Like many engineers, if I’m given a particular problem to solve I have a tendency to research, tweak, modify, and optimize an algorithm for days and days until either a deadline arrives or a supervisor tells me I’m finished.
As she says, you can do a project (a) quickly (b) cheaply (c) well. Pick two.
But then there is the anomaly of Joel Greenblatt’s Magic Formula in the field of investing. It seems to do all three, quickly processing thousands of stocks, using a simple algorithm, and beating stock market indices left and right (or so the historical backtesting says).
Recently, I’ve noticed that it seems to be taking longer and longer for me to plan where to stay for vacations and weekend getaways. I think the main problem is that I’ve accumulated a list of so many helpful travel websites that it seems like I spend hours and hours gathering lots of great data, but have a hard time boiling it down and coming to a decision.
So, I wondered, could there be a simple Magic Formula for picking a great vacation lodging?
Although Greenblatt certainly had a huge set of metrics he could pick from in ranking stocks, he ultimately decided to use just two: P/E and ROIC. Therefore I decided my Magic Vacation Formula would also use only two. I’m sort of after the same thing – pay a little, receive a lot.
Step 1: Sort for Popularity
If you’re a convert to the efficient markets hypothesis or have dabbled in prediction markets, you believe in the wisdom of crowds. That is to say, if most people love a certain hotel or resort, the chances seem high that you will too. So our first step is just to go to tripadvisor, enter our city and dates, and sort the results for popularity. If we’re after crowd intelligence instead of expert opinion, be sure to uncheck the Show Tripadvisor Top Values First box.
Step 2: Sort for Price
If money were no object we might just stop with Step 1. However just because a certain hotel may offer a wonderful experience, it’s not necessarily a good deal for the exact time when you plan to be there. So Step 2 is to again use tripadvisor (or whatever your favorite deal site is) to find the price per night for each of the lodgings from Step 1.
Step 3: Rank and Multiply
For lack of a better algorithm I now do exactly what Greenblatt does. I sort the list for each of the two factors, multiply the resulting ranks, and I now have the combined metric. The best deal for the best experience should rise to the top.
Step 4: Tweakage
Despite my best intentions, in practice I’ve found it hard to just stop with Step 3 and book at wherever had the best combined rank. Like restaurants, some lodgings are popular not so much because they’re great but because they’re dirt cheap. So if it’s for a special occasion, you may want to pass over lodgings rated less than 3 stars. Other places have a potential sample size problem (a high ranking from only a couple reviews). You’ll also want to read a couple of the latest reviews to make sure that something hasn’t recently changed or that a noisy renovation isn’t currently underway. The reality is, the #1 ranked might not be #1 right now or anymore, so keep the top 3 or 4 and use your judgement to pick the best.
Step 5: Modern Portfolio Theory
Use your judgement?! Isn’t this what we’re trying to avoid?! To help narrow it down we can take a tip from Markowitz’s efficient frontier, where when comparing investment vehicles we’d be nuts to choose one that had more risk for equal or less expected return than another. Equivalently, when examining your sorted list, I say go ahead and throw out any lodging possibilities if another exists that has an equivalent or greater star rating for a lower price.
I should close with something Benjamin Graham-ish, like stating that we may not expect the Magic Vacation Formula to work in each instance, but to deliver satisfactory results on average for a basket of vacations. Can you tell that I’ve had extra time on my hands while waiting for the CFA exam results to be released? Use the Magic Vacation Formula at your own risk!