This series will compare the last 2 years of RSI2 performance trading SPY against the previous 10 years to see if the indicator is losing its edge.
RSI2 is a wildly popular indicator. As with any indicator, if everyone is using it, the edge may be eroded or erased. I’m curious to see whether or not we can determine if the RSI2 edge is eroding, changing, or about the same. This post will look at recent performance – from 1.1.2010 to 11.11.2011.
Most traders understand that a low RSI2 reading is a good indicator for a bounce. However, decades ago, a high RSI2 was a good indicator for continued gains, and a low RSI2 was not very successful in predicting a bounce. Instead of predicting mean reversion, decades ago, RSI2 was better as an indicator of trend strength. I have often wondered whether we will see RSI2 slowly shift back to being an indicator of trend strength rather than mean reversion.
Using SPY, I have tested various levels of RSI2, where buys take place once RSI2 crosses below or above a threshold. We want to look at what happens with both low and high RSI2 thresholds.
The tests did not include any commissions or slippage. All buys and sells were made at the close of the same day that RSI2 crossed the threshold.
Results After Buying When RSI2 is Beneath a Low Threshold
Results After Buying When RSI2 is Above a High Threshold
The graphs show that RSI2, when using it with a low threshold, is still a decent short-term indicator of coming mean-reversion. However, after the bounce, performance falls off rather significantly.
The high threshold tests suggest that a high RSI2 reading has tended to lead to more gains or consolidation, and does not work nearly as well to signal mean-reversion as low thresholds do.
The next post will show the same tests applied from 1.1.2000 to 1.1.2010. Then we can draw some conclusions about whether or not RSI2 is losing its edge, or if it is changing in some way.
The timing on when to buy the RSI2 has changed. Most RSI2 oversold stocks are also in strong down trends. So getting in to early is like catching a falling knife.
Thanks Wood. What were you using for the triggers?
RSI2 at the close being beneath or above the thresholds of 10, 8, 6, 4, 2, or 90, 92, 94, 96, 98.
Let me know that doesn’t make sense. I’ll reword it.
Perfectly. Thanks.
Good deal. I haven’t yet run the rest of the tests so I have no idea how they’ll turn out. I’m pretty curious to see what develops.
LOL, I guess I didn’t see the legend on the side of the chart. Maybe it’s time for an eye checkup.
Does weekly RSI(2) perform better?
Scav, I don’t know. I would have to test it. What setting would you use?
Buy below 10. Sell above 90.
BTW, I read sometime in the past that a trading firm backtested RSI(2) with daily prices and found it best to buy below 5 and sell above 70.
Nice work Wood.
You should use something like the 20ma or 50 as a long short bias..
only take a long trade if above the 20 and short below… you need as stated above not to take a long when in a downtrend.
Multiple time frames would be nice also.. weekly daily and hourly all bleow 20 for a long for example.
I use the RSI for oversold and bought conditions.
What about intraday? I wonder what that looks like? Have you tried it?