Adrift in Uncharted Charting Territory

The big downward spike last Thursday has led to a lot of confusion among technical analysts. Is Thursday's data correct? Should it be removed? Should it be adjusted? etc.

Here at StockCharts.com we are getting lots of questions from our users about the down spikes that Thursday has left on our charts. Many people are asking that we remove those spikes because they are "obviously wrong."

Well... it's not that simple.

We are trying very hard to "do the right thing" here but almost everyone agrees that this is uncharted charting territory. Philosophically we want to make sure that the data behind our charts reflects "the will of the market" at each point in time. (This is why we adjust our historical data to remove "artificial" effects like splits and distributions.) So what happened on Thursday? Was it "the will of the market?" Or was it some "artificial effect?"

Currently, the answer is "We don't know."

The only people who can know for sure are the market makers at the various exchanges. Unfortunately it is taking them a long time to figure everything out. In the mean time, we have come up with three key policies that we are going to follow until the exchanges release definitive prices for Thursday:

1.) We are not going to adjust any of our Intraday data.

There is no independent third-party that we can compare our intraday data to. Thus we have nothing to guide us when making those adjustments. Because of that, we have decided to leave our intraday bars "as is" - a "forensic record" if you will of Thursday's events. In addition, the intraday data that we have is what we need to use to help us with our next policy.

2.) In cases where the daily low value is "clearly" wrong, we will adjust it upwards based on whatever intraday data we have.

We are being very conservative with our definition of "clearly wrong" here however. That doesn't mean that we will remove just any downward spike from our charts. We have noticed that many of the daily quotes we received on Thursday had the same low - 0.001. After reviewing the data, we decided to adjust the daily low for all stocks with that value. To adjust that low, we looked at all of the intraday data we have and we (uncomfortably) selected the lowest plausible value from that intraday data. Often that turned out to be the low of the next one minute bar on the chart.

This methodology is admittedly a conservative one. Even after these adjustments, there are still big down spikes on these charts. Our goal was not to eliminate those down spikes but to eliminate the most implausible parts of those spikes. Again, until the exchanges release revised "official" data, our goal is to only change what is absolutely necessary.

3.) We will adjust all of our numbers when the exchanges come out with "official" updates.

So far we haven't seen any new numbers but that could change at any moment. We are starting to see some indications that different (hopefully "better") numbers may appear by tomorrow. Whenever it happens, we will toss out all of the adjustments that we made and go with whatever the exchanges say.

Bottom Line:

We've adjusted everything that we feel comfortable adjusting ourselves.
We're not planning on adjusting intraday charts.
Things are in limbo until the exchanges release official updates.

I realize that some of our users want us to do more. This is clearly one of those "Darned if we do, darned if we don't" situations. If your charts are truly unusable because of Thursday's data, here are some things to try:

Turn off the "Log Scale" setting. Big moves down are exaggerated even more by Log Scale.
Use "Close Only" indicators instead of indicators that rely on highs and lows.
Fall back on trendline and chart pattern analysis for the time being. Your "Mark 1 Eyeball" can overlook the spike even if our indicators can't.
P&F Charts need to set the "Price Field" setting to "Close Only" instead of "High/Low" to avoid heavily skewed charts.

If / When we get updated data from the exchanges, I will post a message about that in this blog.

- Chip

Hi Chip, thanks for the update, Yes these are difficult times, but I wouldn't call it uncharted territory, maybe just a little larger than the usual variety of data errors. We've seen this before, bad ticks, data errors, reversed trades etc and when the official data is corrected, the chart is corrected. (however it normally only involves 1 or 2 stocks at a time) I agree with you, the only official data is the data you have and it should stand unless it is revised with new data from the exchanges. As much as we want to massage many of the ugly charts, they have to be based on the real trades, not just the trades that make the chart look "right ?". This would be like running an experiment to prove a hypothesis and when it doesn't work out, we just adjust the data to fit. Lets say we were to adjust the charts and "reduce" the relative weakness of the stocks that were hit hard, then to be fair, we should also "increase" the relative strength of the stocks which were not hit. Once you start down that road, where do you stop? Anyway its something that happened and I vote that the charts should report the official data version of what happened. "It is what it is", until officially revised by the exchanges, based on the official trades that are left standing.

Posted by: Gord Greer May 10, 2010 at 10:02 AM

Chip, Glad I stayed up late to see your comments...thanks for the updated thoughts. FC

Posted by: Fred Coleman May 11, 2010 at 01:47 AM

I appreciate your problems with the charts and prices. But let's face it, until that falty data is revised or eliminated, the charts and their resulting P&F price objectives are totally worthless. These prices and resulting conclusions therefrom were NOT the result of market conditions. They are the result of a technological breakdown and trading "accommodations and arrangements." And you can't have two different prices for one company at the very same time in two very different markets (NYSE vs. NASDAQ). The failure to have common and shared rules permitted the massive price differentials. This is not reality and "what is, isnt and wasn't." Trades are now being broken and ultimately there was no transaction at the extreme prices. My purchase of Apple at $190 was killed. No transaction = no price. I suspect that just about every ETF for which you supply price charts is bad. You have to eliminate those fake prices by whatever means, wild guesses are better than phony prices. If you don't find a way, then the charts are worthless. Good luck.

Posted by: Hank Riehl May 11, 2010 at 12:41 PM

What about the other direction - massive upticks? Take SCHB for instance. From a technical analysis standpoint, I don't really care and mentally throw out that $80 high, but it completely screws up the scaling of the charts, making the entire daily chart for that security basically useless.

Posted by: Jim Hsu May 12, 2010 at 05:06 AM

Errors/fake prices or not, the bottom line is that some "event" happened in the market last week. I want to plan better for the next one and am currently trying to see some commonalities in securities that did have a "panic" vs. those that didn't. Looking at the data, not all securities show the same (sometimes very) large "hammer", some are short, and others do not reflect any such behavior showing a more "normal" down or even an up day, with what I am assuming is "uncorrected" data. Some (many?) of those affected securities showed price weakness before the event. The negative divergence pointed out by Arthur Hill and Dr. Elder that appears with SPY seems more common in those stocks showing a hammer. Also, stocks that are more "liquid" seem to be more affected although there are interesting anomalies. I suspect this price action, even if ultimately reversed, is a warning/opportunity for future trades in these stocks. Resetting all these data points will loose valuable data.

Posted by: Wayne Wagner May 12, 2010 at 08:41 AM

Thanks for the post on how you will adjust data (or not adjust it) for the crazy day. Falling back on eyeballs and pattern reading etc doesnt work for me though. I am a quantitative investor - and I rely on your advanced scanning. Thankfully I use daily data rather than intraday.

Posted by: Rod Bryant May 13, 2010 at 01:17 AM

Hi Skip, We should not let our emotions or biases let us start changing the history of what happened. It is easy enough to have a mental note that for a "moment" there "may or may not" have been slightly erronious data that will be come a hardly noticable blip on the chart as time passes. I totally agree with the comment above "I suspect this price action, even if ultimately reversed, is a warning/opportunity for future trades in these stocks. Resetting all these data points will loose valuable data". Additionally we are in the middle of political interfernce with the markets that we do not want to errase from history (as some claim has be done with weather data). And, we are after all Technicians, aren't we? Give us the real thing and we can deal with it, no matter which way the markets evenually go. Thanks for you great work and StockCharts.com.

Posted by: Henry Albert Harris May 13, 2010 at 11:20 AM

Subscribe to Chip Anderson to be notified whenever a new post is added to this blog!