August 19, 2018

Caroline and I went to see her grandmother today. She retired a few years ago and wanted some help optimizing her financial portfolio.

We got her set up with a bond ladder. One of the key questions when buying bonds is which term you want: 1 year, 2 years, etc. Longer terms mean higher interest rates (good) but more interest rate risk (bad) and your principal is locked up longer (bad). The solution is a "ladder": you define a length, say three years, and then buy equal amounts of bonds maturing in one, two and three years, reinvesting the one-year into a new three-year bond in a year. When done properly, this gives you a continuous flow of income at the long-term (three-year) rate, even though you get income every year, and your interest rate exposure is decently hedged; all in all, a win.

It's surprising how much laddering comes up in non-financial contexts. I used a similar "laddering" strategy when purchasing reserved compute capacity (RIs) for ops at Crittercism: we built a ladder out of expiring reserved instance contracts, which provide reserved compute capacity at discounted rates over a fixed term. The problem is similar to bonds: you want the cheaper prices, but not the lock-in inherent in a multi-year contract. The solution is to buy a set of multi-year RIs each year, so that they're constantly expiring, giving you the option to renew (at lower rates, if your compute needs are the same) or adjust things as needed.

I love cases like this where you can use ideas in one domain (cloud computing) that come from anohter (finance). I suspect there are many other cases where a laddering-type approach can reduce risk with a little bit of planning; feel free to write a comment if you come across another.