Can spam be used as a leading stock market indicator?
I was going over some of my stock charts last night when I thought to myself "Could the amount of spam hitting my inbox be used as an indicator of which way the market is going to move?"
In stock speculation, there are two types of indicators - confirmation indicators and reversal indicators. Confirmation indicators, as the name suggests, "confirm" the trend. If we get a break-out on high volume, then volume is confirming the price breakout to the upside (and downside, usually). It tells us that everything is in sync.
By contrast, reversal indicators (or contrarian indicators) are more useful (in my opinion). When the indicator moves one way it often signals that the price trend will move in the other direction sometime in the near future. An example of this is the put/call ratio (the ratio of all puts traded divided by calls traded). When the put/call ratio is high, that means that investors are negative on the future outlook on the market. Unfortunately for them, this usually means that the market is going to reverse direction and move higher. The reverse is true when the put/call ratio is low.
To test my theory, I took our total mail seen across our network from Feb 2, 2006 until March 30, 2007. I eliminated weekends and trading holidays to normalize our network traffic to trading days. This means that total mail will have some gaps (and spam runs occur in sequence). I don't have data on stock spam specifically, so I also had to assume that an increase in stock spam corresponds perfectly to an increase in overall network traffic.
I then calculated the correlation between the SP-500, Nasdaq, SP-600 and Russell 2000 (small cap indexes), Russell 1000 and Dow Jones Industrial average. For the entire time period, here are the results:
Russell 2k 0.665
Russell 1k 0.858
This means that for the SP-500, Nasdaq, Russell 1000 and Dow, price movements in those indexes correlate nearly perfectly with the amount of stock spam. Ironically, the small caps and Nasdaqs do not correlate quite as well: ironic because small cap stocks are the ones that spammers pump and dump.
Does this mean that stock spam is a good indicator of the market? I was suspicious. A corrleation of 0.8 is incredibly good. Too good, in my opinion. I decided to calculate the % daily price change in the index and compare it to the % change in total mail. The results are below:
Russell 2k -0.141
Russell 1k -0.141
Interesting. Using the % change in movements, stock spam is a contrarian indicator. In other words, when the market is good, spammers start spamming and naive investors get in thinking that the trend will never end. To me, that would fit in well with my own theories on pumping-and-dumping.
I still wasn't totally convinced about the market and stock spam's relationship. Between July and December, the amount of spam we've seen has gone straight up. Not so coincidentally, the market rallied for 6 months starting in January only recently started correcting. What about the period in-between (from Feb - July)?
I checked the correlation between the above indexes' price movements and the amount of spam we saw between Feb and July 2006. The results are below:
Russell 2k -0.760
Russell 1k -0.756
Results for the % change in movement were statistically insignificant. To summarize these findings, the amount of stock spam acted as a very strong contrarian indicator (a little too strong for my liking). The % change in movement did not have any predictive value. So, like a lot of trading systems on the market, the usefulness of stock as an indicator depends on the historical trading period which is selected.
All in all, what did I learn from my research? I can't really conclude anything; I need more data in order to make a definitive statement to see how well spam trends hold up during market corrections.