Length of Time to Reach 100% Equity Risk

Discussions about Money Management and Risk Control.
Post Reply
Stephen
Senior Member
Senior Member
Posts: 41
Joined: Tue Jul 18, 2006 12:17 am
Location: Jakarta, Indonesia

Length of Time to Reach 100% Equity Risk

Post by Stephen » Thu Oct 12, 2006 11:36 pm

I need some guidance on a proposed strategy to reach 100% total equity risk quickly. I am assuming that reaching full risk quickly is a good thing -- one would want all the available funds working in your system as quickly as possible.

I use Trading Blox Professional. I am developing a DMAC system, ATR-stopped, with a maximum drawdown goal of 32%. Be aware that my global parameter settings are conservative, and I will be using a robo-broker, so my overhead per trade is high. I am running my tests over 20 years of data. Startup account is $65000, trading liquid NA futures.

In my system, in order to limit the maximum drawdown to 32%, I have to set my risk per trade to 1.6%. The problem is, with these settings, it takes 6 years of trading to reach 100% of total equity risk!

My solution is to increase the risk per trade in the first few months only, in order to reach 100% equity risk quickly, then back off to the 1.6% risk per trade for the long term. TB Pro tells me that if I initiate trading with 3% risk per trade I can reach 100% total equity risk in 6 months with a maximum drawdown in that period of 55%.

I am not trading yet. I am learning much from this site (one of the finest on-line communities on the Net!) and, of course, from endless test runs. I would appreciate any comments on the proposed approach, or alternate approaches I should consider.

Thanks,

Steve

dave3076
Roundtable Fellow
Roundtable Fellow
Posts: 87
Joined: Thu Dec 01, 2005 10:51 am
Location: London

Post by dave3076 » Fri Oct 13, 2006 5:18 am

Firstly, in regard to having all your funds working, i would rephrase that to having all your funds at risk! Think black wednesday,9 11,and wall street crash! Also i would be surprised if any system could achieve a true drawdown figure of 32% over all liquid markets with 1.6% risk per trade. I would hesitate a guess that 1985 to 1991 were superb years for your system, which boosted your notional account to levels where the 25% of volume filter kept you out of full position sizing.Run the test with the same account size used to achieve 32% drawdown with 1.6% at risk with the test start date stepping every 80 days! am i right?
In regard to a general point you highlight.....it sounds like you`re testing with an account size that is to small to be position sizing correctly,hence tb is keeping you out of trades with to higher risk for the account size. A quick way to find a pretty good measure of what the lowest account size you need to trade all signals that a system produces is test with say $5,000,000. Look at how many trades you have every year with a $5,000,000 account and then run the same test using a decreasing account size until you hit a level where the trades per year begin to decrease noticably. When you find the lowest figure you need to take all trades/same amount as the 5 million account, then move onto comparing it with the 5 million account using the test start date stepping to ascertain if the annual trades stack up through differing time parameters. If it does you have a good guide over 20 years of data of what the lowest account size needed to trade ALL signals from this particular system. Of course you can trade with a lower account size than optimal, however your risk increases AND not all signals will be entered and your results will be distorted....sometimes for the better and sometimes for the worse. hope this helps

nickmar
Roundtable Knight
Roundtable Knight
Posts: 191
Joined: Tue Oct 26, 2004 12:38 pm
Contact:

Post by nickmar » Fri Oct 13, 2006 12:17 pm

In an attempt to keep risk relatively constant over time (due to the disparity between the number of available markets in the 70s and 80s vs. the number of available markets today), you may want to consider adjusting your position-sizing algorithm to incorporate the number of markets available for trading. The system.tradingInstruments property provides you with the total number of instruments in the portfolio being tested that have trade data, are primed on the current date and are set to allow trading by the Portfolio Manager.

Stephen
Senior Member
Senior Member
Posts: 41
Joined: Tue Jul 18, 2006 12:17 am
Location: Jakarta, Indonesia

Post by Stephen » Sat Oct 14, 2006 2:24 am

Re-reading my submission a day later I already see a basic problem in my analysis: I need to derive my maxiumum drawdown for the startup period from many 6 month 'start up' periods at 3% risk and with the given starting balance.

Thanks to Nick and Davey. Nick, I only have TB Pro so, as I understand it, I cannot change blox properties.

