In the January 2022 issue of Technical Analysis of Stocks & Commodities magazine, we're presented with another indicator created by John Ehlers.
Can this new trading indicator help us make more profitable trades?
Let's find out!
In this article, we're going to test this new indicator across several markets and timeframes to see if it can be used to build profitable trading systems.
The article in question is, "(Yet Another) Improved RSI", which is found in the January 2022 issue. I'm not going to get into the details of the code or the theory behind John's math. You can read the article and download the code if you have a subscription to Technical Analysis of Stocks & Commodities magazine. I've been a subscriber for over 10 years and still recommend it.
The article's main thrust is that this new RSI-based indicator (RSIH) is better than the original RSI because it's a lot smoother. Thus we should reduce whipsaws. FYI: The "H" in the indicator name stands for "Hann window," which is a mathematical concept that I certainly don't understand.
Below is an image of a price chart with the classic RSI indicator (upper pane) and the enhanced RSI with Hann windowing to become the RSIH (lower pane). The RSIH has a zero mean and is smoother than the classic RSI.
You'll notice the HRSI indicator has a horizon line at zero. This is the critical line that generates buy and sell signals. A long trade happens when the HRSI crosses above the zero line. A shoring signal occurs when it crosses below the zero line.
I updated the code provided by Technical Analysis of Stocks & Commodities magazine to test long and short independently and combined. I added an input value called LSB (long, short, both). As an input, I can use TradeStation's optimization feature to optimize the values of -1 to 1 in steps of 1.
I'm not concerned about testing this new RSI against the traditional RSI. I want to see if this new indicator will work across different markets and timeframes. If it does, I can add it to my collection of indicators for consideration when building strategies.
Testing Across Different Markets & Timeframes
I'm going to use Tradesq to test across all available futures markets and time frames. Tradesq is an online tool that will allow you to test your code, and many popular futures markets and across 8 intraday timeframes. You can read more about Tradesq in the article, "How I Trade Like A Hedge Fund.
Futures markets tested include: AD, BD, BO, BP, C, CC, CD, CL, CT, DX, EMD, ES, EU, FV, GC, GF, HE, HG, HO, JPY, KC, KW, LB, LE, MW, NG, NQ, OJ, PA, PL, RB, RR, RTY, S, SB, SF, SI, SM, TY, US, W, XG, YM.
The timeframes tested include:
I had Tradesq vary the lookback period from 5-60 in steps of 5. I also carried the LSB input from -1 to 1 in steps of 1 so I could test long and short trades independently and together.
Best Performing Markets & Timeframes
I took the results from Tradesq and loaded it into a spreadsheet to find the best performing markets. What defines the best performing markets? Good question.
When Tradesq has completed all backtests it generates a table of the results. This table lists the markets tested and all time frames. See the image below.
This RSIH Edge Table does not display all the results from the backtest. Instead it's showing you filtered results. The filter applied to the backtest required that a backtest must...
- Minimum number of trades 100
- Minimum PnL: $5000
- Minimum R2: .55
- Minimum NP vs DD: 1.5
These values may seem too low, and I would agree. Remember, we're testing an indicator, not a trading strategy. I would not expect a single indicator to create stellar results. I'm just looking for OK results knowing that this indicator would be combined with other indicators, price patterns, and filters to make a final system. In short, I'm just looking for an OK performance.
Notice also in the RSIH Edge Table there are numbers in each cell. These represent the successful run that's passed the filter requirements. Remember, we tested different lookback periods. We also tested long trades only, short trades only, and both. These are our different runs.
When a particular run passes the filter, this is called an "edge." The higher edges mean more backtests in that specific category passed the filter. So, many edges show that more backtest produced passing results. The different input values (our lookback period, for example) had a passing performance.
From these results I simply pick the top markets which have the highest "edges".
- HO (Heating Oil). This looks good on the daily chart and the 240-minute, 360-minute, 480-minute, and 600-minute intraday charts. It also appears that both long and short trades are viable so, be sure to test both.
- DB (Euro Bund). This held up on the daily chart and across several intraday timeframes including 120-minute, 180-minute, 240-minute, 360-minute 480-minute, and 600-minute. There is a clear bias to taking long trades only.
- YM (E-mini DOW). Surprisingly this looks best on a 180-minute chart. There is a long side bias, of course.
Below are equity curve of some of the better examples from the top three markets. Notice these equity curves are all based upon intraday charts with slippage and commissions deducted. In the description below each graph will find the net profit vs. drawdown metric (NPDD).
What Can This Be Used For?
Overall, I don't think this particular indicator holds much of an edge. Only a handful of promising results stood out in all the markets and timeframes.
This indicator should not be used as a stand-alone system. Like nearly all indicators, you can't just create a trading system from them. Some of the equity curves look like they almost could be trading systems. Instead, I would attempt to combine the RSIH with other indicators or filters. So, based on the Tradesq results, the RSIH could be the start of a trading system for the most promising markets and timeframes.
Another option is to use this indicator as a possible regime filter to divide the market between bullish and bearish. I'll be adding the RSIH to my library of regime filters that I will use when I build strategies.