Money management - relationship between profit and position size

[bassmann]

I have a mechanical system I trade that has some money management built in. I am looking for some ideas on possible next steps to improve this where possible.

My Money Management
I currently use a fixed fractional method that has 5 equal size positions so a 100k account would have 5 * 20k positions when fully committed. I have a trailing stop that is also used as my Initial Stop at my Entry that is volatility based, and on average would be around 15%. This might sound extremely wide for some of you but I have found tightening it produces adverse system results.

RISK
Based on these money management parameters I am risking 3% per trade and my aim is to reduce my exposure to 2% per trade and maintain system performance if possible. The obvious options to me are a) tighten my trailing stops (which I have not been able to do effectively) or b) increase number of positions in my portfolio (which I have found lowers the systems profitability somewhat).

SIMULATION OBESERVATIONS
I have compared results from 2 to 15 and found;
• Win/Loss ratio does not change much until 5 or less positions where it begins to increase.
• System profit is cut in half when increasing positions from 5 to 10.
• Days for Winning Trades and Days for Losing Trades does not vary with number of positions
• Profit factor increases considerably with less positions
• Drawdown increases with less positions but is more dramatic for simulations of < than 5.

QUESTIONS
1) What is the relationship between position size and profitability of my system? What should I look at next?
2) Is it ‘always’ bad to risk more than 2% per trade in all cases? Are there exceptions where this would be feasible, if so when? Should I be concerned about 3% in my system or am I over-reacting?
3) Does anyone have any suggestions on what I can look at to reduce my risk beyond what I have described above? i.e. Different technique etc.
4) The above focuses on risk per trade. Am I correct in assuming that changing the number of positions does not alter the overall portfolio risk should the market tank. For e.g. 5 * 3% = 15% portfolio risk vs. 10 * 1.5% = 15%.
5) What is the impact of risk per trade on a margined portfolio? For e.g. My account had a 50% LVR so my 100k account now has 200k available to trade. I trade 5, 40k positions with a 15% Stop. Is my risk per trade still 3% or has is doubled to 6% because of leverage? Why?
6) Any other recommendations. Any suggested reading?

Thanks in advance

bassmann

 

[Nick Radge]

I currently use a fixed fractional method that has 5 equal size positions so a 100k account would have 5 * 20k positions when fully committed.

You don't have fixed fractional sizing if you have 5 equal positions.

 

[bassmann]

Thanks Nick, just reading up on this now.

Please ignore the "Fixed Fractional' bit. Rest of what I posed is accurate for my system.

bassmann

 

[Trembling Hand]

What is the impact of risk per trade on a margined portfolio? For e.g. My account had a 50% LVR so my 100k account now has 200k available to trade. I trade 5, 40k positions with a 15% Stop. Is my risk per trade still 3% or has is doubled to 6% because of leverage? Why?

Margin should not be used to increase the size of your positions as you will increase the risk per trade(3% to 6%) unless you tighten your stop by half. Which you have indicated that you don't want to.

 

[nizar]

Margin should not be used to increase the size of your positions as you will increase the risk per trade(3% to 6%) unless you tighten your stop by half. Which you have indicated that you don't want to.

TH, would it be right to say that a correct use of margin would be to enable you to take on more positions?

 

[Trembling Hand]

TH, would it be right to say that a correct use of margin would be to enable you to take on more positions?

Yeah. It doesnt matter if you are trading CFDs with 5% margin or Stocks fully purchased you are still risking the same amount per trade therefore you purchase the same amount of shares per trade. (or change your system to find trades with tighter stops)

So margin enables you to trade MORE position rather than more quantity(shares) of the same trade. (correlated risk aside )

 

[MichaelD]

Woohoo! An intelligent post.

1) What is the relationship between position size and profitability of my system? What should I look at next?

a. Go to fixed fractional position sizing.
b. Reduce position size to 1% or even 0.5%
c. Add to winning positions by pyramiding, taking the position size up to 2 - 2.5% when fully pyramided.

You will find that your results will change very significantly even with simply implementing point a.

2) Is it ‘always’ bad to risk more than 2% per trade in all cases? Are there exceptions where this would be feasible, if so when? Should I be concerned about 3% in my system or am I over-reacting?

Yes you should be concerned.

As position size increases, your risk of ruin becomes significant due to the sheer size of a given position. Once you get over 25% of your capital allocated to a single instrument, you become very vulnerable to black swan events.

You should set a maximum position size % to limit your exposure to one instrument - eg never expose more than 15% of your capital to one instrument.

Most books suggest 2% risk per trade and maximum 25% per instrument. I find these risk limits too high, and personally use 0.5% and 15%.

" Beginners will see 2% and think "so little!" and professionals will see 2% and think "so much!" "

3) Does anyone have any suggestions on what I can look at to reduce my risk beyond what I have described above? i.e. Different technique etc.

