it seems to be useful is to give you a picture of where the stock behaviour is in relation to its normal movement.
Even then you need to be looking at an ATR in excess of 2 x ATR otherwise you may find yourself getting stopped out unecessarily.

Develope the ability to "see" what applies to any individual chart.

At the moment with the % risk model I use I do eyeball the chart and apply my indicators (ATR is not one of them) to set my stop. The example was based on info in the book and I wasn't able to identify how a SL was set with the % volatility method.

Below are two charts of RSG, a trade that I went long on yesterday.
The top chart shows a potential target area and an "eyeball" exit should it all go wrong. Based on the entry and exit and using my $ at risk (< $1000 inc brokerage) I can plot a scenario with the R/R etc.

The bottom chart is my Metastock scan chart, the red line is a 2xATR and can be used as a barometer once in the trade.
In the case of this entry the 2xATR ($1.80) is too close to the action, I could buy a lot more shares but stand a greater chance of getting stopped out so I prefer to "eyeball" the exit trigger.

Don't spend too much time on complicated books etc, paper trade a few commonsense entry and exit scenarios.

Just my

(RSG - click to expand)

I am interested in your decision point that you used. Was it set because 2.9R is your historical expectancy or is the target based on TA?

Setting the target is another area I am still working on.

 

[Jetson]

Re: Position Sizing Confusion Help Please.

The ATR multiple is the stop loss! If you choose a multiple of the ATR of 3, as suggested by Tharp, then whatever that amount is your stop loss. You don't want to use the ATR because you will get stopped out often because it is the average range of the stock price so you use a multiple as per Tharps explanation.

Thanks Wysiwyg,

This makes sense but I have just re-checked the section on % volatility method and it dosent mention 3x ATR. It may have earlier in the book though that I forgot.

It's funny, this was the only section (about 2 pages) in the book that I didn't "get". Starting to feel a bit dense! Lol. I think I will just stick to the % risk model

 

[Boggo]

Re: Position Sizing Confusion Help Please.

I am interested in your decision point that you used. Was it set because 2.9R is your historical expectancy or is the target based on TA?

Its a target based on a form of pattern analysis that always creates a bit of fun around the campfire so I will leave it out of this discussion.
There are a few on here who understand the concept.

Cheers.

 

[peter2]

Volatility model (also risking 2% account).
You need to determine the current volatility over a fixed period eg. ATR(14d).

In your example if the ATR(14) is 1.00, then you may purchase $2000/$1 = 2000 shares using this model. Your SL is $1.00 below your entry price. Small risk sizes give bigger R multiple winners, BUT are much harder psychologically to trade consistently. In his book Van Tharp suggests reducing the FF% from 2% to 1% or even 0.5% when using volatility. Systems with tighter SL's generally have a much lower W%. This makes them psychologically harder to trade.

Most traders cannot trade with a tight SL so use a multiple of the ATR value (eg, 2, 2.5, 3 ... ). They are essentially placing their SL where they feel comfortable rather than using the current volatility. (I'm including back-testers in this group also.)

It is worth understanding that volatility cycles between high and low. Increases in volatility are associated with rising and falling prices. If you trade setups formed when volatility is low then a position sizing model based on volatility will earn much more profit over time than a ps model based on a FF%. This is clearly shown in the Van Tharp's book.

Most people (other than options traders) wouldn't know what the current price volatility is when they buy. Does it make a difference? Check Van's example again.

 

[Gringotts Bank]

Is it possible to stick to the 2% rule without having to use leverage to take all the positions you want?

I guess one just reduces the % at risk.... yes?

 

[peter2]

It depends on the size of the initial risk of each trade.

eg 2% of 100K = 2000
Trade 1: Buy CBA at 65 with an iSL of 64 (Risk = 1) # shares = 2000/1 = 2000 x 65 = 130K (not enough capital)

Trade 2: Buy CBA at 65 with an iSL of 60 (Risk = 5) # shares = 2000/5 = 400 x 65 = 26K (OK can start another trade)

 

[Gringotts Bank]

It depends on the size of the initial risk of each trade.

eg 2% of 100K = 2000
Trade 1: Buy CBA at 65 with an iSL of 64 (Risk = 1) # shares = 2000/1 = 2000 x 65 = 130K (not enough capital)

Trade 2: Buy CBA at 65 with an iSL of 60 (Risk = 5) # shares = 2000/5 = 400 x 65 = 26K (OK can start another trade)

I guess my stops are fairly small. Makes it hard. Thanks.