This here badass piece of metal and wood is an old apple mill. When my wife and found it in the barn of the farm we bought last year, our minds were made up. We were going to plant an orchard. We were going to make cider. We were going to make brandy. No more than 100 gallons, of course, and entirely for personal consumption, in keeping with Connecticut General Statute §30-77. After all, anything else would be illegal.
Subtitle this, “Notes from the Barn”, I guess.
Like good little analysts who don’t want to play at macro tourism, we started from the bottom up. We tested and amended the soil where we were going to plant the new orchard. We researched each of the 19 apple varieties we were going to plant, from Ashmead’s Kernel and Binet Rouge to Vilberie and Yarlington Mill. We built a library of quantitative analysis on the acid levels, tannin content, sugars and yields of each. We assembled a matrix of pollination calendars and harvest calendars, mineral requirements and preferences of different rootstocks. We collected qualitative descriptions of the impressions tasters got from the particular mix of phenolic compounds like anthocyanins, phloridzin, rutin and catechin that exist in wildly different quantities across cultivars. We reviewed empirical data on the results of other producers with different varietals.
We were also thinking about portfolios. You see, a good cider depends on getting the right blend of sugars, acid, bitterness and aromas. A delightful eating apple might make an insipid cider. Your favorite honeycrisp, for example, possesses moderate sweetness, a good bit of acid and practically no bitterness or other aromas. It’ll do as the base for a commercial cider, but it’s not really what we’re talking about when we talk about a hard cider. But then, it is rare that any one apple ever is. Sure, there are balanced apples that can be fermented into a beautiful and curious expression of a single cultivar, but in general, what makes a cider is what is achievable through the combination of a variety of different cultivars with complementary characteristics.
The act of making apple brandy is simply the distillation of cider. Figuratively and literally. In America or England, you’d probably heat up your cider in a copper pot still. In Normandy, you’d probably heat it up in a columnar distillation apparatus called an alembic. When heated, the alcohol in your cider evaporates more readily than the water. Along with the alcohol and some water go esters, fusel oils, methanol and tannins. When you’re done with your distillation (or in the case of a pot still, two!), you’ve got a 130-140 proof spirit that’s ready to take on the character of a wood cask for a few years.
But there is a problem.
Remember that beautiful cider, that expression of acid, sweet and bitter that we built with the help of math, nature, intuition and experience? The mechanics of distillation don’t care about any of that. Once we start applying heat, certain things are going to evaporate and recondense. Others will not, or will do so in lesser quantities. In the alembic, what our cider tasted like doesn’t matter. What matters is what will survive the process of distillation. There is in this an important truth to any process:
The characteristics of a good eating apple are often not the characteristics of a good apple for making cider. The characteristics of a good cider are often not the characteristics of a cider that will be distilled into good brandy. The characteristics of a good asset class portfolio are often not the characteristics of an asset class portfolio that will be combined with others to make the best aggregate portfolio.
That usually doesn’t stop us from trying. One of my first jobs back in my public pension days was to design the framework for – and identify most of the allocations of – an externally managed US small cap equity portfolio. It was something I was going to be judged by professionally. It was a portfolio that would be published and measured against a peer group of US small cap allocations of other institutions. It was a sub-portfolio that would be a benchmark-compared line item in every performance report we sent to the board and our consultants. I had every incentive and every temptation to treat it like The Portfolio. I wanted it to be balanced. I wanted it to be all-weather. I didn’t want it to be dependent on just one factor. I didn’t want there to be any tails that could wag the proverbial dog. Let me tell you – this was going to be some good cider. It’s just too bad it was going to be made into brandy.
Look, I did this to myself, but this also happens at different scales in almost every investment house in the world. We all have strategic portfolio objectives, and we all dole out implementation responsibilities to staff or to third parties. In the interest of measuring these agents – an entirely legitimate aim – we nearly always create incentives for them to smooth out every rough edge of the pool of assets they are charged with overseeing. We prize sub-portfolio Sharpe and Information Ratios when we should be celebrating the marginal impact on the risk efficiency of the aggregate portfolio. The result, very often, is a portfolio which sacrifices conviction for risk reduction that would have been achieved anyway when the portfolio was rolled up.
I cannot tell you how often I see this when we meet with investment firms of every type.
Asset management executives, do your stakeholders a favor. Ensure that the processes followed by your asset class teams and sub-portfolio managers are integrated into your aggregated risk and strategic asset allocation processes. Take caution when you incentivize staff or third-party managers to smooth out every edge of the portion of the portfolio they are managing. Allocators, develop comfort with asset class portfolios that have persistent structural biases within a balanced broader portfolio. Fight the temptation to systematically reward standalone sub-portfolio risk/reward ratios. Understand the factor loadings, environmental biases, betas, tilts and exposures of your asset class portfolios against whatever benchmark you’re using, but stop optimizing your portfolios at each step of the construction of the aggregate.
Stop forcing your people to make cider that’s going to spoil your brandy.
Rusty when I saw the word Alembic – I clicked expecting “The Elder Scrolls IV: Oblivion” images…
Thanks your cider /brandy analogy is very accurate. So true that allocators end up “over-fitting” the mandates of the individual components without thinking of the overall correlations as it is harder to visualise.
I always skipped alchemy, as the temptation to use it to cheat was just too strong for me – thanks for the response!