Introducing The Modified Stick Sandwich – a swing strategy for EUR/USD

In an ideal world, traders want to be able to pick a trend, enter a trade and ride the wave until its conclusion. The strategy outlined below aims to help investors achieve this by offering a high probability setup for entering an uptrending market.

This strategy is based on the observation that, in an uptrending market, the negative days tend to have smaller real bodies (open to close) than the positive days and that their ‘session closes’ tend to occur nearer the highs. These characteristics are noted on the chart below.

The Setup – ‘Your Bread and Butter’

This simple setup comprises of three bars or candles. The first candle is bullish and ‘green’ and needs to be followed by a bar with a smaller real body. Finally the third candle requires a longer real body than the middle candle. The name of the strategy derives from the fact that the middle candle is somewhat ‘sandwiched’ between two longer bodies – in a similar but looser manner to the sandwiching which occurs in the ‘stick sandwich’ Japanese Candlestick technique.

Candle’s 1 and 3 must be bullish and green, whilst the middle candlestick can be either green or red with the two varieties shown below. These correspond with the setups illustrated on the chart of EUR/USD above.

Setup A

Setup B

As far as actually trading the setup, the high of the third candle is used as the place to enter a buy order. By placing the entry order high, the number of bad trades is automatically reduced. Depending on the level of risk/reward desired, you can individually adapt the entry preferences. Stop loss placement is optional and some possibilities are discussed later in the article.

Statistical Perspective – More likely in an uptrend?

Before we go on to look at the results and the different options for optimisation let us first check whether there is a statistical basis for the strategy. The underpinning principle is that the setup occurs more frequently in an uptrend.

Using a simple 21-day moving average (MA) to define the trend we can test this hypothesis. Days occurring above the 21-day MA are defined as being in an uptrend whilst visa versa for those below.

Firstly, categorising the days by trend, using the EUR/USD we find the following:

Total number of days above the 21-day MA = 1426 = 55.42%
Total number of days below the 21-day MA = 1147 = 44.58%

Likewise, if we count the number of setups which occurred above and below the 21-day MA and categorise them as either happening in an uptrend or downtrend, then we have two sets of data to compare. If our hypothesis is correct there should be a preponderance of setups occurring above the 21-day MA. This experiment yielded the following results:

Number of setups which happened above 21-day MA i.e. during an uptrend = 154
Number of setups which happened below 21-day MA ie during a downtrend = 63

In percentage terms this breaks down to:

Uptrend setups = 70.97%
Downtrend setups = 29.03%

Compared to the 55% and 45% distribution for overall days, the 71% and 21% distribution indicates a bias towards the setup occurring in an uptrend, therefore supporting our hypothesis. However, to increase certainty the results could be further checked for statistical significance.

It is generally accepted that, for a result to be significant it has to occur outside of two standard deviations of the mean. Using binomial probability to assess our data we got a mean number of 120.26 for all setups and from that we got a standard deviation of 7.32.
In this case two standard deviations gives 120 + (2 x 7.3221) = 134.6442.

This is still less than the 154 times which the setup occurred in the actual sample data. It suggests that there is only a small probability that the bias we detected could be put down to chance alone.

In percentage terms, two standard deviations from the 55% mean gives 62.17%. Given that our sample shows 70.97% of setups fell within an uptrend, the result is more than two standard deviations and slightly above three standard deviations.

This indicates therefore, that from a purely statistical viewpoint we are probably onto something.

The 5-bar method

Time to test the method in a trading environment – if we trade every setup and use a 5-bar trailing stop what sort of results will we get using a $100,000 lot over a 9-year period on the EUR/USD?

Whilst the strategy above is profitable and has made an average 6.98% annual rate of return – which is a respectable amount, it is not a particularly high yielding strategy. So the question is, is there a way of optimising it?

In theory, using the 21-day moving average, we should be able to filter out the bad trades. If we keep the 5-day trailing stop the results we get are below:

We can see that the strategy, with a 21-day MA acting as a filter produces no real improvement. This may be because the setup itself tends to occur more often on the right side of the market. In addition, the 21-day MA may have ended up filtering out profitable trades, entered too early, when the MA was still above the price.

Perhaps if we threw out the 21-day MA, loosened up the stop, extending it by a day.

This appears to have improved the profitability of the strategy somewhat. The 7-day stop increases returns even more but note, the progress stops at the 8-bar stop and falls dramatically at 9.

Can the dollar trailing stop improve our return?

A set dollar trailing stop can be used as a method for trailing a stop behind a trade and limiting risk. With this form of stop the trade will close once the loss has reached a specific predefined dollar level. The dollar trailing stop works best alongside the 7 day trailing stop tested above. In this method both stops are used together, whichever one triggers first closes the trade.

Previously we tested the following stop sizes: $250, $500, $1000, $1200, $1500, $2000 and $2500. From our initial investigation, the most profitable combination was obtained using a $1500 dollar trailing stop together with the 7-day trailing stop. The results are shown below:

Conclusion

The art of trading using technical indicators involves entering a trend in its formative stages and exiting before it changes. In this experiment we set out to find an indicator which would help us achieve this. The setup we identified occurs with a high probability in an uptrending market and enables traders to access an entry point with a high degree of certainty and confidence in the future direction of the trend. Various methods for stop placements were analysed, with the result that the combination of a loose trailing stop and a tighter, fixed-risk limiting stop produced the best outcome for the period of time studied

Research by:

Joaquin Monfort

Forex4you analyst