Using Quantifiable Edges to Your Advantage – Part 5 – How I Factor In Overbought/Oversold

This post is the next in the series “Using Quantifiable Edges to your Advantage“. The last post discussed how I combine studies to help establish a bullish or bearish bias. Today I will discuss how I factor in overbought/oversold readings when considering whether to take (or hold) a position.

I’m typically averse to maintaining long positions in strongly overbought markets or short positions in strongly oversold markets. Being long in an overbought market, or short and oversold market, can carry a high level of risk since market reversals under such conditions can be sharp. Mean reversion traders strictly abide by this philosophy as they look to take advantage of stretched conditions and then exit once conditions revert to a more normal state. For instance they might look to buy a short-term low and then exit the trade on a moved back up through a short-term moving average. This can be a solid approach, especially if you also factor in the long-term trend of the market.

My approach is a little bit different. Rather than comparing the market’s price to a mean or measuring overbought/oversold with an oscillator, I compare recent price action to recent expectations based on estimates provided by my studies. This typically allows me to get long easier if my studies are suggesting an upside bias, and allows me to get short easier if my studies are suggesting a downside bias. At the same time it protects me from entering a position in the direction of a move that is already strongly overdone.

Let me provide a brief example to better explain. Assume that over the last three days the estimates from my studies suggested that the market should be up a total of 1%. If over that period of time the market rises 2% then I consider it overbought and too risky to hold a long position, even if my estimates for the next few days are for further upside. But if instead the market has only risen 0.75% while my estimates suggested it should be up 1%, then it would not be overbought and I could continue to maintain my long position. And if my estimates were for 1% up, and the market had declined, there again a long position would be justified. The combination of an underperforming market with positive expectations or a market that has outperformed and has negative expectations is a combination that I want to hold a position.

Some recent long signals provide nice examples of instances where a classic mean reversion approach would have to be flat or short, but I was able to maintain a long position. For most of the early December my studies suggested an upside bias. I was quickly taken out of my long position, though, as the SPX became extremely overbought early in the month. On December 15th the market pulled back for the first time in over a week and a long signal triggered. The next day the SPX reversed sharply and closed at a new high. In doing so it would’ve meant an exit for any classic mean reversion strategy. But despite the new high, the SPX was still considered “underperforming” versus expectations over the last few days. For me this meant I could continue to hold, or even establish new long positions. On the 17th the SPX again closed at a new high, but again it was considered underperforming versus my recent expectations. It wasn’t until the 20th when the SPX was making its third new high in a row that my measurement suggested the short-term move up was getting too overheated and it was time to take profits. Even that exit was early as the market continued upward for two more days. My studies remained bullish and my next long signal occurred on December 27th despite the market closing up for the day and only one point shy of another new high.

The purpose of this is not to discuss my trade triggers in any detail but rather to share the idea that overbought/oversold can 1) be incorporated to help reduce risk and 2) be looked at a number of different ways. When I determine my position size I consider both the strength of my current open studies, and the degree that the market is overbought or oversold versus recent expectations. While you may not track studies and generate estimates in the same way I do, you should still consider adjusting your overbought/oversold measures based on market conditions and/or your current and recent outlook.