Quote from: RE on Nov 29, 2024, 12:07 PMYou're supposed to be the best suit at the table according to you.

I am not a suit. Money handlers, insurance salesmen, bankers and financiers, lawyers. Never done any of that in life.

Quote from: REInquiring minds want to know.  If you don't answer or equivocate, you get cooler time.

But of course.

So..here we go.

The standard 6 variable model used to do this type of projecting involves 7 main components, requiring they are ranked in order from most important in terms of overall uncertainty for a given scenario to least. Each of the 7 are then assigned a weighting range, the sum of which on any iteration must equal 1.0. This means that in any given run, by the time the probabilities are calculated from most to least uncertain, if that number exceeds 1.0 on any given run, the remaining 1 or 2 might be entirely eliminated from the calculations.

Brent prices in US$ and all underlying variables examine data from both prior model runs and now historical results and project from them forward. One of the most important of the 7 inputs is the squared deviation from the past 12 model runs compared to what is now history...when running the current month and moving forward in time. High detail information like country level production is grouped into logical regions in order to improve overall run times of the simulation. Some parts of the model are data, some are derived more from a delphi approach, for example expectations of world oil balances in terms of production based on data and expectations of demand based on historical patterns which result in forward looking world storage balance changes. Changing storage balances act as a overall gauge of the supply/demand balance. Stated OPEC guidance is within the model, as are what is called "unstated" guidance which is designed to match actions as oposed to just words. Exogenous events can work their way into now and future casting as well using similar modifcations to the appropriate category among the 7. The weighting and order for the 7 can be adjusted as necessary. Expected forward year oil intensity of national economies is a relatively new addition with the growing offset of energy once hydrocarbon based now something else (renewables, substitution of non-hydrocabon based fuels, etc etc).

Prices are the changing variable to reach a required equilibrium point on each iteration. Iteration results in terms of price required to balance all weighting and other conditions results in a single output, 5000 iteration is usually reasonable for the resulting distribution to stabilize.

These are the monthly expectations of price and probability for 2025.



If you look carefully, of note is that the thin white line across the middle (or close to it) of the box part of the plot is a median, and the eye should be able to pick up naturally that more than 50% of the date tends to reside on the lower side of the medium Brent price for any month. Downside risk is more apparent over the coming year than upside. 5% of the data is above and below the visible data, but 90% of all prices lay between the maxima and minima vertical lines. Also of note is the tendency for late next summer after demand tends to slacken as it does seasonally, prices do not look to recover to the same levels as they are expected to enter 2025. Because all the fractiles of these probability constructs are known, they can be compared directly at each 1$ point, and the odds then calculated as to the over/under at any point in time between months if there was an interest.

The question you asked wasn't about sochastic model results for 12 months in 2025, but what is the average price for the entire year. To plenty of people this is a single number. A single number for those who can't handle sharp objects is pretty standard. I'm betting that folks don't show you what is being provided here because I know I am not the only one doing it. But we all dumb it down to single numbers for internet denizens, newspapers, 2nd grade readers and suits and whatnot.

In order to create a reasonable annual average from 12 distinct distributions of probability, you use all 12 monthly distributions to create a single annual average, and just run 5000 iterations to populate another distribution for the annual answer. And then you present that distribution as the answer for annual oil price....while keeping the proper uncertainty contained within all of the 12 months.

So my answer for the most likely price of oil (in this case a mean) in 2025 based on all the individual months and their accompanying uncertainty is $65.83, give or take. Using that as a reference point on the graph you can then eyeball for yourself the accompanying range and probability of 90% of most outcomes for 2025. Only a 3% chance of the price being more than $70, but it does exist.