To use these patterns you will need to install the DLL software located at: DLL Installation. You should also download one of our pattern strategies to get started. One can be found at: RSI Strategy 1

Now that the initial DLL software development is done, we will release several new patterns in the upcoming weeks.

PS: If you have any trouble getting the software installed and working correctly, please email joe@thestrategictrader.com

This is the first strategy built entirely using patterns discovered from the tstPattern Searching software. The strategy is set to exit after being in the trade for 5 bars. Obviously, 5 bars is not an ideal exit. However, it provides a uniform baseline to build with room to improve in the future. The pattern consists almost entirely of RSI based inputs.

The in-sample testing period consists of 6/1/2001 – 1/1/2011 on 5 minute ES bars.

Pattern Inputs

AddParam(RSI(3,1)[0],”RSI3 “);
AddParam(RSI(7,1)[0],”RSI7 “);
AddParam(RSI(14,1)[0],”RSI14 “);
AddParam(RSI(21,1)[0],”RSI21 “);
AddParam(RSI(30,1)[0],”RSI30 “);
AddParam(RSI(45,1)[0],”RSI45 “);
AddParam(RSI(60,1)[0],”RSI60 “);
AddParam(ADX(3)[0],”ADX3 “);
AddParam(ADX(21)[0],”ADX21 “);
AddParam(ADX(45)[0],”ADX45 “);
AddParam(RSI(ATR(5),3,1)[0],”RSIATR5 “);
AddParam(RSI(ATR(14),3,1)[0],”RSIATR14 “);
AddParam(RSI(Volume,3,1)[0],”RSIVOLUME3 “);

Observations

When we ran the pattern searching software against our testing file a total of 220 patterns exceeding statistical significance were generated. Shown here is the top pattern discovered. Keep in mind, because this is only one pattern, the number of trades is a bit low. Because of this we also validate the pattern against the Dow Mini (YM). This allows us to feel more confident that the pattern is not simply curve fit to the ES data.

Strategy In-sample Results

Profit: $26,192
PF: 2.14
Trades: 650
Win %: 62.26
Expectancy: +$41.40 per contract

Strategy Out of Sample Results (1/1/2011 – 12/1/2011)

Profit:$5,062
PF: 2.19
Trades: 97
Win %: 67.01
Expectancy: +$52.19

Strategy On YM

Profit: +$22,770
PF: 1.70
Trades: 827
Win %: 59.61
Expectancy: +$27.53

Summary

This new method for generating strategies has a lot of promise, but a lot of work left to do. We have built the main pattern reading software into a DLL so that we can distribute it with the strategy. Although we have built this blog to share with the world our techniques, there still exists IP that we do not wish to share completely. Before you can use the pattern below, you must install the pattern reading software. Follow the instructions here: DLL Installation

Download Pattern Strategy
Download Pattern File

Going forward we will be using custom pattern searching software for creating new strategies. The software searches through thousands of potential patterns and only saves the patterns that exceed statistical thresholds. The method for finding strategies is very similar to standard neural networks except with a few custom twists.

How is a pattern found?

We feed raw indicator values and the resulting output into a “training” file. The software then constructs neural networks that “learn” the profitable patterns. For example, we may wish to find a pattern that identifies a high likelihood of being profitable 5 bars into the future. For this training file we would write the delta between today’s bar and 5 bars into the future with the corresponding raw indicator values for each entry.

The software then traverses the training file and determines the best patterns for reaching the desired outcome.

Statistical Thresholds

When searching for patterns in an automated fashion, we have to be careful not to include a pattern just because of random luck. To filter out patterns that are the result of luck, we require the pattern to exceed values on win% and/or trade expectancy. At the start of each training cycle the software traverses the training file and determines the average win% and trade expectancy for various trading frequencies as well as the corresponding standard deviation.

Using this information the software identifies which thresholds need to be exceeded to be statistically significant.

As part of this discovery, I found that the same patterns I was using to filter out losing trades could instead be used as entry conditions. In a matter of minutes I had created a strategy using only newly discovered patterns that drastically outperformed the simplistic strategy outlined in part 1.

Armed with this knowledge, I plan to take this blog in an entirely new direction. Instead of focusing on if indicator does x, then do y, I will feed raw indicator values into the pattern searching software and let it figure out the underlying relationships. Then use these patterns to build new strategies.

