This transcript is intended only to assist automated search tools.
This transcript is not a replacement for the auto-visual presentation.
If you are reading this, you have not installed Flash on your computer correctly.
What is a bad print, and why do you care? Have you seen a chart like this? It happens a lot more than you think. Someone could have typed in the wrong value when making a trade or reporting to the tape via act. Someone could have reported a trade long after it happened. Many data providers will try to correct this, so you may only see this on a chart for a short time. The whole point of the Trade-Ideas alerts server is to point out interesting events. If we are not careful, we will point out more bad prints than anything else. Lets look a real example. On this day some users reported seeing unexpected alerts for Research in Motion. Our server reported some opening range breakouts later in the day. Why did this happen? They dont appear on this chart? Let’s look at a chart form a different data provider. In this example, the chart and price action is dramatically different. In the last example, the price was much higher than anything we had seen in over 3 hours. And the volume for that one print was over the volume that this stock usually trades in an entire day. Trade-Ideas looks for unusual trading patterns and this is about as unusual as they get. We do not have a specific alert for this exact pattern, but we reported several alerts. The volume and the change in price were both too big to ignore. So, if the stock traded like that, why were our users confused? The problem is, these two charts represent the same symbol at the same time of day. The short answer: bad market data. Some data providers drew the chart differently. Eventually Realtick fixed its data too. Depending which data provider you had, and how quickly you brought up the chart, you might or might not have seen the chart pattern with the erroneous print. Here is an example of a bad print that was corrected within 30 minutes. Let us examine another illustration. Look at the chart for KO. I’m using Yahoo! data to compare to our exchange data. This is a good sanity check, to make sure our software didn’t read something wrong. Look at the big spikes on this chart. The tallest one caused the alert in question. So, what does this mean? Was this caused by an error in someone else’s software? Or was this a legitimate print from some overly optimistic investor? Let’s get a second opinion from somewhere else. This data also comes from Yahoo but it contradicts the chart below. According to the Yahoo table, the high for KO for the 18th was 41 dollars. According to Yahoos chart, it was much higher. So, was it a bad print? Probably, but we may never know. An interesting note: Look at the low on the chart for the 20th. Look at the historical Trade-Ideas alerts for that time. On the 18th we reported that KO was up 7 standard deviations. On the 20th we reported that KO was down 7 standard deviations. Those are both big spikes on the chart. Those are both just as unusual according to our statistical analysis. However, the trade on the 18th was corrected on the table from Yahoo! and the trade on the 20th was not. What does that mean? Maybe the second one was real. Maybe the person who corrects the data just hasn’t gotten to Friday’s data yet, but has corrected the data from two days earlier. Again, we may never know. What can we do about this? Trade-Ideas provides a window specific filter that minimizes the affects of an erroneous trade. These filters are called Min and Max distance from the inside market. This compares the last print for this stock to the best bid and offer. If you set the max to 0, you will only see stocks which were trading at or between the bid or offer at the time of the alert. If you set the max to 0.1 you will only see stocks which were trading no more than one tenth of one percent above the offer and no more than one tenth of one percent below the bid. Bad market data does not just consist of bad prices. Erroneous or delayed volume data is also prevalent. Because many of our alerts use Volume as a confirmation prior to triggering the alert, this bad data can also result in a faulty alert. The exchange sends volume continuously so we dont have to compute it. If an exchange found a bad print and removed it, they would likely fix the total volume that everyone sees. This concludes this instructional video on how bad prints.