Money management model for multiple strategy trading method | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

BASKET TRADE Management: Using multiple trade management EA's on basket trades 11 replies

Whats your best money management method? 52 replies

Alternative Trade Management/Money Management Methods 3 replies

Money Management / Risk Management 24 replies

Most popular money management method. 7 replies

Money management model for multiple strategy trading method

  • #1
  • First Post: Edited 7:06am May 17, 2010 4:07am | Edited 7:06am
Hi traders,

I would like to post a discussion about how to work out a money management model for multiple strategy trading method. This is a rather open-end question. Let me elaborate more.

Suppose a trader uses a number of different setup types for his trading and each of his setup type has a different win rate, risk reward ratio and frequency of occurence, and all these setup types show a positive expectency. How should he set up his money management model, so that he know the position size he should use for each setup type.

To simplify the issue, I instantiate an example with number values, although mathematical calculation is not the main theme in this post. To simplify the example, assume that there are no breakeven using or partial profit taking.

Setup....Win...........Risk:................Freq. of
Type.....Rate..........Reward............Occurencey

1..........50%..........1:2 - 1:3..........High
2..........60%..........1:1 - 1:2..........Low
3..........75%..........1:1..................High
4..........80%..........1:1 -1:2...........Very Low

Questions are:

1. How should the trader decide the position size for each of these setup types?
2. What principles should he use when working out his money management model?
3. What else should he take notice of?


kk007
  • #2
  • May 17, 2010 3:59pm May 17, 2010 3:59pm
The goal of MM (staking) is to balance the maximizing of gains with the preservation of equity.

Stake size should be both small (to preserve equity) and consistent (to avoid the situation where one loss can eradicate multiple wins). A higher profit factor (= win rate x average historical RR) justifies higher staking. You could run a Monte Carlo technique over a large trade history to attempt to optimize your stake size.

However, in the absence of more information, I would simply use anything between 0.5% and 3% fixed frac in all of your examples. Fixed frac has its detractors but I have yet to find anything better. Its inherent compounding allows exponential growth, while dollar value of losses is reduced should drawdown increase. Other possible staking methods are discussed in this thread.
  • #3
  • May 17, 2010 4:49pm May 17, 2010 4:49pm
Quoting kk007
3. What else should he take notice of?
hey kk...... couple other items worth notice might be , for each setup,

the average length per trade ..... along with maximum and minimum....

the average negative excursion per trade..... along with maximum and minimum....

//----

....h
to trade and code, keep both simple... no call to impress....h
  • #4
  • May 17, 2010 5:33pm May 17, 2010 5:33pm
I'd set up the problem as follows:

with mixing coefficients c_1 c_2 c_3 c_4 (amounts/risk allocated to each strategy)

estimate the expected Sharpe ratio, expected win per unit time divided by expected standard deviation over that unit time. You may be able to do that with the information in the first post if you can quantify the 'trade frequency'.

Then optimize over the set of c_i's, probably subject to a global constraint on sum of c_i's to control overall risk. (This is pretty much mean-variance CAPM optimization from MBA finance 101)

The problem is dealing with cross-correlation in time & returns with the different strategies, they may not be all independent and so the overall variance would be larger than sum of individual variances.
  • #5
  • May 18, 2010 7:42am May 18, 2010 7:42am
Quoting hanover
The goal of MM (staking) is to balance the maximizing of gains with the preservation of equity.

Stake size should be both small (to preserve equity) and consistent (to avoid the situation where one loss can eradicate multiple wins). A higher profit factor (= win rate x average historical RR) justifies higher staking. You could run a Monte Carlo technique over a large trade history to attempt to optimize your stake size.

However, in the absence of more information, I would simply use anything between 0.5% and 3% fixed frac...
Thanks hanover,

I agree that consistency is one principle in money management. I have been thinking given setups of various RRR and winrate, how position size should be varied to optimize consistence. If one of the goals is to preserve captial, the lose rate should be used to adjust position size.

cheers,
kk007
  • #6
  • May 18, 2010 7:45am May 18, 2010 7:45am
Quoting hayseed
hey kk...... couple other items worth notice might be , for each setup,

the average length per trade ..... along with maximum and minimum....

the average negative excursion per trade..... along with maximum and minimum....

//----

....h
Thanks hayseed,

I think you suggest looking at more deeply into the management of each trade. Nevertheless, I think here the interests rest on the money management model and thus concentrate more on a series of trades.

cheers,

kk
  • #7
  • May 18, 2010 7:48am May 18, 2010 7:48am
Quoting mbkennel
I'd set up the problem as follows:

with mixing coefficients c_1 c_2 c_3 c_4 (amounts/risk allocated to each strategy)

estimate the expected Sharpe ratio, expected win per unit time divided by expected standard deviation over that unit time. You may be able to do that with the information in the first post if you can quantify the 'trade frequency'.

Then optimize over the set of c_i's, probably subject to a global constraint on sum of c_i's to control overall risk. (This is pretty much mean-variance CAPM optimization from MBA finance 101)...
Thanks mbkennel,

Your reply looks relevant. Nevertheless, there are not enough details for me trying to replicate your idea. Can you refer me to a website or other online reference for me to study a bit more, if possible?

cheers,

kk
  • #8
  • May 18, 2010 8:29am May 18, 2010 8:29am
I'm not sure what's being discussed tbh. I would suggest.

