One of my first Options 101 articles was titled something like "Technical Analysis Is for the Birds." I'd been writing for OIN for quite some time by the time I began writing these articles, and I'd focused on technical analysis. However, the markets were going through a period in which technical analysis just didn't seem to work.

We're going through another of those periods. Whether or not you're a conspiracy theorist and believe that someone nefarious--dark pools, traders with co-location agreements with the exchanges or the government--is tinkering with the free market system, you can't help but notice that the normal indicators just aren't working as well as they sometimes do.

Or is that true? Back when I wrote that first article, my conclusion was that technical analysis indeed does work when the markets break a different way than expected. It works to show us something different than what we thought is at work and that our interpretations are wrong. It works the same way that a stop loss does when we set it just above resistance that we expect to hold or just below support that we expect to hold.

Don't we need to know when our expectations are wrong? Let's look at what was happening October 6, as the SPX was testing the upper boundaries of a rising price channel, a boundary at which it had found resistance previously and perhaps would again.

Daily SPX Price Chart:

One easy observation is that RSI is turning down. While the SPX was peaking at a higher relative high, RSI was turning down from a lower one. That's clear bearish divergence unless the RSI turns around and moves higher before prices turn lower.

Another easy observation is that after that break higher on the previous trading day, price action had created a doji day, a day that began and ended at about the same point. The doji day followed the strong rise the previous day, a rise that nevertheless did have some end-of-day selling as is shown by the small upper candle wick or shadow. Taken together, these two candles can be the first two of a three-part reversal signal sometimes known as an evening-star pattern. To be completed, it would be followed by a third candle, a long red candle that closed at least halfway down that tall green candle.

Someone with a bearish cast could certainly have looked at this evidence and said that there was a strong possibility that the SPX was due for a pullback, and perhaps a deep one that took it at least to the 38.2 percent pullback level marked with the Fibonacci bracket, and maybe even to the 50 percent level. Somewhere in that range would also mark the midpoint of the rising price channel, a level that's often tested, too.

I had reasons to hope that a pullback, stopped at one of those likely support levels, would happen. I was in an SPX iron condor. I, and all market observers, know that failed bearish or bullish reversal setups sometime result in a big continuation rally or decline. I didn't want a huge move in either direction.

We all know what happened. The next trading day did indeed see a pullback, but prices closed well above the midpoint of that tall white or green candle, failing to confirm the evening-star formation. Within days, the SPX had edged above the top of that price channel, too, consolidating for a while before taking off again.

Did technical analysis fail? Of course not. Technical analysis showed us exactly the point at which we would know that our suppositions were wrong, whether it was because the markets really were stronger than we supposed or whether there were forces, nefarious or not, at work that we just didn't understand.

I also contend that technical analysis evidence already evident that day hinted that the SPX just might not pull back. What was that evidence? First, look again at that lower high on the RSI. If one goes back to late August, we can see that the RSI had been forming a sort of head-and-shoulders formation. I've found such RSI formations provide a heads-up to possible moves before price action confirms it. There's always a price to pay for such advance information: sometimes it's wrong. However, in this case, the RSI did not confirm the head-and-shoulders formation. Instead, it broke higher again. As it did, it broke above a declining trendline from the RSI's September peak. That little downturn we're seeing at the far right-hand side of the chart could be just a pullback to retest that trendline to see if it now holds as support. RSI was showing us the possibility of failed bearish formation in the price action, too. Such failed formations are often met with big moves in the prior direction.

I confess that I didn't notice that RSI action at the time, but I had noticed something else that troubled me. Through long observation, following comments of a previous writer for the site, I often fit Fib brackets to the move but fit them in such a way that they might predict where the move might end. I do this by extending the bracket beyond the current level, seeing where gaps, consolidations, peaks and valleys would best fit a theoretical anchor point for the Fib bracket. Why? I noticed that after a move is complete, it seems as if someone had previously fit a Fibonacci bracket to the action, so that all the important Fibonacci levels fit with appropriate market action. For example, as a Fib level was approached the first time, prices might have driven above it during the day, but then pulled back, leaving only a candle shadow poking above that level. If several such candles occurred in a row, there was likely to be pullback to the lower Fib level to retest it as support before prices attempted to break higher again. If the market was a gapping market instead, those gaps tended to gap prices above important Fib levels.

