Receive insight on investment-focused marketing trends, statistics, and more right in your inbox every week.


Harvest Data Roundup - Vol. 1

Harvest Data Roundup - Vol. 1

(With help from Austin Teague, Marketing & Data Insights Analyst, and Bob Dryzgula, Chief Marketing Officer)

The world's top investment organizations share their expertise on Harvest. Proprietary technology behind the Harvest platform collects granular data on what these investors are reading and writing. Using machine learning and natural language processing, we can translate content production and consumption into powerful insights. In this series we would like to share with you trends we find from our community.

Give Us The Stars

For this first post we are taking a step back to look at the the difference in focus on growth strategies vs. value strategies on Harvest. Different market climates can make a Growth strategy more viable and produce higher short term returns, whereas value stocks with dividends can be a better bet in the long run1Although we do not speculate on strategy, we would like to share with you some interesting data from our platform.

The interest of growth and value was revealed to be relatively constant for those producing content and those reading content. By breaking up the past year into quarters we followed the trends of each investing season. In Q1 2016 about 40% of posts contained value stock mentions, but only around 20% of opens were on posts that had a value mention (see graphs below). As the year went on, the mentions in the posts approached the opens. Posters gradually picked up on what most investors were looking for: growth stocks. As it has been traditionally, most are looking to find the next “star”, the high growth companies that can quickly gain market cap and turn significant profit for investors2. The wealth of thought leadership and original content on Harvest makes it the perfect place to find the next rising star.

Screen Shot 2017-04-20 at 11.33.17 AM.png
Word clouds capture the frequency of words in a body of content. The size of the word scales with the frequency it appears. There was no overwhelming leader for word frequency in posts (left) most likely due to the broad range of topics and interests covered on Harvest. Our community is a place for niche expertise and general advice. When looking at the readers word cloud it is immediately apparent that “growth” and “risk” were a big focus in 2016. This confirms what we determined above. Another word that stands out is “rate” as the Fed increased interest rates for the first time in five years3Readers find Harvest as a great source of information for investor letters and whitepapers, so they could find information on “portfolio” “returns” and “performance”, another stand out group of words.

Screen Shot 2017-04-20 at 11.33.10 AM.png


Categorizing content is a problem we face at Harvest since a single piece is usually about multiple, complex and subtle topics. Based on stock tickers in the post and the number of opens, we gave each post an interest score for value stocks and growth stocks. For example, a post could mention five stock tickers that are considered growth, and three that are considered value. Instead of bucketing this post as only a “growth” post, we decided to create an interest score by looking at the number of mentions of these tickers. The same goes for opens. If the post had 100 opens, growth would have 500 effective opens for growth, and 300 for value.This gives us more quantifiable data as opposed to tracking mentions of the words “growth” and “value”, which may also be used in other contexts.

Then once we have the stock tickers, we decided to use the industry standard for grouping these stocks: the Russell Growth and Value indexes4.  To get a sample of stocks that are actively traded in each of these groups we found the ETFs that tracked these indexes5,6. Each of these ETFs actually are composed of around 700 stocks; even then some overlap between the two. To make our analysis more specific, we removed the stocks that overlapped from each group. Ultimately, this left around 400 unique value stocks and 300 unique growth stocks to track (see the figure to the right).