1-Split your account into 4 parts.

2-Trade each part as if it was a separate account, one strategy per account.

3-The better strategies will make more money and so you'll end up risking more on them, the weaker ones will grow less and so you'll be risking less as a %age of your total account.

4-Retire and send me my cut (5%).

There is a problem with this of course, the proportions can get completely out of whack, and you'll be risking thousands on one strategy and cents on another. This could lead to a large draw down of the overall account. This means you might want to redistribute the funds occasionally. You also have to decide how to original distribute the account, which might be what you were all discussing having assumed the rest.
  • #9
  • May 18, 2010 6:40pm May 18, 2010 6:40pm
Quoting TraderHotch
I'm not sure what's being discussed tbh. I would suggest.

1-Split your account into 4 parts.

2-Trade each part as if it was a separate account, one strategy per account.

3-The better strategies will make more money and so you'll end up risking more on them, the weaker ones will grow less and so you'll be risking less as a %age of your total account.

4-Retire and send me my cut (5%).

There is a problem with this of course, the proportions can get completely out of whack, and you'll be risking thousands on one strategy and cents on another....
Of course thing can be done in the way you described, but I am looking for something more robust and low-risk than this. As you said, there are some possible high-risk cases in your method.
  • #10
  • May 18, 2010 7:13pm May 18, 2010 7:13pm
I pay a lot of attention to the correlation matrix of my portfolio. Correlation Matrix should be closed to zero or even negative so that you can minimize the portfolio volatility.
  • #11
  • May 18, 2010 7:28pm May 18, 2010 7:28pm
Quoting ForexQuant
I pay a lot of attention to the correlation matrix of my portfolio. Correlation Matrix should be closed to zero or even negative so that you can minimize the portfolio volatility.
Thanks for your suggestion.

I haven't thought about correlation between the setup types yet. I have been thinking about the simpler cases that each of the trade is independent to others.

mbkennel's approach seems to relevant to what I am looking for, still waiting for his input.
  • #12
  • May 18, 2010 8:20pm May 18, 2010 8:20pm
Quoting kk007
Thanks for your suggestion.

I haven't thought about correlation between the setup types yet. I have been thinking about the simpler cases that each of the trade is independent to others.

mbkennel's approach seems to relevant to what I am looking for, still waiting for his input.
Just study the modern portfolio theory then you know what he tried to tell you. Portfolio weighting and correlation is the important elements in MPT. However my experience told me that correlation is much much more important than how you choose your portfolio weighting.
  • #13
  • May 18, 2010 8:53pm May 18, 2010 8:53pm
Quoting ForexQuant
Just study the modern portfolio theory then you know what he tried to tell you. Portfolio weighting and correlation is the important elements in MPT. However my experience told me that correlation is much much more important than how you choose your portfolio weighting.
Thanks ForexQuant,

I have ever studied modern portfolio theory some time ago, but do not remember the details. If you say as far as I know the details of the modern portfolio theory, I will know what he said, then I studied it again.

Now I worry that these portfolio theories to me are about holding investments. Nevertheless, in day-trading, positions are not hold over-night.

cheers,
  • #14
  • May 18, 2010 9:08pm May 18, 2010 9:08pm
Quoting kk007
Thanks ForexQuant,

I have ever studied modern portfolio theory some time ago, but do not remember the details. If you say as far as I know the details of the modern portfolio theory, I will know what he said, then I studied it again.

Now I worry that these portfolio theories to me are about holding investments. Nevertheless, in day-trading, positions are not hold over-night.

cheers,
The same idea is applicable to day-trading. I use the same idea for my day-trading portfolio.
  • #15
  • May 18, 2010 9:10pm May 18, 2010 9:10pm
Quoting ForexQuant
The same idea is applicable to day-trading. I use the same idea for my day-trading portfolio.
Thanks, then I have to re-study the portfolio theory. cheers.
  • #16
  • May 18, 2010 9:36pm May 18, 2010 9:36pm
Quoting ForexQuant
Just study the modern portfolio theory then you know what he tried to tell you. Portfolio weighting and correlation is the important elements in MPT. However my experience told me that correlation is much much more important than how you choose your portfolio weighting.
About correlation, if a trader doesn't hold more than one trade simultaneously, does it mean correlation is not important and MPT is not applicable (I think MPT is about correlation between positin holdings)?
  • #17
  • Last Post: May 19, 2010 2:28am May 19, 2010 2:28am
Quoting kk007
Hi traders,

I would like to post a discussion about how to work out a money management model for multiple strategy trading method. This is a rather open-end question. Let me elaborate more.

Suppose a trader uses a number of different setup types for his trading and each of his setup type has a different win rate, risk reward ratio and frequency of occurence, and all these setup types show a positive expectency. How should he set up his money management model, so that he know the position size he should use for each setup type.

To simplify...
Perhaps, I should work out the position size of a trade for each setup type, so that the max drawdown (95%CI) in a trade series are the same for each of the setup type.
0 traders viewing now