The basis of comparing different position sizing methods?

[joesan]

Hi, I am testing different position sizing methods and come up with a general question in this category: how should we compare different position sizing methods?

As we know, if you decrease the delta in a fixed ratio algorithem, or increase the risk percentage in a fixed fractional algorithem ,or increase the risks in other models, you can easily boost up the system performance.( given the trading strategy do have a positive expectancy), so system performance is virtually a function of the risk you would like to stomach here. Then in order to compare different sizing models, is it appropriate to make the risk imput same? If so, does it make sense to only compare the system performance of different models with same (or similar, to make the calculation convenient) maximum drawdowns?

 

[kut2k2]

Quote from joesan:

Hi, I am testing different position sizing methods and come up with a general question in this category: how should we compare different position sizing methods?

As we know, if you decrease the delta in a fixed ratio algorithem, or increase the risk percentage in a fixed fractional algorithem ,or increase the risks in other models, you can easily boost up the system performance.( given the trading strategy do have a positive expectancy), so system performance is virtually a function of the risk you would like to stomach here. Then in order to compare different sizing models, is it appropriate to make the risk imput same? If so, does it make sense to only compare the system performance of different models with same (or similar, to make the calculation convenient) maximum drawdowns?

"As we know" ??

Rather than just assuming that we all share the same knowledge set, why not define what you mean by "fixed ratio", "fixed fractional", etc. ? That way, everybody is on the same page and you'll have a better chance of getting a coherent answer. Remember, this is the forum where members can't even agree on what is meant by "trend".

 

[alex.samant]

yeah. no need to start the thread like superman here ...

currently i have a sizing dilemma also.

My system usually yields 50% winners.
I usually go for the 1:1.5 ratio. i settle for nothing lower than that.

I size my positions in two ways, and i am still testing out the best one:

------------------------------------
The first:

I evaluate the risk on every trade and i size each of my individual positions in order to risk a constant 2% of account.

This means that for each 2% risked I can get a 3% return.
Because the system is 50/50 W:L, it means that out of about 10 trades / week (intraday trader) on average i get:

5 x 3% - 5 x 2% = 5% / week

PRO: it quickly boosts your account if you have at least a 50/50 win/loss ratio.

CONS: it can really output a bad performance if you get below 50/50. And on bad weeks, you go home bruised.


---------------------------------
The second:

At the beginning of the month i evaluate the following:
a) the average risk / trade
b) how much does 2% of the account represents at the start of the month

According to that i set a fixed contract size that will be used on all trades for the following month.

The system still gets 50/50 and i still go for a 1:1.5 risk/reward.

PROS: When you have bad weeks, the loss is not that great as you can have trades that had a low risk.

CONS: You can't capitalize enough on trades that have a low risk as you keep your sizing constant.


So these are my tools. Anyone care to expose his ideas?

 

[ronblack]

Quote from alex.samant:

So these are my tools. Anyone care to expose his ideas?

All sounds good until an unexpected streak of losers comes along with your name on it.

Ron

 

[ronblack]

Quote from joesan:

Hi, I am testing different position sizing methods and come up with a general question in this category: how should we compare different position sizing methods?

Open as many accounts as your different position sizing methods and at the end of the trading year compare the results. Make sure you trade the same signals in all different accounts and you just vary the position size according to your methods.

Look, there is no other way to test trading ideas other than actually risking money.

Quote from joesan:

As we know, if you decrease the delta in a fixed ratio algorithem, or increase the risk percentage in a fixed fractional algorithem ,or increase the risks in other models, you can easily boost up the system performance.( given the trading strategy do have a positive expectancy), so system performance is virtually a function of the risk you would like to stomach here. Then in order to compare different sizing models, is it appropriate to make the risk imput same? If so, does it make sense to only compare the system performance of different models with same (or similar, to make the calculation convenient) maximum drawdowns?

Please define:

1. Delta in a fixed ratio
2. Algotithem
3. Risk percentage
4. Fixed fractional algorithem
5. Positive expectancy
6. Sizing models
7. Maximum drawdowns

I tell you most people here do not care about these buzzwords.

