Flostock news #9 The volatility in commodity prices

 

Alternative 1 to the Flostock forecasting method: Extrapolation

Immer Gerade Aus.

Simplest forecast is a quantitative extrapolation of the past. This assumes that the future is somehow a logical function of the past. The simplest extrapolation is a straight line, but if it is a bended curve or even an S-curve we can still call it extrapolation. One may add trends, seasonality and cyclical variations. Major advantage of extrapolation is that it is simple and easy to understand. The disadvantage is not failure to address Black swans, because that is true for all forecasting systems. Nay, the major disadvantage is that all non-linear effects are omitted. For example the inventory effects of the supply chains are certainly non-linear and thus can be surprisingly counter-intuitive. 

The simple extrapolation has been made more complex over time by using a weighted moving average,  by Exponential smoothing, by Autoregressive moving average (ARMA), or by Autoregressive integrated moving average (ARIMA). It all remains the same: extrapolation of the past. 

Stocks have a memory, according to Flostock’s 9th Law.

Stock of DRAM has a memory

Stock memory is like potential energy, ready to start flowing. A lake full of water for hydro-energy. A bucket of marbles balancing on a hill top.  A syringe that is drawn vacuum. An under-stocked warehouse. Helper T cells. A hill full of snow in spring. An empty shelf. A jungle full of pharmaceutically interesting plants. An empty terrace with a good view. DRAM memory. All savings on the bank. A pension fund. A group of hooligans in a bus. A full battery. Hot water in a boiler.  So I use the word “stock” as in stock & flow, not just as a physical inventory. The flow from the stock gets direction from the equilibrium level: when snow melts it flows down. When people retire the pension fund sends them money. When the syringe is released, it will close. An empty shelf will be filled. Low inventory creates replenishment orders to fill a gap.  In thermodynamics  they call this equilibrium a “lower entropy level”.

In Flostock models this property is used for a flying start: a model can be started at any moment if the stocks (memories) are filled to the right level. This memory property makes it also possible to forecast non-linear: or, to say it in other words: the absence of stocks in a forecasting model makes it impossible to predict anything else but straight lines. This, we believe, is the main reason why most institutions have their forecasts wrong at every turn of the economy.

Measuring forecast accuracy with MAPE

Hit the MAPE Bull's eye

The most frequently used measurement is MAPE: Mean Absolute Percentage Error. Basically it is the error divided by the actual value. The Flostock models contain these calculations by default.  Dynamic as we are, we offer them in two varieties: as a value, so the average MAPE for a period, or as a curve, so the MAPE as it develops over time. A value is easier to talk about and easier to compare, but  the question always is how far into the future are you doing this? The average MAPE value of 90 on a scale of 0-100 is less impressive one month into the future than 12 months into the future. As they say: “Forecasting is difficult, especially the future.” And the more future, the more difficult. 

Flostock models for commodity pricing

Price setting is easy

In a free market, price depends on the supply/demand balance. Flostock has established that in the huge supply chains of big bulk products, with strong capacity constraints in production and in storage, the supply/demand balance also determines the price. As a complication, the strong volume constraints ensure that  price elasticity up and down is very high.  If a product is short, price goes through the roof because customers cannot refuse to buy. If it is long, it drops to variable cost or even lower because producers do not want to close their installation. If demand is within capacity + storage boundaries, the price is reasonable.

Run-aways in demand and price can be modeled for the past and their likelihood for the future can be predicted. This price/supply/demand balance can be coupled to the standard Flostock model that already contains demand, inventory, capacity,  cost and margin. The strong volume restrictions in commodities can thus explain most of the price volatility.