Friday, May 30, 2008

Risk/Reward & Percent Risk Position Sizing

Things are not always what they seem.

I am away this weekend and will not have time for a regular study but I wanted to cover something that will be an important concept moving forward. For those already familiar with these concepts, I apologize in advance for boring you but we will be adding a little twist at the end.

To-date on this blog, we have not really introduced money-management, position sizing, and compounding to any of our tests. The reason for this is to keep the raw test results as pure as possible without the introduction of additional variables and complexities. As we began to introduce these elements to our tests, I will primarily use percent risk position sizing on this blog and wanted to briefly describe how it works and point out a few key structural components with a very straight-forward example.

Let’s assume the following:

Trading Account: $100,000
Risk per trade: 1%
Protective Stop: 8%
Profit Target: 20%
Share Entry Price: $50

By choosing to put 1% of our equity at risk per trade, we are risking up to $1000 per trade: $100,000 * 1% = $1000.

Our system calls for an 8% protective stop and we know we have an entry price of $50 so that means we are risking $4 per share: $50 * 8% = $4.

This means our protective stop gets placed at $46: $50-$4=$46

Now to calculate how many shares to buy, we simply divide our risk amount per trade by our risk amount per share: $1000/$4 = 250 shares.

Easy enough. Stockbee has an easy-to-use calculator for this on the website.

Now here is where things get interesting. We know that if our protective stop gets hit, we lose $1000 but if we hit our profit target, our gain will be $2500: $50 * 20% * 250 shares = $2500. Obvious enough, but what happens if we use a 4% stop instead? Run the math. We now are buying 500 shares at the same risk level. This means if our protective stop gets hit, we still only lose $1000 but if our profit target hits, we now make $5000 instead of $2500. The risk/reward ratio is 2x as much.

But wait, there is always a catch. We also know that the tighter the stops, the lower our win rate. The question you must ask his how much lower is it? This is really what expectancy is all about. If we refer back to one of our earlier studies and look at the raw data here, we see that in this example, our 8%-20% stop combination had a win rate of 41.94% compared to only 26.04% for the 4%-20% combination.

If we calculate expectancy for these two scenario (Chris Perruna will show you how) using the same number we did at the beginning, we get an expectancy of $467.90 for the 8%/20% combo and an expectancy of $562.40 for the 4%/20% combination – 20% more! The reason for this is simply because the tighter stops allow us to put on larger positions at the same risk level and this only becomes evident after the implementation of percent risk position sizing.

If we take this one step further and make – say 200 trades with each pair of stops in a year, the 8%/20% combo would generate $93,580 in profit while the 4%/20% combo would make $112,480 in profit and still does not include compounding. Furthermore the tighter stop will loosen up dead capital much quicker and will allow us to make more frequent trades with less margin. The only thing standing in our way now is the psychology of a 26% win rate.

Additional info:
Trader Mike


Dave M. said...

Trader Mike also did a nice report on this back in 2005.


bhh said...

thanks for the link, I am going to post this in the main entry.

Anonymous said...

And that psychology will be tough to overcome for most traders!

I use a similar method for position sizing trades in my trend following system. First, I use a multiple of ATR to establish the size of the stop loss, then I finish the calculation to get position size.

Pradeep Bonde said...

To have lower stop a non breakout entry is better is another way to look at it.

bhh said...

Hi Pradeep - I think the swing low entry you discussed has an enormous amount of potential in light of this. The raw performance stats aren't quite as good as the breakouts but the added position sizing advantage might more than make up for it.

I will be testing some of these in the future and would love to hear any other ideas you had for playing the retracements discussed in the link above.


Pradeep Bonde said...

Pullback/retracement/weakness with a timed exit will show you good results in testing.
So it will be condition like the pullback entry I discussed or variation with holding period of 5 days or earlier if it goes up more than 10% before five days.
There are more longer term retracement kind of methods also possible.

Woodshedder said...

I really want to test results of the pullback after breakout entry for a breakout/trend following system.

While I'm sure short term it tests well, I'd like a longer term strategy to complement other short term strategies.

I would love to be able to put on larger position sizes for a longer-term trend following/breakout system.

bhh said...

I have little doubt that a high volume breakout over solid earnings/fundamentals with with increasing avg. volume as a sign of institutional accumulation is one of the best prospects for long term growth. I think the key is tracking these breakouts for a period of time with a lookback period of 1-2 weeks as a setup criteria and then accumulating after a light low-volume correction or base. It will require a lot of testing though as there are so many variables at work... breakout criteria, lookback period for breakout, specific pullback entry criteria, etc.

bhh said...

We are basically looking for the 40%+ reliabilty of a good breakout with the position size of a 25% reliable entry.

Woodshedder said...

Bhh- that is exactly what I am looking for.