Ron

 

[joesan]

Hi, I have mentioned in my original post that this is a "general question", hint that without go to specifics, I hope to talk about the basis for comparing different position sizing methods. So it is not important whether it is fixed ratio or fixed fractional or fixed amount per contract or any other methods, the basis of the comparision should be the same .

Maybe it is not appropriate to assume that we all know something , but assuming "most of us know nothing" will not be helpful either. fixed fractional and fixed ratio method are quite common position sizing skills discussed in many books and I believed they have been frequently discussed in ET here. This week I justed suggested to the administrators of ET to create a subforum dedicated to money management topics and fortunately they has accepted it.

http://www.elitetrader.com/vb/showthread.php?s=&threadid=99171

Hope we can talk about money management skills just as frequently as we talk about trading strategies.

Quote from kut2k2:

"As we know" ??

Rather than just assuming that we all share the same knowledge set, why not define what you mean by "fixed ratio", "fixed fractional", etc. ? That way, everybody is on the same page and you'll have a better chance of getting a coherent answer. Remember, this is the forum where members can't even agree on what is meant by "trend".

 

[joesan]

Ron

Do you trade every trading idea you have in mind with REAL money ? trading strategies need thourough test before go alive, so do the position sizing models.

Quote from ronblack:

Open as many accounts as your different position sizing methods and at the end of the trading year compare the results. Make sure you trade the same signals in all different accounts and you just vary the position size according to your methods.

Look, there is no other way to test trading ideas other than actually risking money.

A book by Ryan Jones " Playing numbers to become millionaire"( I do not remember the exact name, as it is now 6:12 am here and I am styill lying in bed ) can answer all these questions. It is a book of money management, some people think it is a classis, some others think it is not very impressive. You may not agree with all his opinions, but he gave perfect definition to all these buzzwords.

you can also find the difinition of most words with google.

Quote from ronblack:


Please define:

1. Delta in a fixed ratio
2. Algotithem
3. Risk percentage
4. Fixed fractional algorithem
5. Positive expectancy
6. Sizing models
7. Maximum drawdowns

I tell you most people here do not care about these buzzwords.

Ron

 

[alex.samant]

Mr Ronblack know it all.... Why don't you expose your sizing methods?

You just dismissed what i have been using for years. And by the way, i exposed two strategies, of which one is "losses string" safe.... ( I just stated that my system produces 50% on average and trust me i have about 7000 trades in my diary and the ratio is pretty constant. I also said that there is no more than 2% risk per trade. What do you take me for? Some kind of an idiot? Who are you after all?

But, i guess that's how the story goes... Some of us like to talk the talk while some of us like to actually walk the walk ....

 

[ronblack]

Quote from alex.samant:

Mr Ronblack know it all.... Why don't you expose your sizing methods?

Look, I don't know what you mean by that but my position sizing method is very simple. I don't risk more than 2% of my available equity at any given time and that is cumulative risk that takes into account all open positions. At this point in time I have 6 long and 2 short open positions in stocks. I have purchased just enough shares for each position so in case all of them turn losers my cumulative loss will account to 2% of my equity. What part of this you don't understand?

Since you are actually the one who knows everything can you tell us the difference between fixed fractional and fixed ratio position sizing? I want to learn.

Ron

 

[sim03]

Quote from ronblack:

Look, I don't know what you mean by that but my position sizing method is very simple. I don't risk more than 2% of my available equity at any given time and that is cumulative risk that takes into account all open positions. At this point in time I have 6 long and 2 short open positions in stocks. I have purchased just enough shares for each position so in case all of them turn losers my cumulative loss will account to 2% of my equity. What part of this you don't understand?

So, when you opened the first of those 8 positions, did you already know in advance that there'd be exactly 7 more to follow eventually, therefore risking 0.25% on that first position? Or do you always open 8 positions at 0.25% risk each at the same time, then don't open any new ones unless one or more of the existing ones are closed?