Do Our Strategies Work On The E-Mini S&P 500 Futures Contract (ES)?
Still hunting for clues...
The hunt for the holy grail of automated trade strategy is an evolutionary process. Our success is tied to the development of every clue. The hope is to then use those clues to unlock some basic truths about strategy development.
Most of our research is concentrated on the E-mini NASDAQ futures contract (NQ). For an overview of backtest results on the NQ contract, click here. In this series, we’re looking at how our strategies perform when backtested against other futures contracts.
In this post, we’re asking the question:
Do Our Strategies Work On The E-Mini S&P 500 Futures Contract (ES)?
Before getting into the answer, let’s look at the research that came before it.
Predecessor Research
The following research came about after the discovery of Strategy 33. Based on preliminary results, Strategy 33 is one of our most robust strategies. One surprise that came out of this research was the performance of the E-Mini Russell 2000 Index Futures (RTY) contract. It posted a profit factor higher than anything we’ve seen from the E-Mini NASDAQ 100 Index Futures (NQ) contract.
The next step was to conduct a backtest on RTY to see if the performance for all strategies is comparable to Strategy 33 or if Strategy 33 is an anomaly.
RTY Backtest: The Results
The chart below is the RTY backtest from November of 2020 to November of 2021. It is based on trading 1 RTY contract.
The results varied. On the whole, RTY posted lower net income and higher drawdowns, but there were a few noteworthy developments: Strategies 17 and 33.
The strategy with the highest profit per trade for RTY is: Strategy 33. With only 57 trades for the entire year, it was profitable 65% of the time. It also had the highest net profit at $78K, a profit factor of 3.29 and a cumulative drawdown low that never went below $7,645.
The strategy with the highest profit factor and return on max drawdown for RTY is: Strategy 17. With only 31 trades, it was profitable 61% of the time. It also has a profit factor of 4.99. This is quite remarkable. It means that the ratio of gross profit to loss is nearly 5x. In other words, the strategy made 5 times more than it lost during the period. Perhaps the most interesting attribute of Strategy 17 is that it has the lowest max drawdown percentage at -1.24% — in dollar terms that’s -$1,450 — which gave it the highest return on max drawdown at 1,237%. It also has a cumulative max drawdown low of $1,615. What does this mean: it means the account value never dropped below $1,615 (cumulative max drawdown), and the most the account value ever dropped from its high was -$1,450 (max drawdown).
The next step was to conduct a backtest on E-Mini Gold Futures (GC) to see how our strategies might perform with a non-equity based contract.
GC Backtest: The Results
In general, our strategies performed worse when backtested on the GC contract compared to the NQ futures contract. It could be that the inverse relationship that gold has with equities is creating a kind of headwind ‘noise’. We tested this theory out on Strategy 38 and it confirmed our suspicions. To read more about that click here.
The chart below is the GC backtest from November of 2020 to November of 2021. It is based on trading 1 GC contract.
The results varied, but there were a few noteworthy developments: Strategies 10, 30, 33 and 34 are all worth looking into more, especially Strategy 10.
Strategy 10 made over 10K trades for the year and makes ~48 trades per day. As a result, it has a profit per trade of only $8, so it’s best to trade with a flat rate commission plan. Still, with 10K trades it continues to have a profit factor of 1.20. In general, the more trades, particularly if they are in the same year, the more consistent performance metrics tend to be. The most interesting attribute of this strategy is that it has a drawdown of only $4,970 or -2.85%, giving it the highest return on max drawdown in the GC backtest at 2025%. So, this strategy made $100K over the last 12 months with a cumulative drawdown of only $5K.
Next, we decided to go back to equities with a backtest using the ES contract. What did we find?
ES Backtest: The Results
On the whole, we found less volatility, lower net incomes, and lower drawdowns. In total, our strategies made $1.1M for ES compared to $3.1M for NQ over the same period. Likewise, the average profit factor is 1.41 for ES and 1.50 for NQ.
The chart below is the ES backtest from November of 2020 to November of 2021. It is based on trading 1 ES contract.
The results varied, but there were a few noteworthy developments:
You will notice a lower trade count for ES compared to NQ. The number of average trades per day is 3 for ES and 12 for NQ. And, this is after we used 18 range instead of 36 range for the data series as a way to increase trade count.
