1-2% is the max to risk per trade. Math behind?

Quote from nonlinear5:

http://en.wikipedia.org/wiki/Kelly_criterion
In your first use case of risk/reward ratio = 1/1 and profitable trades = 55%, the optimal bet size would be 10% of your account value.

In your second use case, you need to specify the odds on the winning bet.
More...

Thank you for your shared info.

Would you please explain how did you use the formula to get the 10% results?

Physician here not mathematician :)
 

Attachments

Quote from J.Joseph:

Here's a screenshot from a paper I wrote on the subject.

The issue is that, even with a 99.99% probability of winning, you still have that 1 in 10000 chance of losing. And when you do, you'll have to make up whatever you lost. So, even at those odds, I would never risk 100%. Or else every 10,000 trades I'd be wiped out (assuming every trade had those odds). Likewise, even with a 55% chance of winning on a trade you still have a fairly good chance of having a long string of losses. I mean when was the last time you flipped a coin and were surprised to see heads like 8 times in a row? It does happen, and the mere possibility of it happening means protection from it is mandatory for long term survival.
Aside from those two instances, the amount you lose vs. amount you need to win to get back to even is logarithmic as the attachment explains.

Finally, don't confuse risk with "amount traded." If I buy a stock for $100 per share and set a stop at $99 per share, that's only a 1% risk. I'm not risking $100 because I put $100 up to purchase the stock. Rather I'm only risking $1 because that's where I redeem the remaining value of the stock if I'm wrong (of course I'm leaving out commission, slippage, and spread for simplicity). I assume you knew this already, but just wanted to be sure.

Hope you find this helpful.
More...

Good chart.

Another way to look at it is in terms of points:

If you lose x points, you need to gain x points to get it back.
 
Quote from J.Joseph:

Here's a screenshot from a paper I wrote on the subject.

The issue is that, even with a 99.99% probability of winning, you still have that 1 in 10000 chance of losing. And when you do, you'll have to make up whatever you lost. So, even at those odds, I would never risk 100%. Or else every 10,000 trades I'd be wiped out (assuming every trade had those odds). Likewise, even with a 55% chance of winning on a trade you still have a fairly good chance of having a long string of losses. I mean when was the last time you flipped a coin and were surprised to see heads like 8 times in a row? It does happen, and the mere possibility of it happening means protection from it is mandatory for long term survival.
Aside from those two instances, the amount you lose vs. amount you need to win to get back to even is logarithmic as the attachment explains.

Finally, don't confuse risk with "amount traded." If I buy a stock for $100 per share and set a stop at $99 per share, that's only a 1% risk. I'm not risking $100 because I put $100 up to purchase the stock. Rather I'm only risking $1 because that's where I redeem the remaining value of the stock if I'm wrong (of course I'm leaving out commission, slippage, and spread for simplicity). I assume you knew this already, but just wanted to be sure.

Hope you find this helpful.
More...

Very helpful.. Thank you..

However, let us build on that to say my exact situation.

Assume you have a system that could have 100% hypothetical winning. So, on a simulator, it will never lose. However, on live situation, there are many factors could affect this outcome such as broker-related insufficiencies. For instances as you know; broker could slip you greatly, or could even NOT execute certain order for you or could execute partial size.

What do you think the optimal mathematical way to approach this?. I know these issues are hardly to be calculated but is there any math-related solution?


Thank you..

Your help is much appreciated.

McGene
 
Quote from bln:

1% risk per trade does't work.

I make 20 swing trades in one year, 40% percent is losers and the rest is winners. so I end up with 8 losers of 1% and 12 winners of 2%, that gives me a return for the full year that is less that 15%, sorry that is not good enough, even buy and hold is better, I need more risk.
More...

No, you need more reward.

Why would your average Risk/Reward be 2:1? It seems you should be looking for either less risk with same reward, or same risk with greater reward. I never even take a trade that has an RR of less than 3. Most of mine are 12 or better. Risk 0.17% gain 2% with stocks or risk 3% gain 36% with options. And when pyramiding is added to this, some trades end up well over 100% with options with an initial risk of 3%. Losses are important to manage, but I wouldn't start to think that there is a direct relationship between risk and reward on every trade.
 
Quote from intradaybill:

http://www.elitetrader.com/vb/showthread.php?s=&postid=2909887#post2909887

Look at the second link.
More...

I have read/watched that TradeWorx has never been ended in any day with loss for very long stretch of days. Taken into consideration, those guys are HFTers so they are putting thousands or orders in every second. A considerable number of these orders will be executed. So, their probability of winning is extremely high.

On this perspective, if those guys have 100% probability winning ratio, how do they approach optimizing their trading sizes. I know they have fine science for that but is there any idea to know how could a retail trader having similar situation " 100% hypothetical win ratio" could approach this based on math?

As a disclaimer, I do not have the holy grail or even close to it.
 
Quote from J.Joseph:

If I buy a stock for $100 per share and set a stop at $99 per share, that's only a 1% risk.
More...

No, it's not.

I would have thought that after 1987, 1997, 2000, and 2008 people would have the sense to realize that stops cannot be counted on for protection when TSHTF.
 
Quote from Random.Capital:

No, it's not.

I would have thought that after 1987, 1997, 2000, and 2008 people would have the sense to realize that stops cannot be counted on for protection when TSHTF.
More...

Don't be so freaking literal. I think everyone understands the possibility of gaps. Most of all, someone who uses stops.
 
Quote from mcgene4xpro:

Thank you for your shared info.

Would you please explain how did you use the formula to get the 10% results?

Physician here not mathematician :)
More...

The formula is very simple, as it uses only three terms: probability of winning, probability of losing, and the net odds. Middle school math should be enough to calculate the result:

f = (b*p - q) / b = (1*0.55 - 0.45) / 1 = 0.1 = 10%

For in-depth discussion, you may be interested in this book:
http://www.amazon.com/Fortunes-Formula-Scientific-Betting-Casinos/dp/0809045990/
 
You must log in or register to reply here.
Back
Top