Explore pyramiding as mentioned above. It's much more challenging psychologically to trade as you get much longer runs of losers, but the losers are much smaller and the winners when they come are much bigger. The nett result is in favour of pyramiding.

4) The above focuses on risk per trade. Am I correct in assuming that changing the number of positions does not alter the overall portfolio risk should the market tank. For e.g. 5 * 3% = 15% portfolio risk vs. 10 * 1.5% = 15%.

Correct, but the market rarely tanks evenly, and a diversified portfolio suffers less drawdown than a concentrated portfolio.

5) What is the impact of risk per trade on a margined portfolio? For e.g. My account had a 50% LVR so my 100k account now has 200k available to trade. I trade 5, 40k positions with a 15% Stop. Is my risk per trade still 3% or has is doubled to 6% because of leverage? Why?

Your risk has doubled. The amount available to you with leverage is NOT the amount you use to calculate risk. You ALWAYS calculate risk on your base capital.

Your risk exposure is basically the maximum amount of YOUR capital that you are prepared to lose at any one time should all your stops get hit at the same time.

If you define your maximum risk exposure as 15%, and fully committing your capital exposes you to 10% risk, then you can use leverage to increase your risk exposure up to your predefined limit. That is the smart way to use leverage. You will note that using leverage this way never even comes close to the margin limits.

WARNING: A prudent trader is not likely to use leverage for at least 2-3 years. If you feel even the slightest twinge of regret at closing a position that has hit its stop loss then you are not ready for leverage. On the other hand, if it is simply illogical to you to not take a stop when it is hit, then you are probably ready to use leverage.

6) Any other recommendations. Any suggested reading?

Van Tharp - Trading Your Way To Financial Freedom

This introduces the concept of your risk per trade being defined as R, and your maximum risk exposure as an R-multiple. It is much more efficacious to consider risk in R-multiples than in % figures.

At a system level, ultimately, the maximum risk you are prepared to expose yourself to is directly related to the maximum drawdown that you are prepared to tolerate.

 

[MRC & Co]

Good post Michael in answering those questions.

On the maximum %s, I also find most books too risky for me, I prefer 0.5-1% and 15% as maximums, but this definately comes down to your own personality.

 

[tech/a]

SIMULATION OBESERVATIONS
I have compared results from 2 to 15 and found;
• Win/Loss ratio does not change much until 5 or less positions where it begins to increase.
• System profit is cut in half when increasing positions from 5 to 10.
• Days for Winning Trades and Days for Losing Trades does not vary with number of positions
• Profit factor increases considerably with less positions
• Draw down increases with less positions but is more dramatic for simulations of < than 5.

Strongly indicates that your system is profitable due mainly to one or 2 trades which could be viewed as outlier moves.
Remove the top 3 best trades and the worst 3 trades then look at your result. Less than satisfactory??

4) The above focuses on risk per trade. Am I correct in assuming that changing the number of positions does not alter the overall portfolio risk should the market tank. For e.g. 5 * 3% = 15% portfolio risk vs. 10 * 1.5% = 15%.

Read up on Portfolio heat.

I'm sure you'll find your system has serious problems with regard to consistent profit. I would suggest your equity curve is less than smooth.
Have you run this through Montecarlo analysis to see the standard deviation of results? This is where you'll see clearly my suspicions.

How many trades were taken?

Nizar
Correct use of margin is highly correlated to correct calculation of Fixed Fractional Position sizing.

 

[bassmann]

Thanks to all that have replied so far.

MichaelD
Lots of leads for me to investigate, all makes sense, next I will go off and do some testing / reading over the next few weeks which is bound to provoke further questions.

I also thought of trading 2 non-correlated systems but in the same market (i.e. ASX), e.g. a long trending system (my current one) and say a mean reversion system to reduce my market risk. Only thing is the 2nd system is still in development.

tech/a
I have re-run my backtesting without the best 3 / worst 3 trades. This does reduce the profit but it is still very satisfactory (i.e. I will continue to trade it).

SIMULATIONS (excl outliers)
The simulation starts from 1/01/06 to present with 157 trades processed.
I've run my simulation with 'Equal Dollar Units' and 'Pyramid Profits'. The equity curve is quite linear and picks up 139 trades.

Also run with 'Equal Percent Dollar Units' and 'Pyramid Profits' and its EC is smooth expotentially and picks up 66 trades.

MONTE CARLO
Ran a MonteCarlo, and the Profit STDEV on Equal Dollar Units is 2% of mean profit and 4% on 'Equal Percent Dollar Units'. Is this reasonable over 10,000 permutations? What would be a sound target? The % winning only changes by 2% and drawdown is fairyly consistent as well.

