At the end of 2015, the S&P 500 closed at 2,043, just under the original end of 2015 targets of sell side strategists predictions at the beginning of the year. Looking at the target prices of analyst reports, most of the time you would think that everybody is using the same model or has the same outlook on the market.
I think it is extremely difficult to predict the target price of anything, so I am not going to talk about how these targets are off or inaccurate because a lot can change over the course of a year to make these original predictions irrelevant. However it’s the nature of the dispersion of forecasts that I am more interested in. Theoretically the set of forecasts is unbound. You can publish a report saying the market will rise up to 10,000 next year if you wanted to but it’s highly unlikely, so realistically it is bound. Let’s say it’s bound by E[Y(t+1) = [70%Y(t),130Y(t)] a reasonable 60% range in which the market could fluctuate. Then you make your model and you find that you have some result like 90%Y(t) or 90% of where the market is trading at the end of 2015. You’re happy with your result and you go to publish it, but just before you release it a coworker suggest you check other forecasts. You notice that all of the forecasts on the street are at 105%Y(t). Do you change your forecast on this new information? Is it relevant?
In a few ways it is relevant. If you were camping in a forest and you knew that there was 10% chance of seeing a bear, but then passing by another camp of people, they tell you that there’s a 40% chance of seeing a bear you’re not going to ignore that information. They too have been camping in the forest so you value their thoughts and are going to update your odds accordingly. Now let’s consider a publication environment. If there exists a cluster of predictions, is there a difference between being at the edges of the cluster versus being outside of the cluster? Well if I am bearish on the market and the lower end of the Street is 2100 do I really gain more by saying 1900 versus 2050? By being closer to the cluster, I limit my downside in the chance that the market rallies during the year while having similar upside in the chance that the market falls. In the upside scenario, my prediction is relatively more accurate than everyone else’s. It’s simply an English auction in which even though you may have an intrinsic value/prediction, your outcry value is going to be linked to the publicly available information.
- How to Be Less Terrible at Predicting the Future: A New Freakonomics Radio Podcast
- Episode 677: The Experiment Experiment
- How The FiveThirtyEight Senate Forecast Model Works
- A Fantasy Sports Wizard’s Winning Formula
- Wall Street’s forecast for 2016: Worse than 2015’s
2015 in a lot of ways was the year of the podcast. It marked another year of consistent growth of listeners, with the share of US population who have listened to an audio podcast up 17% from 15% (keep in mind the additional population growth of the US population). Podcasts in a lot of ways reflect what online video platforms (more YouTube than Netflix) were to television. In the past, entertainment was produced and targeted towards the median consumer. This makes sense because in a time when channels were broadcasting to the general public, you want to produce content that the median consumer can relate to and find entertaining. YouTube changed the way entertainment is made. Channels now are more focused and niche than ever before with ever increasing “personalization” of entertainment. This makes sense because YouTubers don’t broadcast or stream in real time, they make videos that are viewed after the fact. So in static consumption, the optimal move is to create content that viewers can relate to more than what is currently out there (mainstream). You take this and iterate over a few periods and you’ll see that YouTube Channels focus on specific markets and demographics rather than the “median YouTube user.”
Podcasts have the potential to do the same. 60% of the US population listened to the radio at least one time a day in 2015 (an extremely large market). Radio channels have greater diversification in interests and demographics than TV channels because their startup costs are smaller, but podcasts lower the startup costs even more. Like YouTube, the key here is cost of production. Online podcast platforms help new podcasters find their target audiences and create on demand content that is more relatable and personalized than their radio counterparts. Does this mean radio is going to die off? Of course not, they will move to podcasting as well.
- Podcasts Rise In Popularity, Funded By Advertisers And Listeners
- The Rise of Podcasting: The Ultimate On Demand Content [Infographic]
- So, Like, Why Are We So Obsessed with Podcasts Right Now