Drop a pin. Get your offsets, decline curves and first-year oil forecast in seconds.
OffsetIQ finds every public horizontal near your location — single wells and multi-well pads — fits Arps decline curves to Texas RRC production, and forecasts first-12-month oil with a model trained only on real public wells. EUR is shown as secondary context only: on held-out wells it does not beat a county median. Coverage gaps are shown, not hidden.
4,716
horizontals with well-level production
+ 1,661 multi-well pads (6,972 wells, lease-level)
19,955
RRC horizontals mapped (incl. gaps)
4,058
Arps decline fits
2026-06
RRC production through
Why offsets
The problem
Every location decision starts with “what did the wells around it do?” Today that means pulling RRC data lease by lease, fixing APIs, fitting declines in spreadsheets — days of an engineer’s time per prospect.
Offsets are the prior
In a factory-mode basin, nearby wells with similar laterals are the best public evidence for a new well. OffsetIQ makes that evidence instant, quantitative and auditable.
Honest by design
Only real RRC wells. Multi-well horizontal leases (where Texas reports production at lease level) are shown as pads with lease-total production — never split into invented per-well volumes. Leases that mix horizontals with vertical wells are flagged as coverage gaps. Every number is reproducible from the public pipeline.
Proof on real wells (held-out 2022+ wells)
PRIMARY FORECAST
21%
median error, first-12-month oil
vs 24% for a naive offset average and 27% for county median · n = 2,306
57%
of wells within ±25% (12-mo oil)
trained on 2013–2021 wells only, tested on 2022–2025 wells
SECONDARY
32%
median error, EUR
vs 31% for a county median · n = 1,462
Limitation: EUR forecasts do not beat a simple county median on held-out wells — use OffsetIQ for first-year oil and offsets; treat EUR as context, not a number to bank. What the model gets wrong →
Sources: Texas Railroad Commission Production Data Query bulk dump (2026-09-26) and RRC digital map well layers. Screening tool — not a reserves report. Full data provenance and model card →