These observations were in concert with comments from Tom Williams in his book, Master the Markets when he describes what big money does when it wants to force a continuation of a rally. "To encourage these old locked-in traders not to sell," he writes on page 29, "professional traders will mark-up, or gap up the market, through these potential resistance areas as quickly as possible." Williams' premise is that we must always look at the volume considerations, too. Those are not available on my SPX charts, but we can still note times when the SPX is gapped up above next resistance. I notice that when a move is complete, we can often discover that such gaps--whether up through resistance or down through support--tend to happen around what turns out to be important Fib levels. It's as if the market participants know where the market is headed or where they want it to head in advance. And maybe they do. Or maybe it's the oddness of the whole Fibonacci thing: flowers certainly don't know how to produce pedals that are Fibonacci numbers. Whatever it is, it's not a foolproof outcome, but knowing what often happens causes me to sometimes fit Fib brackets to ascertain where I think an index is going to go.

And that Fib bracket just didn't fit well enough for my comfort. First, the gaps aren't through important Fib levels. Just as important, I notice that after a move is completed, we'll find the biggest congestion band, the place where prices stall and kind of trade back and forth in a tight pattern for a while, to be near what ultimately turns out to be the 50 percent mark. Maybe all the congestion is just below it, with candle shadows poking above it, or maybe it will be just above it, with candle shadows poking below it. Maybe it will be right across it, with the 50 percent level tending to slice through the middle of most candles.

As of that October 6 snapshot seen in the first chart, most of the congestion was occurring far higher, near what was then the 23.6 percent level. Although back then, I wanted to believe this move was about over, it looked possible to me from this Fib bracket evidence that the SPX could go far higher. I didn't want to believe it because that was going to ruin my iron condor trade, but I was wary. That possibility alerted me that I needed to hedge my deltas if there was a breakout and not trust that the SPX was going to turn around.

Here's where the SPX was as of midmorning on Friday.

SPX Daily Chart as of 11:00 AM CT on Friday, November 05, 2010:

This fits better, doesn't it? We do see congestion near the 50 percent level. We also see action that gapped the SPX through what will now be the 61.8 percent level if there's a retracement. We also see some congestion near the 23.6 percent level.

There's nothing scientific about this. Perhaps I should have started the Fib bracket at the early July low rather than the August low. Either way, I have to say that I'm still not entirely sure of the fit of the Fib bracket. It's not telling me with any certainty whether the move is likely completed or not. I can see the possibility of the move being completed but the fit seems just as good and maybe better with a further move, perhaps up to the 1238-1240 range and maybe even higher. It depends on how you configure that 50 percent level around that band of congestion from the middle of September and into early October, and whether you anchor the bracket off the July low of the August one.

It's more art than science. Recently I was struck by something market psychology expert Denise Shull said on a seminar for Think-or-Swim. I'm paraphrasing, but she said that we traders like to think of the probabilities we use in our trades and in our use of technical indicators as the probabilities we find in physics. In truth they're more like the probabilities we encounter when betting on the odds of some team or another winning a football game. We can know the makeup of the team and the opposing team. We can watch their past performances on certain fields and against certain teams. We can determine what we believe will happen, but we have to make judgments along the way, too. Therefore, I use this fitting of Fib brackets the same way that I use Keltner channels or RSI patterns or CCI or MACD or moving averages. I use them to devise a likely thesis and then I use them to tell me when that thesis is wrong. Technical analysis certainly hasn't failed when it tells me that a predicted action didn't occur. It alerts me that either I don't understand what's happening or that something odd is going on the markets and I need to step back and reassess.