We all now how it works, an RSI reading under 20 and we get ready to buy, over 80 we sell.
With some patience and practice it works great... until the market picks up a strong trend and you get stopped out on every consecutive trade.
The biggest problem with trading RSI signals naked
No filters on an RSI is a disaster in the making.
In order to utilize the RSI (Relative Strength Index) we need to make it relative to something besides itself!
This is where additional filters come in to limit the amount of false signals during a strong trend.
Now the trade-off with this is that we won't get as many trade signals as before... that's bad news for your broker, good news for you!
A trade signal filter increases the quality of your trades, therefore increasing your win rate, which increases your expected return.
It's quite simple, multiple your percentage of winning trades by your average winner, do the same for your losers, then subtract the two.
A positive integer means you are expected to make money, a negative means you're expected to lose over time.
Anything you can do to increase the inputs to the formula will work in your favor to make you a profitable trader.
That's what it's all about.
As an algorithmic trader the first thing I do when I run into an opportunity like this is to program a strategy for back-testing and genetic optimization.
I'm more concerned with the numbers over a series of trades, I could care less about being right every time.
First to get an idea of what we're working with you'll want to run the RSI buy/sell signal by itself on many markets.
The following is a simple always in the market strategy. It's a lightweight piece of code and can be ran on every market for many years in a matter of seconds to get a base idea of performance:
IF the RSI is above 80 and the %K crosses below the %K, then sell at market.
IF the RSI is below 20 and the %K crosses above the %D, then buy at market.
Often times the results will be great, sometimes disastrous, it mostly depends on the time period it was running; this makes it very risky to trade live right away because what if the market does not give good RSI signals right now?
At best such a simple strategy is a gamble.
To give yourself a shot we start adding additional filters to smooth out the false signals.
The idea is that we can some filter create an adaptive strategy that is robust to avoid disaster if it strikes.
Now here's where it gets tricky, we must fine tune the system to accept changes in market behavior, but not too much that it is over-fitted to the data.
The strategy must be ADAPTIVE and ROBUST if it is to survive the bad times.
I've put together a checklist to help reduce false signals in an adaptive manner.
What to look for when trading an RSI:
>Support based on swing structure; price should have reacted to this level at least once before.
>Overextended price action based on higher time-frame analysis; does the bigger picture support the trade?
>Is there enough trading activity to ensure the trade doesn't fall short and stagnate?
>Are we entering the trade in combination with a price action signal proven to increase the probabilities of success based on an algorithmic back-test?
An RSI has its place in ranging activity, be sure to have signal filters that support the RSI where it functions the best and you'll add an edge to your trading right away.