Steve

dave3076
Roundtable Fellow
Roundtable Fellow
Posts: 87
Joined: Thu Dec 01, 2005 10:51 am
Location: London

Post by dave3076 » Sat Oct 14, 2006 10:43 am

i would still research the minimum account balance you need to trade all the signals your system produces. Because i would be very very surprised if a system that trades all NA liquid futures with an account balance of $65,000 could take all the signals the system produces. Run the system with an account of $5,000,000 and check how many trades you had in the first year and second and third etc, and then run the system with $65,000, and you will see that you would have been kept out of numerous trades, hence your risk/heat would have been lower on paper, but in reality you wouldnt have really been trading the system you designed.

Stephen
Senior Member
Senior Member
Posts: 41
Joined: Tue Jul 18, 2006 12:17 am
Location: Jakarta, Indonesia

Post by Stephen » Mon Oct 16, 2006 12:33 am

I ran the system (see system description in the first post) adjusting the following: 1) startup account size, 2) trading universe, and 3) risk per trade – taking note in each case of the time taken to reach 100% equity risk. It is easy to see the rejected trades in Trading Blox Professional under the Trades tab, and the graph that plots Total Risk Profile reflects this. Again, the issue I have is how to get all of my capital working as quickly as possible. Here are the results:

TIME TO 100% EQUITY RISK

Varying Startup Account Size:
$65,000 - 6 years
$100,000 - 3.5 years
$150,000 - 1.6 years
$250,000 - 10 months
$500,000 - 6 months
$1,000,000 - 6 months
$5,000,000 - 6 months

Varying Trading Universe (subsets of liquid NA futures):
very low to medium volatility - 16 markets - 16 years
very low to high volatility - 21 markets - 12 years
very low to very high volatility - 26 markets - 10.5 years
all liquid markets - 31 markets - 6 years

Vary Risk Per Trade:
1.6% - 6 years - 33% maximum drawdown (DD)
1.8% - 5.5 years - 35% DD
2.0% - 3 years - 40% DD
2.5% - 1 year - 46% DD
3.0% - 6 months - 54% DD

What to make of all this? What to do?
1. As Davey has suggested, figure out a way to start with a larger initial investment. $500,000 seems the magic number for this particular system.
2. Expand the trading universe to include many more markets. My test data from CSI incorporates only the NA futures, so I cannot test this. I wonder how the ‘time to full equity risk’ would change if I included a significant number of liquid non-NA markets into the mix?
3. Start the engine with more heat, then back off. Easy to do on paper, but if there is one lesson coming through clearly on this forum, it is not to overestimate one’s tolerance for risk. Using the rule of thumb ‘prepare for 50% more than your maximum drawdown’ then at 54% DD this is an 81% DD worst case scenario. And I already know the other rule of thumb ‘nothing is certain’.
4. Trade an alternate system. This system is a double moving average crossover system. In my testing so far I have not found another system (within the built in set available with TB-Pro) that brings available capital to work quicker than the DMAC. Am I missing something?

Thanks,

Steve

dave3076
Roundtable Fellow
Roundtable Fellow
Posts: 87
Joined: Thu Dec 01, 2005 10:51 am
Location: London

Post by dave3076 » Mon Oct 16, 2006 2:43 am

Steve.....

I would seriously go back to the drawing board with your thinking in terms of risk and having your money working for you. This is not how the game is played. I can see how you have come to this conclusion but if you have 100% total equity risk, how are you going to meet margin calls? What if a global catastrophic event happens whilst 100% of your capital is tied up at various levels, like black wednesday? Read the threads by other members about the days they lost 50%+ of their accounts. As for testing optimal minimum account balance, test from 250k to 500k. Then once you have done it, you need to test different start dates. Just because the 500k account took all the trades the $5 million account did starting in 1985, doesnt mean if you started in 1986 it would be the same.Its likely they would be similar figures but just check. A possible solution to this problem is 2 fold.....1.trade mini`s 2. Financial spread betting. The latter will allow you to position size to the penny. Just re-consider your thinking on risk, because i hate to say it, but you are heading for trouble. Better to say it than live it though. Goodluck....

Stephen
Senior Member
Senior Member
Posts: 41
Joined: Tue Jul 18, 2006 12:17 am
Location: Jakarta, Indonesia

Post by Stephen » Mon Oct 16, 2006 4:12 am

Thanks Davey, I appreciate your comments. I just want to make sure we are on the same page with regard to Total Risk, so I have attached the definitions provided in the Trading Blox Professional Help File

It is my understanding, through reading this forum and taking the Modus Trading Course, that with a well diversified, non-correlated portfolio, risk can be effectively controlled through position sizing as a percent of portfolio, even with 100% of the portfolio invested in markets. Trading Blox simulates margin calls in its historical testing.

Maybe a Trading Blox veteran could help us out here on definitions and such.

I am certainly willing to admit I've got it all wrong.

On running system tests using stepped start dates: I agree that this is important and I do this in all my testing.

