Simple RSI trading strategy based on 40- and 60-point zones
are they important?
Note: it is highly advised to read 3 articles (part 1 part 2 part 3) about Relative Strength Index before continuing. One of the observations about Relative Strength Index that comes with experience is that reversals around 70-points and 30-points lines occur only in horizontally trading markets. In trending markets, the support/resistance zone is shifted by 10 points (relates to RSI(14) only!). So, in the uptrend, the RSI will find support in 40-point zone and resistance in 80-point zone. In the downtrend, the zones will be 20 and 60 respectively If we take this idea a little further, we can assume that 40-60 zone is “no man’s land” and only breakout from this zone can either provide a signal for a trend change or a signal for a trend continuation. Of course, both numbers (40- and 60-point lines) are not set in stone. Sometimes the reversal occurs little lower or little higher, but for RSI(14) we can assume that these zones are important. If you read previous articles about RSI, you already know that what RSI indicator is really showing (badly calibrated) is a distance between price and exponential average of corresponding length. You also know that for RSI(14) 40-point line is approximately -0.5*scaled volatility and 60-point line +0.5*scaled volatility from ExpMov(27). We will use RSI Predictor indicator to create a trailing support/resistance line based on predefined RSI levels. But first, let’s make one thing perfectly clear: a trailing stop can only move in one direction! For example, a trailing stop for an existing long position can only go up when the criteria are met or stay at the same level when predefined criteria are not met. It can never be lowered. The opposite (move down only) works for trailing stop for short positions. At any given bar we can calculate where price needs to move (next bar) for our RSI(14) to show either 40 or 60 points. If we start in our “no man’s land” we have one number calculated above current price, and another one below it. If price does not break out, we calculate it again, and again… After a few iterations, a resistance line is drawn above current price and a support line below it:
To clear out the above picture, we remove RSI Predictor band and add a single trailing stop line that will either be above or below current price. And when the price crosses it, it will mean that the previous long position got stopped out and/or the trend just changed:
We just created a simple trading strategy that can trade long, short or both positions. Upper line will serve as a trailing stop for short position or trigger for long position, and lower line will serve as a trailing stop for long position or trigger for short. Besides direction (long/short/both) the strategy will have 3 inputs:
• RSI length We can see that this particular strategy is way more efficient for longs than shorts, but sometimes (2008) it can be really good for short positions as well. We checked just one combination of 3 different inputs (14/60/40). Of course, we can run the full 3-dimensional optimization process, but to keep it simple, RSI length will be fixed at 14, and we will only run 2-dimentional optimization to check which combination works best for long position and which one works best for short positions. For testing, we will use daily S&P500 e-mini futures data for last 20 years. Test will be run in TradeStation platform. First let’s take a look at long only strategy. All tables are presented as a combination of long/short triggers. Long trigger is an RSI level above 50 that is used to calculate resistance line. Short trigger is an RSI level below 50 that is used to calculate support line. The upper table presents Net profit, middle number of trades and lowest Profit Factor (sum of all gains/sum of all losses). From the upper table we eliminated the area where profit is lowest. From the middle table we eliminated the area where we have less than 1 trade / year, because it is more of a buy-and-hold strategy, than a systematic trading system. It leaves us with a small area where we can find 2 best combinations of long trigger / short trigger. One being 61/25 and the other one 61/33. First one means that we have a buy signal when price breaks above resistance line based on 61-points RSI line and position is closed when support line based on 25-points RSI line is breached. Second one takes 33- instead of 25-points as support. Doing the same thing with short positions, we see that any combination of long/short triggers ends up generating losses. But we still can pick the best combination. In regard to short positions, the best combination is 59-61/31-33. This applies to >=20 total positions. If we allow fewer positions, the best (or the least bad) will be 61-63/17-19, which means that the short position is opened only when an extremely strong down move occurs. Switching back to the S&P500 chart, we can see the differences between different sets of indicator inputs. In long only window the differences are small, but in short only window we see that 62/18 combination means that we almost never trade and wait for another 2008 to profit from this kind of strategy. By now, you probably noticed that for resistance line the best “window” of parameters we get is between 59 and 63 – which supports our initial assumption, that a 60-point area is important. But 40-point area … not so much – at least not for S&P500 in last 20-year window. This simple strategy can be used in trending markets to both initiate a trade and to close it. The important thing to remember is to check both sides separately, because most likely the combination of inputs for long and short side will differ slightly.
|
