Hypothetical scenario - how much would you risk?
- Thread starter Cutten
- Start date
I clipped this see below:
So for 1 $10,000 unit of margin I invariably invest $20 so thats a 2% bet size.
Could you do the math again?
Michael B.
So for 1 $10,000 unit of margin I invariably invest $20 so thats a 2% bet size.
Could you do the math again?
Michael B.
http://www.elitetrader.com/vb/showt...029&highlight=money+and+management#post372029
There is 2 things to distinguish:
Risk management and Money Management in the restricted sense of optimisation (position sizing,..) used by Ralph Vince for example - the author of the classical "New Money Management". Of course there is some overlap between the two.
The most important one is the Risk Management because it relates to the alpha risk type in term of probability theory which is the risk of being in error whereas optimisation relates to the beta risk type which is only the risk of missing some opportunity. Missing opportunity cannot conduct you to go broke (although it should be modulated in details but I simplify here) whereas not knowing the alpha risk can get you broke.
People focus too much on Money Management thinking that it is the "secret" of wealth - perhaps because of sellers of books/systems which have interest to put the accent on Money management rather than on risk management. Yes you will see people who become rich suddenly because they applied martingale rule (BTW so called anti-martingale rule is also a martingale rule mathematically it is amusing how marketing tried to disguise to people things through fake name !) but what you see is the "survival bias" of chance that's why the same person when ruined later is incapable to reproduce the same exploit. Money Management is only useful when Risk Management is already under state control. Then it becomes easy to optimise that is to say use Money Management (using probability with or without Monte-carlo, Linear Programming).
In conclusion: "PREMATURE optimisation is the root of all evil"
It is in fact Knuth's famous maxim in software programming but as you can see it can apply to other field since Money management is about optimisation. Doing Money Management before doing the essential part of Risk management is also the root of all evil (it will conduct to overfitting and constant tweaking for example).
There is 2 things to distinguish:
Risk management and Money Management in the restricted sense of optimisation (position sizing,..) used by Ralph Vince for example - the author of the classical "New Money Management". Of course there is some overlap between the two.
The most important one is the Risk Management because it relates to the alpha risk type in term of probability theory which is the risk of being in error whereas optimisation relates to the beta risk type which is only the risk of missing some opportunity. Missing opportunity cannot conduct you to go broke (although it should be modulated in details but I simplify here) whereas not knowing the alpha risk can get you broke.
People focus too much on Money Management thinking that it is the "secret" of wealth - perhaps because of sellers of books/systems which have interest to put the accent on Money management rather than on risk management. Yes you will see people who become rich suddenly because they applied martingale rule (BTW so called anti-martingale rule is also a martingale rule mathematically it is amusing how marketing tried to disguise to people things through fake name !) but what you see is the "survival bias" of chance that's why the same person when ruined later is incapable to reproduce the same exploit. Money Management is only useful when Risk Management is already under state control. Then it becomes easy to optimise that is to say use Money Management (using probability with or without Monte-carlo, Linear Programming).
In conclusion: "PREMATURE optimisation is the root of all evil"
It is in fact Knuth's famous maxim in software programming but as you can see it can apply to other field since Money management is about optimisation. Doing Money Management before doing the essential part of Risk management is also the root of all evil (it will conduct to overfitting and constant tweaking for example).
Ok, now let's try taking this thread into the real world. You think your odds on "can't lose" trades vary around a mean of 5:1 reward/risk, 60-70% win rate, but you can't be sure. Maybe it is 3:1 on 40% win rate, and you just had a good market environment in the last 2 years, who knows? So, given real world uncertainty, what is the optimal bet size, given that you have probability estimation or "model" risk too?
Also "ruin" must be less than 100% loss. After all, who wants a 90% or even 50% drawdown? And if you have a reasonable but not too large capital base, or run public money, then a realistic maximum acceptable drawdown might be more like 25%. In this case, do you just reduce the position size proportionate to the lower ruin number i.e. with ruin as a 25% drawdown, you would trade on 1/4 the size that you would normally choose on the basis of the Monte Carlo simulation?