Thanks,

Steve
Attachments
Total Risk.doc
(31 KiB) Downloaded 902 times

dave3076
Roundtable Fellow
Roundtable Fellow
Posts: 87
Joined: Thu Dec 01, 2005 10:51 am
Location: London

Post by dave3076 » Mon Oct 16, 2006 5:12 am

Steve

Yes i agree that diversification can smooth out seriously bad periods, however, it is still not a good idea to have all your trading equity in the market at one time. Even though a diversified portfolio may be uncorrelated in the way that one commodity is doing one thing as another commodity is doing another, you should check volatility correlation as opposed to directional around key past events. The ripples are felt through some to most of the portfolio. Also in regard to tb`s ability to simulate margin calls, it only does so in a fixed manner. i.e. It doesnt make a copper contracts margin explode from $1000 to $10,0000, or crudes from 1200 to 7000. With me? Dont forget aswell that to my knowledge, tb doesnt account for maintenance margin. If you`ve got 100% of your account at risk, where is the maintenance margin coming from? I may be wrong on that one but i see no reference to maintenance margin in any kind of settings.

Stephen
Senior Member
Senior Member
Posts: 41
Joined: Tue Jul 18, 2006 12:17 am
Location: Jakarta, Indonesia

Post by Stephen » Tue Oct 17, 2006 12:16 am

I am committed to a system development approach based strictly on testing potential systems against historical price data. An important assumption I make is that the 'worst case' scenarios and the 'best case' scenarios are contained in the historical data. Monte Carlo simulations in Trading Blox allow me to see these more clearly, so that I can determine what I would be comfortable with in the future over the long term. I know nothing is for certain, but this is, for me, the best of the uncertain approaches. And I admit that I am on a learning curve here.

So, Davey, your suggestion regarding 'checking volatility correlation around key past events' doesn't click with me. If I decided to take such an approach, then I am immediately faced with a subjective decision: How do I decide what past events are the key ones? Your approach is different, and it obviously works for you, and I respect that. But it is not for me.

As to your comments regarding margins within Trading Blox -- it is probably best that Tim or c.f. respond to this. I trust that margins are handled in the most feasible way in Trading Blox.

Back to the subject of this post. Does anyone else have suggestions/experience on how to quickly ramp up the efficient use of capital in a small start up account? (Note: I have read the earlier excellent threads on this site regarding small accounts).

Thanks,

Steve

dave3076
Roundtable Fellow
Roundtable Fellow
Posts: 87
Joined: Thu Dec 01, 2005 10:51 am
Location: London

Post by dave3076 » Tue Oct 17, 2006 1:57 am

steve

This is what i know to be true......yes, you are right to base your decisions on market data, and yes you are right to look at how your system performed in all kinds of markets in terms of market conditions, and yes you are right that if it performed ok in all differing conditions then this is the best we can hope for in terms of its future performance. However, if you are trading with 100% total equity risk for the duration of 20 years, then there are certain distortions in testing that apply to you. Distortions that have a ripple effect down to the very core of the entire system. They are diluted to those with total equity risk of say 40-50%. And it brings us back to margin. Tb does simulate margin calls. However, it doesnt simulate rising and falling margin figures. You put your margin figure into the section in futures dictionary and it stays static for the duration of the test. Theres is the option to select a margin interest rate in global parameters, but tb doesnt make margins go up and down like they would in real life. So when margin does explode in a certain commodity or rises steeply, where is the money going to come from to support that position?.Are you going to close out? what effect may that have had to your system? what are your contigency plans if something like that happens?what if it happened to the commodity that was involved in the biggest move of the year?. So your clinical approach and terminology in the first half of your message are only good and true if you simulate true conditions. And you need to, because these are conditions that are very applicable to you. They are not so applicable to those with space in their accounts to have had a reserve of capital, so knowing that kind of passifies that effect from testing to real life. Anyway, we can agree to disagree, and i wont interrupt your original topic anymore. Goodluck....

Stephen
Senior Member
Senior Member
Posts: 41
Joined: Tue Jul 18, 2006 12:17 am
Location: Jakarta, Indonesia

Post by Stephen » Tue Oct 17, 2006 3:25 am

Thanks. All points well taken! And I appreciate the effort you've made to respond and explain.

Steve

sluggo
Roundtable Knight
Roundtable Knight
Posts: 2986
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo » Tue Oct 17, 2006 9:50 am

It's possible to construct systems for which "100% Total Equity Risk" is not a particularly useful or meaningful observation. I'll give an example here; it may help you crystallize a general principle.

