Why Not To Use Stop-Losses. Becoming a Better System Trader part 2.

In Blog

This is part two in my series “Becoming a better systems trader”. The first part explained the importance of compound growth and re-investing your trading profits back into your strategy and not into a handbag collection for your wife. You can view the post here: Becoming a Better System Trader Part 1


This post will detail on whether adding a hard dollar stop-loss to your strategy is beneficial to performance (hint: it usually is not)


I recently had a consulting client (contact me if interested) that disagreed with me over the use of hard dollar stop-losses. I could not convince him stop-losses were not ALWAYS optimal or even beneficial. I thought this view on stops was more omnipresent; however, after reaching out to other clients and subscribers I realized I need to make this more publically aware.


I am currently refining a new strategy concept and it has proven to be the quintessential example for this idea. That is, it tests better in almost every, if not all, metrics with OUT a dollar risked stop-loss. That is right; a hard dollar stop-loss does more harm than good and here is an example to prove such.


I tested this sytsem from April 2002 (arbitrary) to October 9, 2015 (today). I tested it on the eMini S&P 500 futures contract trading only 1 contract per signal and not re-investing trading profits back into the system (non-compounding growth).


Let us compare the strategy with out a stop loss, with a $2,000 per trade stop loss and then with an optimized dollar stop-loss. I optimized from $1,000 to $3,000 incrementing by $100 and found the “best” performance or optimal stop to use was a $3,000 stop loss. However, that performance still paled in comparision to the system with NO stop. So even after checking every stop loss amount from $1,000, $1,100, $1,200… $3,000 it still makes more sense to trade this strategy with out a hard dollar stop-loss.


Here is a table comparing the 3 different test’s results –


**Figure 1-A. Profit Factor is Gross Profits/Gross Losses. Essentially, how many dollars was made for every dollar lost**

Here are their respective equity curves and performance reports ( **from left to right** no stop, $2,000 stop and $3,000 stop)…


NoStopReport StopReport  Stop3KReport



Yeah, I know the $3,000 stop made the most money. However, it made a measely $300 more while withstanding a $6,000+ greater drawdown. As a fellow trader, I’m sure you can understand that is not the risk:reward we are looking to add to our models.


A simple and rudimentary way to normalize and compare system results for risk is to look at profit/drawdown ratio. There is a clear winner here – I will let you decide. (refer back to figure 1-A).


Moving forward, my client expressed some concerns about trading without a stop because of the higher leverage in the futures arena. I suggest trading with an “oh s**t” stop. Some amount that is not going to affect system performance, but assures you that you will not be completely wiped out given some tail risk on a one off event. Also, technical levels or dynamic measures such as using some units of volatility as your stop can prove beneficial. This post is to dispell the use of hard dollar stops, only!


I also would like to point out some systems, typically mean reversion concepts, show similar results to the system in this post. That is, better performance and risk measures withOUT a stop at all. However, some trend and momentum strategies with very low winning percentages, but large payoffs DO require a stop. This is the importance of testing and knowing your system! Each system is different, but it is impossible to know without testing your ideas. If you need help testing your system please contact me! It is a low cost service we offer that comes with a warranty!


Obviously, in this case, I did not test some sort of technical or quantative stop in this basic example. For example, something more dynamic than a hard dollar stop might prove useful; however, one cannot know until they test their system/idea. A few quick ideas are: entry price – k*ATR, entry price – k*volatilityUnits, lowest(low,n bars ago), etc.


Usually most clients prefer a dollar stop because it is psychological, but we all know the market doesn’t care what we think is a good dollar amount to risk on a given trade at a certain time. This is why dollar amount stops are determinental, test poorly and should almost always be avoided.


In summary, stop trading everything you read in trading books. If all these authors advocating stop-losses were any good they would not bother writing books. Dollar amount stop losses just feel good, but in reality they do not test well!


Keep a lookout for post 3 in this series,
David <- click my name for an About Me blog post


P.S. This algorithm is not currently finished. For a FREE trial to my finished algorithms please visit here: Active Elite Free Trial or Pro Elite Free Trial

Recommended Posts
Showing 5 comments
  • David

    Here is a note where Goldman Sachs explains how Stop-Losses contribute to the severity of market crashes!

    “Trading with Stop-Loss orders can accelerate a sell-off. Consider that retail investors tend to trade around the open and can often trade with stop-loss orders. A stop-loss order is triggered when a security trades below a certain threshold, and prioritizes the speed of execution. In other words, it sells agnostic of price. In a Flash Crash, stop loss orders can be triggered, and this second wave of automated selling can worsen the sell-off, and can trigger sales of securities sharply below the threshold if liquidity dries up. Some have argued to place limits on individual trading of ETFs if the discount or premium to the NAV is greater than, for instance, 5%. Increased investor education, regardless, seems key.”

pingbacks / trackbacks

Leave a Comment