Equal position size

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
Bollinger
Senior Member
Senior Member
Posts: 40
Joined: Wed Apr 16, 2003 2:42 pm
Location: New York City

Equal position size

Post by Bollinger » Tue Apr 22, 2003 10:49 pm

Forum Mgmnt,

One of the interesting points that you have made both in the Original Turtles discussion and separately on this board is that it is important for a trader to maintain equal position size across the various positions she has on. The Original Turtles system does this by establishing the approximate volatility of each market at the time a trade is put on, determining the value of that volatility, and ensuring that at the time a trade is put on the value of that volatility multiplied by the number of contracts put on is equal to a fraction of the account equity value.

With a long term system, unfortunately this fails to ensure that all positions have equal weighting at any particular point in time. Why? Because over the course of a trade (1) the volatility of individual markets will change, and (2) the total amount of equity in the account will change.

As an example, assume the following facts. Trader has a $1,000,000 account on March 1, observes that a new June Widget futures position should be put on, and after observing that the average true range of the June Widget futures contract is 20 points decides to go long 5 contracts. Based on the point value set by the relevant exchange for Widget futures contracts, Trader has calculated that this position will risk 1% of equity, or $10,000. Trader is a long-term trader and will likely hold this position approximately 2-6 months.

Example (1): After 2 months, Trader's account has grown to $1,100,000 as a result of Trader's adroit trading. Trader still has the open Widget position. June Widget futures still have an ATR of 20, so that Trader believes the weight of the Widget position is still $10,000. Trader observes that it is time to put on a new June Buggy Whip futures position. However, because Trader's account has grown to $1,100,000, the new Buggy Whip position will be weighted at $11,000 to achieve 1% risk. The result is that the position weights will not be equal -- the Buggy Whip position will be more heavily weighted than the Widget position that was put on when the account was smaller.

Example (2): Instead, after 2 months Trader's account remains at $1,000,000. Trader observes that the volatility in June Widgets has doubled from an ATR of 20 to an ATR of 40. Thus, the weighting of Trader's Widget position is now 2% of equity, and will be substantially more than other positions put on at 1% of equity.

It would seem that the simple solution to these problems would be to periodically rebalance position sizes across the portfolio based on a renewed analysis of position weighting that considers the current equity level and individual market volatility. Have you put any thought into this?

Best,
Neal Stevens

blueberrycake
Roundtable Knight
Roundtable Knight
Posts: 125
Joined: Mon Apr 21, 2003 11:04 pm
Location: California

Re: Equal position size

Post by blueberrycake » Wed Apr 23, 2003 1:11 am

Bollinger wrote: It would seem that the simple solution to these problems would be to periodically rebalance position sizes across the portfolio based on a renewed analysis of position weighting that considers the current equity level and individual market volatility. Have you put any thought into this?
I was thinking the exact same thing. So I modified Aberration, whereby if any position increased to the point where it's weight in the portfolio was twice the initial allocation percentage, I would sell half the position to bring the risk in line. The result was a noticeable decrease in profits, with very little commensurate reduction in drawdowns.

-bbc

Chuck B
Roundtable Knight
Roundtable Knight
Posts: 469
Joined: Thu Apr 17, 2003 6:34 am

Post by Chuck B » Wed Apr 23, 2003 6:55 am

Hopefully the forum will soon allow uploads of images like LeBeau's forum. I have done a lot of research on open risk management as you are referring to, and I could post a few graphs to illustrate the results. With my long term systems, there is some gain to be made in reward to risk measures (i.e. Annual ROI versus max and average drawdown, portfolio heat, etc.) by using what amounts to open risk scale outs. I looked at the Tom Basso method of open risk control that puts a lid of pure open risk and also position volatility risk.

Forum Mgmnt
Roundtable Knight
Roundtable Knight
Posts: 1842
Joined: Tue Apr 15, 2003 11:02 am
Contact:

Post by Forum Mgmnt » Wed Apr 23, 2003 6:59 am

The problem with rebalancing is that it is in-effect changing the system itself.

Taking equal risks across each market and balancing based on volatility is a way of ensuring two things:
  1. Equal Initial Risk - The risks across each market will be the same in terms of money
  2. Equal Stops - The probability of getting stopped out is the same
The idea that seems to work is to take the same initial bet. Some will work out well and pay for the ones that don't. If you rebalance you are, in effect, changing this equation since you are truncating the profits for the positions that are winning. That is why I don't think it works at all. You end up cutting short your best winners.

This is very much the kind of thing that mutual funds do. It improves divesification but reduces profits (an unintended side-effect).

I haven't found a means of doing what you suggest that makes money.