MONTHLY PROFIT
How even should my monthly profit be? How would one quantify this? The monthly profit varies a bit however there are only 3 losing months from jan-06to present. The months with zero profit represent when the system has switched off (i.e. market gone bearish). This feels reasonable to me considering the level of volitility is recent times.

My own conclusion is the profits are fairly consistent, but maybe I'm not considering the right metrics. Any thoughts?

bassmann

 

[tech/a]

• System profit is cut in half when increasing positions from 5 to 10.

Well I'm at a loss.
the only reason I can figure for a 50% decrease in profit with a 100% increase in number of positions is that the trades added were losers---big ones.
The 2% deviation in montecarlo sims doesnt compute taking the statement above into consideration.

Anyway if your happy with it keep on keeping on.

 

[bassmann]

As a next step I have run tabulated 4 simulations of my system (both sim and monte) as follows (see attached summary);

Simulations
1) Fixed Dollar, 5 equal positions
2) Fixed Dollar, 10 equal positions
3) Fixed Dollar Percent, 5 equal positions
4) Fixed Dollar Percent, 10 equal positions

The idea being 1 compares to 2 and 3 compares to 4. I have also removed the best 3 / worst 3 trades.

Observations
1) The profit change is reduced without outliers but is still a factor of 1.46:1 moving from 5 to 10 positions.
2) Scenario 3 (Fixed $ %, 5) has the lowest number of trades. Maybe my system does not have enough signals, so when I run 5 positions a greater % of my money is in the market and greater % of the time explaining increased profit.
3) Scenario 3 also has the greatest drawdown so my main focus needs to be managing my risk.

Running the above will be a good base for me to compare results when I try the suggestions you have all contributed.

Regards,

bassmann

 

[tech/a]

Why have you not included brokerage.
100 trades at $25 in and out = $5000 hows that effect your bottom line?

 

[nizar]

Bassmann and others.

I have attached an Excel file which may be of interest to you.
Looking at the relationship between Percent risk and Profit and also Percent risk and CAR/max.DD.

 

[bassmann]

hi tech/a - didn't include brokerage because I got halfway through and realised I had it switched off. I usually factor 0.12% of turnover this typcially is not a problem for 100 trades over 28 months on a medium account size.

hi Nizar, I assume this is off your system.

1) The best return here of 41.8% is based on 2.5% fixed risk.
2) 0.5% Fixed Risk produces lowest drawdown
3) 1.5% fixed risk offers the best ratio of DD to CAR.

I think I know where you are going with this. Back to more testing.

Bassmann

 

[howardbandy]

Greetings all --

In my opinion, THE expert on money management is Ralph Vince. He has published at least four books, the most recent of which is "The Handbook of Portfolio Mathematics." Amazon carries it:
http://www.amazon.com/Handbook-Port...=sr_1_1?ie=UTF8&s=books&qid=1210284037&sr=1-1

He has three earlier books, all are excellent.

Thanks,
Howard

 

[GreatPig]

Hi Howard,

THE expert on money management is Ralph Vince.

I just had a look at his books on Amazon. Is it possible for someone without a PhD in statistics to understand them? They look pretty high-powered.

Cheers,
GP

 

[howardbandy]

Hi GP --

There is some math, but nothing too scary. Being reasonably comfortable with algebra is necessary. Some background in probability and statistics will help. There is very little calculus.

Thanks,
Howard

 

[bassmann]

Hi there,

As part of my money management project, I also wish to revisit my stops.

Can anyone recommmend any books dedicated specificially to stops. Heavy reading is ok by me.

Bassmann

 

[howardbandy]

Hi there,

As part of my money management project, I also wish to revisit my stops.

Can anyone recommmend any books dedicated specificially to stops. Heavy reading is ok by me.

Bassmann

Hi Bassmann --

Based on my research, the best way to exit from a trade is through a signal. That signal might be based on the same logic that caused entry to the trade with the same parameter set, but in the opposite direction. Or by the same logic but with a different parameter set. Or by a completely different signal.

The second best way to exit depends on whether the typical holding period is long -- a few weeks or longer -- or short. If it is long, then consider a trailing exit (trailing stop) such as chandelier or parabolic that follows along below the trade and protects profit. If it is short, then consider a profit target or timed holding period. Neither timed holding periods nor profit targets work well with long holding periods; trailing exits do not work well with short holding periods.

The final way to exit a trade, and the least desirable one with the poorest performance by far, is the maximum loss stop.

Whichever ways are used to exit, and a system may have more than one, be certain that each is tested independently, occurs often enough to be at least somewhat statistically significant, and passes reasonable tests of significance. Be particularly careful to avoid selecting exits that are tuned to specific historical situations rather than general conditions.

Remember -- only out-of-sample results have value in estimating the future performance of a trading system.

Thanks for listening,
Howard