This is the second part in the ultimate guide to stop losses on stocks. In the first part, I compared some standard stop loss techniques using random entries.

In this second part of the article, I take the stop losses as mentioned in the previous article and apply them to three different types of trading systems, so we can see how well the stops work with three already developed trading strategies.

It is argued that by setting stop losses on shares, you can manage risk, cut losses and become a more profitable trader/investor. So let’s investigate.

## Strategy 1. SPY EMA Crossover Strategy

First of all, let’s load up 20 years of historical data for the S&P 500 SPDR ETF \$SPY between 1995 and 2015. This is a popular ETF that mimics the S&P 500 Index.

In this first test, we will buy the SPY with 100% cash whenever the 20 day EMA crosses over the 50 day EMA and we will sell whenever the 20 day EMA crosses under the 50 day EMA.

Transaction costs are set at \$0.01 with a starting capital of \$100,000 and trades are entered on the next day open.

*The objective function used here is the CAR/MDD ratio, which is the compounded annual return divided by the maximum drawdown. This is an ideal way to compare different trading strategies since it evaluates return relative to risk. The higher the CAR/MDD, the smoother the resulting equity curve and the better the trading strategy.

### Base result

This basic strategy results in a compounded annual return of 8.92% from 40 trades with a maximum drawdown of -31.68% (a CAR/MDD of 0.28). This compares to a buy and hold return (nominal) of 9.76% with a maximum drawdown of -55.19% (CAR/MDD = 0.18).

### Introducing fixed stops

I then introduced various fixed stop loss orders into the equation (from 1% to 70%), the results are shown in the chart below:

The distance of the stop (from the entry price) is plotted on the x axis and the CAR/MDD value is plotted on the y axis. As you can see from the graph, The best stop loss trigger price was 10% or more and this is because when the stop loss was 10% or higher, the stop loss was never triggered.

In effect, the best tactic with the EMA crossover strategy was to have no stop loss at all.

### Introducing trailing stops

In the next test we buy SPY with 100% cash whenever the 20 day EMA crosses over the 50 day EMA, however, this time we sell by trailing stop only.

Different trailing stop levels are tested from 1% to 100%. The results are shown below:

As you can see from the chart, very tight trailing stops inhibit strategy performance. Looser trailing stops give the market room to move up and are more successful at locking in profits.

In this example, the best performance came from a 11% trailing stop.

## Strategy 2. 200 Day Breakout (Portfolio)

The next system we will look at is a portfolio strategy using stocks from the S&P 1500 universe. (The data comes from Norgate and includes historical constituents).

When a stock from the S&P 1500 universe makes a new 200 day high, the stock is bought and added to the portfolio and when that stock makes a new 200 day low, the stock is sold.

The maximum portfolio size is set at 10 stocks and the starting capital is split equally between each position. Commissions are set at 0.5% per trade. Ranking is based on volume (strongest first) in order to choose between multiple signals.

### Base result

Running this portfolio strategy on the S&P 1500 universe gives a compounded annual return of 15.77% between 1995 and 2015 with a maximum drawdown of -53%, (a CAR/MDD of 0.30).

### Introducing fixed stops

Now we can see how adding different lengths of fixed stop losses affect the performance of the strategy.

In this instance stocks are sold either by the exit criteria or at the fixed stop loss level:

As you can see from the chart, there is a clear correlation between system performance and stop loss length.

Generally, the further away the stop is placed, the less often it is called into action and the better the performance of the strategy. More data is needed, but it appears fixed stop losses work better when greater than 25%.

### Introducing trailing stops

The next test is the same as above, however, this time the stock is sold by trailing stop only:

As you can see from the chart, wide trailing stops are preferable with tight trailing stops resulting in negative returns. There appears to be a wide sweet spot in the 30%-60% range.

### Introducing Chandelier stops

This time, instead of using a % based trailing stop we will use a dynamic trailing stop based on the Average True Range indicator (ATR), this is also known as a Chandelier stop.