This is the Trading Blox Builder builtin system "Dual Moving Average", applied to a portfolio of 29 futures markets. A stop is used, however it is placed rather far away from the entry point (ten full ATR's away!!), so the stop is very seldom hit. Rather, trades are exited when the two moving averages cross the other way.

I've run the system twice. Once with the stop at 10 ATR's and a risk per trade of 1.5%, and then a second time with the stop 4X farther away and the risk per trade 4X bigger. We expect these two variations to give the same performance because the system uses the ratio (risk% / StopInATRs) in its calculations; and these two variations have the same ratio: (1.5/10) = (6.0/40). Thus we expect them to take the same trades at the same prices on the same days using the same number of contracts.

And, as expected, the TBB results indeed are essentially identical (see below); these are essentially the same system with essentially the same performance. (The only difference is that approx 2 trades hit the 10 ATR stop before the MA's cross, whereas approx zero trades hit the 40 ATR stop before the MA's cross).

However, look at the "Total Risk" plots. They are very different! Yet these two systems are essentially the same! I wouldn't blame you for concluding that, in this case on this system, the Total Risk plot is not meaningful.

+SLUGGO+
Attachments
part2.png
Run it again with stop 4X farther away and risk-per-trade 4X bigger
part2.png (37.95 KiB) Viewed 16265 times
part1.png
The Dual MA system with a stop at 10 ATRs and risk-per-trade equal to 1.5%
part1.png (41.17 KiB) Viewed 16265 times

AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2040
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner » Tue Oct 17, 2006 1:21 pm

Rising open equity and a stop which does not move may increase theoretical on-going risk above start trade risk. Not that that in any way detracts from the point Sluggo makes.

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

Post by Forum Mgmnt » Tue Oct 17, 2006 5:08 pm

AFJ Garner wrote:Rising open equity and a stop which does not move may increase theoretical on-going risk above start trade risk.
This is very often the case when looking at long-term trend-following systems. I am certain the graph is correctly showing the fact that there is a lot of open position profit that could theoretically go away were the markets liquidated at the stops.

I do sometimes find the graph to be useful when comparing shorter-term systems where the stops are much closer to the market and much more likely to be the reason for an exit.

- Forum Mgmnt

Tim Arnold
Site Admin
Site Admin
Posts: 8540
Joined: Tue Apr 06, 2004 1:41 pm
Location: Boston, MA
Contact:

Post by Tim Arnold » Tue Oct 17, 2006 11:24 pm

I thought it would be interesting to look at the different components of total risk. This is the DMA system mentioned above, with 28 futures trading. It uses the same large stop (10 ATR) for illustration purposes.

The graph shows three things: 1) Total Risk, Open Equity Risk (distance between close and entry price), and Stop Risk (distance between entry price and stop). The definitions and code are here:

Code: Select all

totalRisk = test.equityRisk

openEquityRisked = ( test.totalEquity - test.closedEquity ) / test.totalEquity * 100

riskToStop = totalRisk - openEquityRisked
Note there is about an equal amount of risk in open equity as there is from the initial stop based entry risk.
Attachments
riskComponents.jpg
riskComponents.jpg (120.11 KiB) Viewed 16170 times

AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2040
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner » Wed Oct 18, 2006 3:32 am

sluggo wrote:It's possible to construct systems for which "100% Total Equity Risk" is not a particularly useful or meaningful observation.
+SLUGGO+
Sluggo is an excellent teacher. He likes to provoke thought and experimentation in those who listen to him. He also knows that it is possible to construct systems where the graphing of total equity risk is both useful and meaningful. It makes it possible to look at portfolio heat in a more constructive way than simply considering start trade risk.

sluggo
Roundtable Knight
Roundtable Knight
Posts: 2986
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo » Wed Oct 18, 2006 10:01 pm

AFJ Garner wrote:... graphing of total equity risk ... makes it possible to look at portfolio heat in a more constructive way than simply considering start trade risk.
In my own personal case, I haven't found a lot of utility in looking at a plot of total equity risk. In fact I've deselected that plot in my TBB Preferences. Others feel oppositely, as is their right.

The one circumstance in which I did use and profit from a Total Risk plot, was when I simultaneously plotted (a) Total Risk and (b) Number of Active Positions, on the same graph. This makes it possible to see whether changes in total heat were caused by changes in the number of positions, or changes in the distance-to-stops. (Change in distance-to-stops can be caused by either movement of price, or movement of the stop, or both.) This 2-in-1 plot provides the system developer many opportunities to scream unkind things at his monitor, such as "Holy moly, we should be tightening our stops right HERE, but we're not. Let's dig in and see why..."

Post Reply