The goal is to build a strategy that is still profitable after taking out both slippage and commissions. Since most of our articles to date have dealt with the 5 minute time frame on ES, we will continue using those parameters.

The Setup:

In order to build a strategy, we have to have a place to start. For this strategy, we will start by combining the best aspects of each of the trading methods we have tested thus far. Starting with the most recent Aroon Oscillator method.  We identified AroonOscillator(8) crossing below -80 as possessing a slight edge, so we will put a stick in the ground here.

Initial Results (Pre commission/Slippage):

Profit: +$54,987
PF: 1.20
Max DD: -$4,662
Sharp: .58
Trades: 6925
Win %: 58.51%
Average Trade: +$7.94

This is a good start, but far from profitable once commissions and slippage are applied.

Before we move on to filtering out trades or tweaking the exits we need to add more potential “entry conditions”. In this step we go back through all the indicators that showed a slight edge and see if an optimized parameter combination for it improves overall performance.

The main purpose of this phase is to get as many entries as possible while still remaining over our “statically significant” thresholds. Each individual entry criteria does not need to remain above the threshold, but as a whole when used in an if/or combination the strategy needs to at least exceed on percent profitable or pure profit.

1. Additional Aroon Oscillator Parameter Combinations
   • No additional combinations improved results significantly. AroonOscillator(27) <-99 helped, but not enough to justify inclusion
2. Cross below Bollinger Lower Band
   • Bollinger(2.5,7) was added to the strategy
3. MACD cross below signal line
   • MACD(2,12,4) was added to the strategy
4. RSI Hidden Divergence
   • RSI Hidden Divergence 29,10,4 was added to the strategy
5. RSI Hidden Divergence
   • RSI Hidden Divergence 14,22,6 was added to the strategy
6. RSI Hidden Divergence
   • RSI Hidden Divergence 18,6,2 was added to the strategy
7. Aroon Hidden Divergence
   • Aroon Hidden Divergence 12,17,4 was added to the strategy

Results:

Profit: +$81,925
PF: 1.15
Max DD: -$4,700
Sharp: .59
Trades: 13600
Win %: 57.32%
Average Trade: +$6.02

Summary:

Several key metrics decreased from the initial entry condition, but this is to be expected. Now we have a basic set of entry conditions providing over 13,000 trades. In the next several articles we will look at different ways of filtering out the non-performing trades, and implement an exit that makes more sense than a static 5 bar exit.

Eventually we will incorporate advanced concepts such as data mining and neural networks to improve the final product.

1. Buy when aroon oscillator crosses above 0
2. Buy when aroon oscillator crosses above some value x
3. Buy when aroon oscillator crosses below 0
4. Buy when aroon oscillator crosses below some value x

Strategy #1: Buy when aroon oscillator crosses above 0

Almost all parameters result in negative expectancy

Strategy #2: Buy when aroon crosses above some value x

This strategy performs fairly well, as long as x is < 0.  No parameter combination was able to exceed our target thresholds. However, a few came pretty close. The closest one would be AroonOscillator(40) crosses above -99

Strategy #3: Buy when aroon oscillator crosses below 0

Performs better than when crossing above 0, but is not enough of an edge to use in a strategy.

Strategy #4: Buy when aroon oscillator crosses below x

There were several parameter combinations that exceeded both profit and winning % after 5 bars. The best combination was when aroonoscillator(8) crossed below -80

Summary:

Using the aroon oscillator to trade “with the trend”, does not look like a viable strategy. However, buying when aroon oscillator crosses below certain values can yield a small edge. May be useful in filtering out bad trades as we start to build out a strategy.

Source Code:

Download Source

1. Buying when the price crosses below the lower band
2. Buying when the price crosses above the upper band
3. Buying when the price crosses back above the lower band
4. Buying when the price crosses back below the upper band

Test #1: Buy on cross below lower band

This method tested out well. There were a few parameter combinations that narrowly exceeded our profit and percent profitable targets set for 5 bar exits. The best parameter combination was 1.5 standard deviations and 4 for the lookback period. Here is the equity curve achieved:

Test 2: Buy on cross of upper band

This method did not yield a single profitable result.

Test 3: Buy when price crosses back above lower band

This method is similar to test #1 except we wait for confirmation of price turning around by crossing back above the lower band. Common sense would dictate this strategy should perform well. However, the actual results were closer to Test #2 than Test #1.