On the whole, average drawdown is less. Average drawdown for NQ is -6.85%, whereas average drawdown for ES is -5.50%. In dollar terms, that equates to an average of $19K and $11K, respectively.
If we had to choose one shining star for ES, it would be Strategy 34. It only has 175 trades for the year, but it also has a profit factor of 2.28 and a net income of $92K — the highest net income of all strategies tested with ES.
Another interesting note is the ‘lowest daily net profit/loss’. Strategy 10 has a ‘lowest daily net loss’ of -$1,887 with 2259 trades, $44K net profit and 1.21 profit factor. A low ‘lowest daily net loss’ is normal for a strategy with a low trade count, but not for a strategy with over 2,000 trades like Strategy 10. Strategy 10 was a star with the GC contract so we’re not surprised to see this performance under ES. We’re going to add Strategy 10 (ES) to our primary performance chart to track over time in our next update.
Holy Grail Comparison
Our goal is to find the holy grail of automated trade strategy. With each backtest we grow closer, but we’re still on the hunt.
It’s been a year. Where do we stand now?
When backtested with NQ, Strategies 37 and 38 met four out of seven of our criteria for the holy grail of trade strategy (marked in bold print).
Profit factor greater than 3
Annual drawdown less than 3% *
Annual return on max drawdown greater than 500% *
Maximum daily net loss of -$1,000
Avg Daily profit greater than $1,000
Less than 5,000 trades annually *
More than 253 trades annually *
When backtested with RTY, Strategy 17 met five out of seven of our criteria for the holy grail of trade strategy (marked in bold print).
Profit factor greater than 3 *
Annual drawdown less than 3% *
Annual return on max drawdown greater than 500% *
Maximum daily net loss of -$1,000 *
Avg Daily profit greater than $1,000
Less than 5,000 trades annually *
More than 253 trades annually
When backtested with GC, Strategy 10 met four out of seven of our criteria for the holy grail of trade strategy (marked in bold print).
Profit factor greater than 3
Annual drawdown less than 3% *
Annual return on max drawdown greater than 500% *
Maximum daily net loss of -$1,000
Avg Daily profit greater than $1,000
Less than 5,000 trades annually *
More than 253 trades annually *
When backtested with ES, Strategies 6 and 10 met three out of seven of our criteria for the holy grail of trade strategy (marked in bold print).
Profit factor greater than 3
Annual drawdown less than 3%
Annual return on max drawdown greater than 500% *
Maximum daily net loss of -$1,000
Avg Daily profit greater than $1,000
Less than 5,000 trades annually *
More than 253 trades annually *
So, we’ve achieved the following attributes:
Annual drawdown less than 3%
Annual return on max drawdown greater than 500%
Less than 5,000 trades annually
More than 253 trades annually
But, we always seem to be three to four criteria away from the holy grail of automated trade strategy. In particular, we’re having a hard time with the following attributes:
Profit factor greater than 3
Maximum daily net loss of -$1,000
Avg Daily profit greater than $1,000
Several strategies have a profit factor over 3, like Strategy 17 (RTY) and Strategy 33 (RTY), but both make less than 253 trades annually and have an average daily profit that is much lower than $1,000.
Strategy 3 (NQ) makes a daily average profit greater than $1,000, but it only has a profit factor of 1.12.
Strategy 17 (RTY) also has a maximum daily net loss of -$680, but again, it only makes 31 trades annually.
Clearly, there’s something we’re missing when it comes to unlocking the relationship between net profit, profit factor and number of trades. The issue is frequency. We know how to find profitability, but not on a frequent basis. We know where to hunt, but it could take weeks or months before we find something. Is there a way to leverage what we do know in an effort to increase the frequency of profitable trades?
What are we missing?
When you have eliminated the impossible, whatever remains, however improbable, must be the truth. – Sir Arthur Conan Doyle, stated by Sherlock Holmes.
We’re hunting something that isn’t intuitive. So, we must rely on our observations to steer us in the right direction, wherever that may lead. We haven’t found the holy grail, but we can leverage what we’ve found along the way. What we’re finding are clues that help to inform a logic around the physics of strategy formulation.
We started creating this list of clues in the post: How Can You Tell If An Automated Trading Strategy Will Perform Well Over Time? Now that we’ve backtested NQ, GC, RTY and ES, we can add a few more clues to the list: