How Value My Well estimates what a lease is worth
The data behind every figure on this site, the decline-curve method used to forecast it, the prices and discount rate applied to it, and the measured error of the result. Production data is public record from the Railroad Commission of Texas.
Where the production data comes from
Every barrel and every thousand cubic feet on this site is a figure an operator reported to the Railroad Commission of Texas, the agency that regulates oil and gas production in the state. It is public record. We do not estimate production, adjust it, or fill gaps in it — where the Commission’s record is silent, so is this site.
The record we hold begins in January 1993 and runs to the most recent month the Commission has published. Every page states the month its own data runs to, because that month moves and a figure without it cannot be read a year later.
A lease is identified the way the Commission identifies it: by its type (oil or gas), its Commission district, and its lease number. Lease names are not unique — there are several hundred Texas leases called some variant of “Smith” — so the name alone never identifies a lease here or anywhere else.
How the forecast is made
Future production is forecast by fitting an Arps decline curve to the production the lease has already reported. Arps is the standard engineering model for how a well’s rate falls over time, and fitting it to a lease’s own history is what turns a record of the past into a projection of the remainder.
A lease rarely declines along one smooth curve. Wells are added, wells are worked over, and a lease can sit shut in for months. The fit therefore splits the history into regimes and fits each one, rather than forcing a single curve through a shape that is not one curve. Months with no reported volume are treated as reported zeroes rather than as missing data, because for a shut-in lease that is what they are.
The implementation is in src/arps.js and src/forecast.js.
How the forecast becomes a dollar figure
Forecast volumes are priced at the forward curve for oil and gas as it stands on the day the page is computed, not at a fixed price and not at yesterday’s spot. A regional differential is applied, because Texas crude and gas do not sell at the benchmark, and post-production deductions are applied to gas.
The result is discounted to a present value at 12% a year for a royalty interest. A royalty owner bears no drilling or operating cost and carries no operational risk, which is why a royalty is discounted more gently than a working interest, which we discount at 18%. Ad valorem tax is deducted at 2%, because it falls on royalty owners too. Income tax is not deducted, because it depends on the owner and we do not know them.
The implementation is in src/economics.js and src/prices.js.
Whose value the figure is
The figure shown on a public lease page is the value of the whole lease, undivided — 100% of the remaining production. It is not any individual owner’s share, and it is much larger than one.
An owner holds a decimal interest, printed on their royalty check stub, and their share is that decimal multiplied by the lease figure. We show the whole lease rather than assuming a royalty fraction because any fraction we picked would be wrong for almost every reader, and a plausible wrong number is more dangerous than an obviously whole-lease one.
When an owner uploads a check stub, we read the decimal interest off it and value their actual share instead.
How accurate it is, measured
A forecast is a range, not a promise, and the range shown on each lease page is measured rather than asserted. We backtested the model across 310,906 leases: the fit was given only the history up to a cutoff date, and its projection was compared with what those leases went on to actually report.
Accuracy is not uniform, and each lease page says which kind of lease it is looking at. Large, steadily producing leases forecast well. Small, intermittent leases forecast poorly, because there is little signal in a handful of irregular months. The error band shown on a page is the measured error for leases producing at that rate, not a confidence of our own devising.
The implementation is in src/backtest.js and src/evaluate.js.
What this is not
The figures on this site are estimates of value. They are not an offer to buy, not an appraisal, and not investment, tax or legal advice. Nothing on this site is for sale.
An estimate cannot know things that are not in the Commission’s production record: the terms of the lease itself, a title defect, a depth severance, a pending assignment, or a well the operator is about to drill. A material decision about a mineral interest should involve a landman, an attorney, or a petroleum engineer looking at the actual documents.