I am basically trying to establish the approx best position size for trades where you have huge conviction. On gut feel I say around 3%, maybe 5% if you don't mind higher drawdowns, but I'd be interested in seeing what someone with mathematical & statistical modelling skills would come up with.
Also "ruin" must be less than 100% loss. After all, who wants a 90% or even 50% drawdown? And if you have a reasonable but not too large capital base, or run public money, then a realistic maximum acceptable drawdown might be more like 25%. In this case, do you just reduce the position size proportionate to the lower ruin number i.e. with ruin as a 25% drawdown, you would trade on 1/4 the size that you would normally choose on the basis of the Monte Carlo simulation?
I am basically trying to establish the approx best position size for trades where you have huge conviction. On gut feel I say around 3%, maybe 5% if you don't mind higher drawdowns, but I'd be interested in seeing what someone with mathematical & statistical modelling skills would come up with.
Professor Harry,
This is over my head.....would you pm a spreadsheet to help me calculate an NQ bet size? Do terms such as "winrate" or "avg win to avg loss ratio" have any bearing on proper money management. I prefer a dynamic approach rather than a fixed value approach, but perhaps you have simplified it.
I find using Kelly and dividing by 4 or 6 keeps the long stints of digging out of drawdown to a minumum.
Michael B.
This is over my head.....would you pm a spreadsheet to help me calculate an NQ bet size? Do terms such as "winrate" or "avg win to avg loss ratio" have any bearing on proper money management. I prefer a dynamic approach rather than a fixed value approach, but perhaps you have simplified it.
I find using Kelly and dividing by 4 or 6 keeps the long stints of digging out of drawdown to a minumum.
Michael B.
Why do you need this spreadsheet since you have the marvellous Kelly formula in your java applet hee hee ! There's nothing wrong with this formula, the problem is to put numbers that are CONSISTENT STATISTICALLY SPEAKING and that is another problem to prove that these numbers are consistent: that's the hard part of the work but since it is not the question of this thread I won't extend 
.
.Quote from harrytrader:
Why do you need this spreadsheet since you have the marvellous Kelly formula in your java applet hee hee ! There's nothing wrong with this formula, the problem is to put numbers that are CONSISTENT STATISTICALLY SPEAKING and that is another problem to prove that these numbers are consistent: that's the hard part of the work but since it is not the question of this thread I won't extend
So an "average" is not a valid way to prove consistancy and calculate with?
One would need to calculate with standard deviations from the average right?
Student :Michael B.
Great thread! I like to see these types of discussions on ET.
I think your original question put forth here with 60-70% success and 5:1 reward/ risk scenario is however, unrealistic.
I currently trade a system on a particular market with a trade success rate of 68% on avg., however the reward to risk rate is much lower, around 2:1. I will risk between 3-4 % of my capital on a trade. I will not risk any more than this on any trade, no matter how strongly I feel about a trade. There are too many instances were something will or may go wrong, Murphy's Law!
People who voted for 2% or 5% are in my risk camp! There's no need to be greedy with a good trading system. Limit your risk to appropriate levels, ie. 2-4%, and let the Power of Compounding take over.
I think your original question put forth here with 60-70% success and 5:1 reward/ risk scenario is however, unrealistic.
I currently trade a system on a particular market with a trade success rate of 68% on avg., however the reward to risk rate is much lower, around 2:1. I will risk between 3-4 % of my capital on a trade. I will not risk any more than this on any trade, no matter how strongly I feel about a trade. There are too many instances were something will or may go wrong, Murphy's Law!
People who voted for 2% or 5% are in my risk camp! There's no need to be greedy with a good trading system. Limit your risk to appropriate levels, ie. 2-4%, and let the Power of Compounding take over.