What % equity to risk on high expectation trades?

[Arthur W. Cutten]

What % of your net worth (not account equity) would you risk if presented with the following hypothetical trade opportunity:

Reward to risk ratio: 5 to 1
Probability of profit: 75%

Assume the market will not gap beyond your stop, and that you either take the full profit, or get stopped out for the full loss.

Also, in your answer could you please say what is the maximum drawdown in your net worth that you would be prepared to tolerate from trading losses.

[Ted Annemann]

I would risk 70.00 percent of my net worth on each play of this game.

I expect this would result in

Code: Select all

50% probability of a drawdown of 98.91% or worse
25% probability of a drawdown of 99.19% or worse
10% probability of a drawdown of 99.76% or worse
 5 % probability of a drawdown of 99.90% or worse
 1 % probability of a drawdown of 99.95% or worse

I further expect this choice would give

Code: Select all

50% probability of a string of 3 or more losses in a row
25% probability of a string of 4 or more losses in a row
10% probability of a string of 5 or more losses in a row
 1 % probability of a string of 6 or more losses in a row

teda

[leonardo]

I would risk at least 20% of my net worth per trade. And 25% per trade of my trading account.

I always want to bet on the low side of optimum.

Leonardo--

[edward kim]

how many trades do we get, or do we trade in perpetuity?

[Arthur W. Cutten]

how many trades do we get, or do we trade in perpetuity?

Trade in perpetuity. Assume the opportunities are infrequent e.g. 3 or 4 such trades per year.

[edward kim]

if i am going on my own, i can handle a 40% drawdown, so i would take about 3.5% per transaction.

if i am managing client money, a lot of investors who want really high returns start getting nervous in the 30% range. after several thousand runs, there is a tiny % chance that we will experience a 10 or 11R drawdown (with your expectancy), so I guess 2% risk will be the max for this situation, with a 25% drawdown at the most.

there is a huge difference between the maximum profit extracted, what a group of people can handle, and what you can handle.

[Arthur W. Cutten]

if i am going on my own, i can handle a 40% drawdown, so i would take about 3.5% per transaction.

if i am managing client money, a lot of investors who want really high returns start getting nervous in the 30% range. after several thousand runs, there is a tiny % chance that we will experience a 10 or 11R drawdown (with your expectancy), so I guess 2% risk will be the max for this situation, with a 25% drawdown at the most.

there is a huge difference between the maximum profit extracted, what a group of people can handle, and what you can handle.

Interesting. Would dynamically adjusting the bet size during a drawdown help to alleviate this problem? E.g. could you start betting 5%, then cut down to 4%, 3%, 2%, 1% if you got consecutive losers? This would allow you to trade larger most of the time, whilst maintaining an equally small risk of an unacceptably large drawdown.

Also, do you know a good way to calculate the chance of a drawdown of a given size, given the probability and reward/risk figures, if you use this dynamic position sizing method?

[blueberrycake]

Interesting. Would dynamically adjusting the bet size during a drawdown help to alleviate this problem? E.g. could you start betting 5%, then cut down to 4%, 3%, 2%, 1% if you got consecutive losers? This would allow you to trade larger most of the time, whilst maintaining an equally small risk of an unacceptably large drawdown.

Also, do you know a good way to calculate the chance of a drawdown of a given size, given the probability and reward/risk figures, if you use this dynamic position sizing method?

I completely agree with you. Though it obviously makes more sense to reduce your bet size as a function of your account drawdown, rather than consecutive losers (ie bet 5% w/ account at new equity highs, and ratchet down your bet size as your drawdown percentage increases).

Best way I know to experiment with this is to write a short monte carlo simulator that takes a distribution, a bet strategy, a number of games to play and then plays them a few million times and gives you a distribution of final equity & drawdown outcomes(or any other numbers you care about). If you have any experience programming, it shouldnt take more than 15 minutes to write.

-bbc