There are ways to scale back positions that work, however, they are all based on the price action, not the return of the instrument in money terms, or percent of account terms. My research shows that adjusting the stops based on risk pays off in some situations, and this is probably analogous to Chuck's reply, but not just arbitrarily thinning positions.

Gerry Gunter
Full Member
Full Member
Posts: 14
Joined: Sun Apr 20, 2003 4:10 pm
Location: Washington State

"Open-Risk"

Post by Gerry Gunter » Wed Apr 23, 2003 7:19 pm

Open-Risk doesn't adjust stops based on risk. A trader would maintain a certain % of open risk based on the traders equity and use that as a way of scaling out, to keep there ongoing risk at a certain level.
Yes it can cut down on profits but it also cuts down on the drawdown and the trick for the trader is to decide on how much drawdown(Heat) he/she can handle.

Gerry

Forum Mgmnt
Roundtable Knight
Roundtable Knight
Posts: 1842
Joined: Tue Apr 15, 2003 11:02 am
Contact:

Post by Forum Mgmnt » Thu Apr 24, 2003 9:08 am

Gerry,

Have you found that this works (i.e. improves risk-adjusted returns) in your testing? Interesting.

Using what kinds of entries?

Gerry Gunter
Full Member
Full Member
Posts: 14
Joined: Sun Apr 20, 2003 4:10 pm
Location: Washington State

Open-risk

Post by Gerry Gunter » Thu Apr 24, 2003 10:49 am

Hey Forum Mgmnt,

Chuck can probably elaborate more on open-risk since he has graphs to illustrate it in action. It's one thing reading about it , but if readers could visually see Chucks graphs it will paint a clearer picture.
Open risk has nothing to do with entries. You trade your normal [entry] into the market and once in a successful trend and adjusting your [stop] according to your system,you can monitor your open risk and keep it at a certain level.
Hey Chuck--can you help me out here?

Gerry
Last edited by Gerry Gunter on Thu Apr 24, 2003 5:31 pm, edited 1 time in total.

Forum Mgmnt
Roundtable Knight
Roundtable Knight
Posts: 1842
Joined: Tue Apr 15, 2003 11:02 am
Contact:

Post by Forum Mgmnt » Thu Apr 24, 2003 11:31 am

I'm looking into what is required to add the ability to upload images to our server. I might have this in place by the end of the weekend. No promises, but I am working on it. I'd love to see those graphs.

If this works, you must be using a different concept than I tested. I'm very happy to find out that I missed something. That's one of the reasons I started this forum, because I figured I'd learn from the interchange with other successful traders.

The reason I asked about entry was because I am trying to figure out what kind of trend you are talking about, long-term, short-term, medium, etc.

- Forum Mgmnt

Chuck B
Roundtable Knight
Roundtable Knight
Posts: 469
Joined: Thu Apr 17, 2003 6:34 am

Post by Chuck B » Thu Apr 24, 2003 1:41 pm

From testing I've done on long term systems (i.e. something like a 70/17 breakout system, or a 55 in/trailing volatility exit, you get the idea), you can improve the reward to risk measures with open risk scaling out and volatility risk scaling out (i.e. the "Tom Basso" method, I say Tom because I first learned this from him about 10 years back at a seminar he was part of and he also published his method in a booklet the CME produced about becoming a CTA). By reward to risk measures, I am referring to Annual ROI as reward and for risk: peak and average drawdown, peak and average portfolio heat (open risk).

I recall looking at initial position sizing between the range of 1 to 2.5% of equity and then holding max allowable open risk to 2.5 times initial risk, so if initial risk was 1.5% of equity, open risk was limited to 3.75% of marked to market equity. If on the close, any position had a risk to tomorrow's exit that was greater than 3.75% of equity, its position size was reduced on the next open enough to keep the risk under this value.

I recall generally setting the volatility risk limit to an amount equal to the initial risk or thereabouts. Here volatility risk was calculated using a 10 or 20 day ATR. If on any day's close, the volatility risk of a position was greater than this amount, on the next open the position was reduced to stay under it.

I'm actually just recently getting back into additional research on my long term systems, but I'm drawing this stuff above from 5-6 year old memory banks here -- however, I think I've recalled it properly :) .

By the way, note that this was all done looking at end of day risk. Consider all the other possibilities such as intraday open risk targets (i.e. if open position risk today reaches "x times" initial portfolio risk, sell (buy) "y" contracts to hold max open risk to that limit). These types of scale outs can be generated as limit orders for each day's trading.

Obviously these types of algorithms are not conducive to shooting for the moon in trends like I view the turtles having done; they are most suited to smoothing the equity curves when managing large amounts of money and especially other people's money.

Chuck B

Post Reply