So if Optimal f is less than optimal how much should a trader risk on a per trade basis?

Although there are many ways to trade the market, I've never heard of any successful trader who risks more than 5% of their account at a time no matter the market, timeframe or anything else. In fact of the pros I know the vast majority risk 1-2% of their account. I'm guessing there is a reason for that.

In my view here is the problem with risking anything close optimal f or kelly criterion. If you test your system for 30 trades that gives you about a 90% confidence level, 100 trades about a 95% confidence level, 1000 trades about a 97.5% confidence level. It would take well over 10000 trades tested to reach a 99% confidence level. That means that even if you could be 99% sure that your expectancy was accurate (within a margin of error) you still would bust your account at least 1 out of every 100 trades with an optimal f risk. Now that might sound like a good gamble to you until you consider this - it is virtually impossible to reach a 99% confidence level because once you test over about 150-300 trades the market makes earlier trades irrelevant.

In other words - the market always is changing. I've known folks to test systems of thousands of trades over decades only to find that the system will break (after being robust for over a decade) and the expectancy will shift (even if the system bounces back eventually). In reality I've found it impractical to achieve much more than a 95% confidence level. That means if we push "the optimal solution" we will go bust 5% of the time. Now I don't know about you but to me that is not the optimal solution. That is the gamblers folly.

So let me define for you what I believe is the optimal risk solution. To trade with as low a risk as possible (thus smoothing my equity curve to as close to a straight line as possible) and still achieve gains that exceed the industry standard (about 25% per yer with no more than a 25% drawdown). And for those of you who aren't impressed with those sort of gains consider the fact that if you can achieve those numbers you could manage other people's millions. Hence we are back to risking less than 5%. I actually prefer about 2-3% for my systems. Of course if I had a million dollar account I might risk 1% or even 0.5%.

Obviously we can certainly do better than the "industry standard" as small traders. However, most of the beginning traders I know are trading too small and want to make 1000% per year. They end up over optimizing everything about their system, taking too many trades, taking trades not in the plan etc. Their own boredom and fears and greed overwhelm them.

The best words that any trader (Phil McGrew) has ever said to me, "Good trading is boring". So be bored. Make 50%-100% per year (instead of dreaming of making 500-1000%). Take so little a risk on each trade that you absolutely don't care about the result of that single trade. Think in terms of at least 30-50 trades at a time. Evaluate your results quarterly or annually instead of weekly or monthly. In 5 years you will be the millionaire you wish to be.

LT

Although there are many ways to trade the market, I've never heard of any successful trader who risks more than 5% of their account at a time no matter the market, timeframe or anything else. In fact of the pros I know the vast majority risk 1-2% of their account. I'm guessing there is a reason for that.

In my view here is the problem with risking anything close optimal f or kelly criterion. If you test your system for 30 trades that gives you about a 90% confidence level, 100 trades about a 95% confidence level, 1000 trades about a 97.5% confidence level. It would take well over 10000 trades tested to reach a 99% confidence level. That means that even if you could be 99% sure that your expectancy was accurate (within a margin of error) you still would bust your account at least 1 out of every 100 trades with an optimal f risk. Now that might sound like a good gamble to you until you consider this - it is virtually impossible to reach a 99% confidence level because once you test over about 150-300 trades the market makes earlier trades irrelevant.

In other words - the market always is changing. I've known folks to test systems of thousands of trades over decades only to find that the system will break (after being robust for over a decade) and the expectancy will shift (even if the system bounces back eventually). In reality I've found it impractical to achieve much more than a 95% confidence level. That means if we push "the optimal solution" we will go bust 5% of the time. Now I don't know about you but to me that is not the optimal solution. That is the gamblers folly.

So let me define for you what I believe is the optimal risk solution. To trade with as low a risk as possible (thus smoothing my equity curve to as close to a straight line as possible) and still achieve gains that exceed the industry standard (about 25% per yer with no more than a 25% drawdown). And for those of you who aren't impressed with those sort of gains consider the fact that if you can achieve those numbers you could manage other people's millions. Hence we are back to risking less than 5%. I actually prefer about 2-3% for my systems. Of course if I had a million dollar account I might risk 1% or even 0.5%.

Obviously we can certainly do better than the "industry standard" as small traders. However, most of the beginning traders I know are trading too small and want to make 1000% per year. They end up over optimizing everything about their system, taking too many trades, taking trades not in the plan etc. Their own boredom and fears and greed overwhelm them.

The best words that any trader (Phil McGrew) has ever said to me, "Good trading is boring". So be bored. Make 50%-100% per year (instead of dreaming of making 500-1000%). Take so little a risk on each trade that you absolutely don't care about the result of that single trade. Think in terms of at least 30-50 trades at a time. Evaluate your results quarterly or annually instead of weekly or monthly. In 5 years you will be the millionaire you wish to be.

LT

Figure Note: The curved lines are "efficient frontiers" showing the most optimal risk-return values for different two assets portfolios against the variation in return. Each curve represents a different correlation ρ between the two assets. The minimum variation (point MV) represents an example of a minimum variance portfolio (Ross et al. 2002). The figure shows that past a certain point risk and return are exponentially positively correllated.

## 5 comments:

Hi LT

nice article

you mentioned that "I've never heard of any successful trader who risks more than 5%"

I have heard of it. It is quite commonly used in trading competitions. Most of these guys will use around 11% position sizes per trade. this is how they get the 600% style trading results and this is also why the majority of competitors blow their accounts.

Another place I have seen it used is in trading small accounts aggressively. For example, your total account might be 50 - 100,000 and you divide off 1,000 and then trade it aggressively with 10 - 30% position sizes. you blow up very frequently, but occasionally you get a good run and will make a 100,000 plus. mystic genie over at forex factory has a good thread on this.

caprica,

I guess I don't consider trading competitions professional as the timeframe used does not promote sustainable trading. Trading for these short periods tells you nothing of any statistical relevance.

I've heard of this, but I think of this merely as semantics. When you consider the positions taken in relation to the whole account then risk is not above 5%.

LT

Hi, i already added you to my blog roll,if you do the same that will be great.Ty.

prd trader

Done and thanks for the blog roll.

LT

I read about Advantages to Trading the Forex in http://forex-currencymovement.blogspot.com

Post a Comment