By using ATR, we can adjust the trailing stop based on the volatility of the stock. So when the stock is more volatile, the stop is further away and when the stock is less volatile, the stop is tighter.

In this case, we buy when a stock from the S&P 1500 makes a new 200 day high and we sell by chandelier stop only:

As you can see, these results are fairly consistent across all of the values. Wider Chandelier stops appear to work best and very tight stops produce negative returns.

## Strategy 3. SPY Mean Reversion Strategy

For this test I loaded up SPY again to see if stops have any merit when used with a mean reversion trading system, as opposed to a trend following system.

In this example, we buy the SPY ETF whenever the RSI(2) crosses under 30 and we sell the SPY whenever RSI(2) crosses over 70. Transaction costs are set at \$0.01 with a starting capital of \$100,000 and trades are entered on the next day open.

### Base result

Running this strategy between 1/1/1995 and 1/1/2015 produces a compounded annual return of 8.45% with a maximum drawdown of -39.74%, a CAR/MDD of 0.21.

### Introducing fixed stops

Next we can see how the strategy performs with different lengths of fixed stops:

As you can see from the chart, there is once again a clear correlation between the distance of the stop and the system performance. Once again, the best strategy is shown to be the one with no fixed stop whatsoever. When the fixed stop loss is tight, system performance is severely restricted.

### Introducing trailing stops

This is a repeat of the test above. We buy the SPY ETF whenever the RSI(2) crosses under 30 but this time we sell by trailing stop only:

Once again, wide trailing stops are preferred. In this instance, a trailing stop of 23% was optimal and trailing stops above ~55% were not called into action.

### Conclusion:

We can see that there is hardly any situation where setting a fixed stop loss order is beneficial to a trading strategy and this was also seen in the previous part of this guide.

In nearly all cases, it is better to exit using the system logic, ie. a change in trend or reversion to the mean, or by a dynamic stop such as the trailing stop. Trailing stops appear to be effective because they lock in profits (not just losses), though they need to be given enough room to do so.

Overall, this is useful information when forming your own trading method as it confirms that fixed stop losses should nearly always be treated as a last resort.

There is one caveat, however. If you use leverage or high amounts of risk, you should still consider using a fixed stop-loss.

Why? Because sometimes, long tail events happen and if you are too highly leveraged, you can be completely wiped out. A good example of this is the recent move in the Swiss franc which destroyed FXCM and many forex traders.

The solution is to keep risk so small that large moves will not wipe you out. Keep the fixed stop loss so far away that it will only very rarely get hit. The stop thus acts as a worst case scenario and gives some peace of mind.

In the next installment of the ultimate guide to stop losses I intend to look at some more advanced methods that can be used to manage risk and exit trades. Thank you for reading.

### You May Also Enjoy:

Trading systems created with Amibroker using survivorship-bias free data from Norgate Premium Data.

Recommended Educational Resources:

Tags: ,

### One comment

• Anonymous

• January 22, 2017

• 10:35 am

I love this piece! Superb write up packed with essential information and no fluff. You have a great talent for explaining the various strategies using real data, a testament to your fundamental understanding of the subject matter. Further more I’ve been looking for ways to test simple indicators and strategies on historical data unaware that this level of automation was available to the average person.

In my current system I use a trend following strategy with a wide trailing stop loss calculated by ATR then trail accordingly, letting profits run of course while adding further positions at strategic points along the length of the trend as it unfolds. Your analysis drives home the point that catching a decent trend requires space for the trend to develop but most importantly risk management is key, only risking 0-2% per position.

I also prefer going long on stocks since if a stock falls from 50 to 25 then that’s a 50% move down but if a stock increases from 25 to 50 that’s a 100% return!

Its admirable that you practice what you teach. Finally a genuine figure who has invaluable experience within the professional/institutional side of trading willing to guide & help novice traders like us. A breath of fresh air.

Kind Regards.