Test 4: Buy when price crosses back below the upper band

This strategy was unable to exceed our preset thresholds. Attached is the equity curve for the best parameter combination: 2.5 std, 24 period

Summary:

We discovered that buying on oversold bollinger band conditions can be a viable strategy. However, more work is needed to fully exploit this potential edge.

Source Code:

Download Source

In this article we will explore the viability of trading the 5 minute ES when the MACD crosses the signal line. The signal line is nothing more than a moving average of the MACD values. To start with, we will test a very common strategy (going long when the current MACD  value crosses above the signal line).

Results: Exit at 5 Bars
A common theme is starting to emerge. Buying on up moves is generally a losing strategy. Buying when MACD crosses above its signal line and then holding for 5 bars is a pretty bad idea. Not a single optimized parameter set was able to turn a profit in this scenario. Winning percentages were decent, but not above the threshold required to be meaningful.

Results: Exit at 5 Bars
Although we are still buying on an up move, the fact that we are above the zero line appears to make a difference. Based on previous testing results related to the zero line, I did not expect the strategy to do well. However, there were quite a few combinations of parameters that were profitable.  The 4,2,2 parameter combination was even able to (slightly) exceed our percent profitable threshold, but fell short of the overall profit goal.

Results: Exit at 5 Bars
Not surprisingly based on the results from Strategy #2 and Strategy #1, this strategy does not perform very well.

Results: Exit at 5 Bars
Some profitable combinations, but no where near significant.

Results: Exit at 5 Bars
Once again, we come very close to passing our thresholds on percentage and profit, but fall short.

Results: Exit at 5 Bars
As logic would dictate, because Test #5 performed well, Test#6 did not do so great.

Summary
MACD signal line crosses mostly reinforce the theme that buying after an up move is a losing strategy. However, strategy 2 had an unexpected result. If you buy on signal line crosses and the MACD is above zero, the strategy performs well. In future tests we will need to examine this behavior more closely.

Source Code:

Download Source

Related Posts:

I am hoping to get back into the swing of things, and eventually incorporate the software I have been working on as a way of further discovering what works and what does not work. The software is great at finding an “edge containing pattern” if one exists in the provided data set. However, the most challenging aspect is figuring out what data to provide and the basic constructs of the pattern you are interesting in.

That is why its important to continue down the path that I started within this journal. To most effectively find patterns within the market, you have to have a way of weeding out some of the useless indicators and ideals. Without performing this degree of pre-filtering, the software can not run fast enough to sift through all of the data.

As this software becomes more polished, I will release a version here in this journal as well as utilize it in further weeding out the garbage indicators. For example, by focusing on a single indicator and feeding it into the software, you can “fairly” quickly see if any edge is possible. If the software is not able to extract any repeatable patterns (resulting in an edge), then that indicator is most likely useless in your trading. Unfortunately, the software is not quite polished enough for prime time. In the mean time, I will continue to test indicators the way I previously approached the issue while slowly migrating towards utilizing this data mining software. Eventually replacing most of my analysis of indicators with results found through the data mining software approach.

MACD is commonly used to enter trades based on its position relative to the Zero Line. Our first test will enter long whenever MACD is above the zero line.

Results: Exit at 5 Bars
These results turned out very similar to what we saw with RSI when above 50. Very few combinations were even positive with no results coming close to demonstrating a tradeable edge.

Results: Exit Using 1:2 Risk/Reward Ratio
Not a single combination resulted in a net profit. Clearly it is not wise to buy when MACD is above the Zero Line.

Now we will test the MACD below the Zero Line. Based on the above results, we expect being below the Zero Line to outperform.

Results: Exit at 5 Bars
No combination was able to surpass our targets for total profit. However, a large number of combinations surpassed our percent profitable threshold. This indicates a potential edge, however the edge is likely too small to surpass trading costs such as commission. May be useful as a filter to filter out losing trades.

Best Combination: Fast = 2, Slow =4 Smooth =4

Results: Exit Using 1:2 Risk/Reward Ratio
No combination came close to our thresholds. This combined with the failure to reach profit thresholds using 5 bar exit, further underscores this strategy as containing a very weak if any edge.

Summary
There may be a slight edge when entering trades while below the Zero Line. However, by itself it is not tradeable. In future articles we will look at ways of using this filter to improve our overall edge.

Source Code:

Download Source

Related Posts: