So, it looked to me that Hollywood could finance an asset (movie), by selling “options” (pre-sales), in a very specialized, over-the-counter market. That meant that the risk of a movie (like Lara Croft’s Tombraider) being a box-office disappointment was passed to third parties, in exchange for those guys getting the distribution rights in their respective territories.
My analysis also showed that the pricing mechanism was very inefficient, because it was not based on any analytical techniques, but on subjective opinions about the ‘hot’ stars, genres & directors of the moment. As an additional hurdle, Hollywood was subject to the capital constraints of their parent corporations, which are exposed to risks larger and diverse than those faced by their Hollywood branches.
With a lot more trips to Hollywood and a little more data gathering (it is hard to find the right data and you have to go to the source), we were able to build a “Black & Scholes” model mixed with Data Science that attempted to price movie rights as stock options, territory by territory, with the US being the most valuable for US movies at that time. (Btw, the main picture in this post is one I took of the swimming pool at the Roosevelt Hotel, one of my favorite spots in that city and where I usually stay).
After many trips to Los Angeles and meetings with famous producers, directors and even stars, we ended up developing a better version of what the Hollywood guy from the Forbes article was using. I want to take the opportunity to thank my friends Kevin Hicks and Lawrence Kubik, who you might recognize from his 80s TV shows and movies “Hunter” (my father’s favorite), “Death Before Dishonor”, and his mentorship of Sly Stallone during his Rocky I days, Linda Lichter, and Ruth Vitale (former President of Paramount Pictures Classics), for helping us with introductions in Hollywood.
Continuing with the description of some aspects of our model, in addition to Monte Carlo simulation, it used an interesting methodology that was rumored to be used in some aspects of Google’s search engine: Markov Transition Matrices and lots of optimization techniques. By building some web scrapers with Python, I was able to gather a lot more data from public sources and feed it to our model.
Anyway, we set up another meeting with “Steve”, and started to ask basic questions to square our view of the facts. Then, my key question: “Steve, if instead of $10 million for your 10 movies, we source $100 million from hedge funds and investment banks; could you get this-and-this star and director for your movie?” After a few days, “Steve” came back with a positive answer for a few of the names we gave him. The model was indicating that the right combination of inputs could make a few of Steve’s movies analogous to an “indie” Lara Croft. At least, with more capital backing and bigger budgets, the movies would be less risky to produce, contrary to conventional wisdom, and with some creative financial engineering & accounting, and tax twists, we might have been able to produce a “zero cost” asset!
So this whole “Steve” incident that started as an interesting meeting with an independent producer ended up with me going to work for Lehman Brothers in 2008, and my partner Ralf Voellmer running SAGA Capital, in charge of teaming up with hedge funds to launch the SAGA Capital Alpha Movie Fund. But before accepting the job offer with Lehman, I had met with hedge fund Elliott Management Associates, 2 other large hedge funds, with Goldman Sachs, and with a Moody’s analyst, who were interested in my model. They had seen the Hollywood guy’s model, and were not particularly interested in it.
I was impressed with the Goldman Sachs Quants – they tried to drill me in the analytics left and right. I gave them as much information as I could without disclosing the ‘secret sauce’. Once I informed them that the math was coming from hybrid financial quantitative analysis and data science models I developed (let’s call it “Data Investing”), they were convinced about the effectiveness of my model. “Mr. X” at Goldman told me that their Quants were starting to experiment with precisely that sort of analytics applied to movies, so they were confident that my model was using the right framework.
Unfortunately, Goldman could not do business with SAGA at that moment. The timing wasn’t right, and other things that we did not know were happening in the background, potentially affecting structured finance deals not just for Goldman but the whole banking sector…. Lehman Brothers was also very interested, but they had not done as much work as Goldman in the film area. And Ralf and I still had one issue to solve: the model was forecasting results for US movies beautifully, with an out of sample R squared factor of over 90%. Nevertheless, forecasts for countries other than the US required a little more work on our part, and a little more traveling, but not to LA this time. The Cannes Film Festival was our next destination…
To be continued..