UPDATE: My nerdy bin packing Monte Carlo simulator strategy has been replaced by my nerdy bin packing estimate comparator. The original page for the Monte Carlo simulator continues below so you can see the evolution in my Fantasy Movie League strategy.

I have a BS in Computer Science from UC San Diego and have worked in the computer industry since 1992. If those facts don’t scare you, read on.

Have you ever packed the trunk of a car for a vacation? You have bags of varying sizes you somehow have to make fit together in a limited space. It turns out, that style of problem is pretty common and generally called bin packing. In computer programming school you learn a lot about bin packing because resources like CPU, memory, and disk space are finite regardless of whatever problem you’re trying to solve.

## Fantasy Movie League = Bin Packing Problem

Fundamentally, playing Fantasy Movie League is a bin packing problem. You typically have 15 movies to choose from per week, each with an associated “fantasy bux” value (aka, $FB). You have up to 8 screens you can slot those movies into without exceeding $FB 1000 to create a legal combination.

Depending upon how the movies are priced in $FB each week, there are roughly 200,000 legal combinations you can chose from, but many of them are stupid choices. For example, the cheapest $FB movie on one screen and nothing else is a legal choice but not one that makes a whole lot of sense if you want a high score.

So, the first thing my nerdy bin packing Monte Carlo simulator does is figure out all the choices for a given week and then it eliminates the stupid ones. That usually leaves a couple thousand combinations of 7 to 8 movies that I refer to as “valid combinations”. That’s still too many to go through manually to make your final selection each week, so I needed another way to whittle down the list to something a human could reasonably look through.

## Tuning with Monte Carlo Simulation

Do you remember when Nate Silver got famous during the 2008 and 2012 Presidential elections by predicting correctly how every state would vote? I love that guy, read his book, and visit fivethirtyeight.com daily.

Essentially what he did there was, he weighted polls in each state according to their historical bias and then ran a Monte Carlo simulator with the error margins. I do the same thing to find the best combinations available from the couple hundred that the bin packing analysis leaves me with. Here’s how.

I find an estimate of next week’s box office for each movie, usually from ProBoxOffice.com. Then I just assume it is correct within +/- 10%. I run a simulation of the week where, for each movie, I let a dice role determine where along that +/- 10% that particular movie will generate box office revenue and I use those numbers for each movie to tabulate a Fantasy Movie League score for each of the couple hundred combinations. That’s a single simulation. Rinse and repeat 10,000 times and for each simulation, I tabulate whether or not a particular combination was in the top performers.

That gives me a list of around a couple dozen combinations which most often came out as top performers. That’s reasonable for human consumption. Usually what happens at this point is there are a few main combinations featuring permutations of high $FB movies that surface, which I refer to in articles as “broad strokes” combinations.

## The Human Element

Finally, I use my own brain and information like reviews, movie genre, cast, and other things to make my final picks. It’s probably 85% nerdy math and 15% me. I pick and write about a top 3 and throw in a wild card each week just for fun. That’s my secret sauce. Use it wisely.