Party draw simulations

The current combined model was rerun after the rulebook review. It includes 294,912 draw runs: six species × three pools × eight scenarios × 2,048 runs. Four party/choice-linkage scenarios are each run with two capacity assumptions. Shared draw positions preserve dependence between the three selected hunts, and each group must fit in full.

For 35 deer codes, including private-deer competitors hidden from the explorer, realized awards were below the available license total. The restored-capacity scenario retains observed awards, adds unissued nonresident and outfitter capacity up to floor-rounded 6% and 10% shares where possible, and assigns the remainder to residents. For resident-only hunts, existing outfitter-category awards are retained; that category does not establish nonresident eligibility. Unissued licenses are assigned to the resident category in this scenario. The observed-award scenario is also retained. These are sensitivity assumptions, not the Department's complete rounding or rollover algorithm; mixed-residency parties and discretionary adjustments remain unmodeled.

License reductions already reflected in the 2026 inputs—such as 2B's 400-to-320 change—are not applied a second time. The main simulated draw percentage is the average of the eight scenario probabilities, displayed to two decimal places so small party-size differences are easier to see. Extra display precision does not establish accuracy. Positive estimates below 0.01% and estimates between 99.99% and 100% use threshold labels; 100% is explicitly labeled as a simulation result. It is not the midpoint of the model range. The model range is shown in “How we estimate this” on this page and includes scenario variation, Monte Carlo uncertainty, and the reconstruction sensitivity allowance described below. It is not a validated real-world confidence interval. More simulations reduce sampling noise, but do not reveal missing linked choices or party composition.

We tested whole-party tag allocation with exact mathematical examples and 144,000 synthetic elk draws. See how party size and competing applications affect the results for 16D, 16A, and 16C First Archery.

Read the simulation results

These experiments test the draw mechanism and sensitivity to missing applicant information. The application outlook and its per-choice estimates use the same shared simulation dataset for six species. Historical research screens use separately labeled statistics.

Planning year and starting selections

2027 planning with 2026 data. The application labels identify the draw year you are planning for. The latest available draw data, currently 2026, is the baseline for application and party estimates. Supporting historical records cover 2024–2026. We are not forecasting next year’s applicant numbers. Dates shown with hunts identify their reference season.

Outfitter suggestions compare each selected hunt’s contribution at its choice position in your current pool and the outfitter pool. Each comparison keeps your party size and other choices fixed, using the same simulation as Your choices. A suggestion appears for that choice only when both estimates are available, the selections pass eligibility checks, and its unrounded outfitter estimate is more than five percentage points higher. Application totals do not trigger these suggestions. The pool setting still applies to your entire application; opening a comparison does not change your selections.

New applications start with the most-applied-for eligible hunts at each choice position in the latest report, without repeats. These are examples, not recommendations. Popularity does not mean better odds. You can leave second or third choice blank.

Choice-specific draw history

Each historical choice rate uses successful applicants divided by applicants for the same hunt, applicant pool, and choice position in the official report.

Historical rates include all reported party sizes; they are not solo-only rates. In Draw Strategy, the headline and choice cards use the same application simulation for the selected party size and ordered hunts. Historical percentages and annual counts remain available under “Draw history & details.” Hunt Explorer and Ask Hunt Smart keep their separately labeled historical statistics and party-size approximations.

Party-size approximation

A party needs enough remaining licenses for every member. For each included year, the approximation subtracts one award for each additional party member, floors the result at zero, and divides the combined adjusted awards by combined entries. It has the same capacity-penalty shape as a simple one-hunt lottery with solo competitors. Real competing parties and linked choices can behave differently.

Hunt-specific party limits apply to the whole selection. GMU 5A private-land deer, bear, and bighorn allow only one applicant; ibex and oryx allow up to two. Private-land-only deer hunts remain excluded from customer selections. A missing or zero modeled pool capacity suppresses this approximation; a party larger than a positive modeled capacity produces zero within that scenario. Modeled pool capacity is not a verified legal allocation.

Year-to-year comparisons

“What changed?” now includes annual rulebook-table comparisons as well as applicant counts, awards, and historical draw rates. Rulebook changes cover dates, total license quotas, weapon, bag limit, area wording, and extracted eligibility and restriction labels. Each comparison links to the exact source PDF pages.

The review covers 6,847 big-game table rows from 2020–21 through 2026–27. All 869 single-row current-code comparisons now have source-checked 2025 and 2026 rows. The 18 unmatched codes are private-land-only deer hunts excluded from customer selections. Bighorn's nested area choices remain separate and are not included in the combined model.

We corrected 74 historical pronghorn matches that had selected a private-land quota table instead of the public-draw table. The same code can appear in both. Wrapped youth and military labels were checked, and the general resident-only rule for antlerless elk was applied across years; the 2026 label does not establish a new restriction.

Boundary review: the official GMU boundary rule, 19.30.4.8 NMAC, cites August 25, 2020 as its replacement date. That rule supplies the outer GMU definitions for the 2024–26 comparison window. Hunt-specific exclusions and named management areas are compared separately. An unchanged GMU does not establish an unchanged hunt area or public access. We do not infer replacement codes or certify parcel access from these tables.

Earlier records are omitted from draw and harvest averages when the area, weapon, bag limit, eligibility or specified restrictions differ or remain unresolved. This changed the included draw years for 50 hunt codes. Routine spelling and punctuation changes are normalized. Date and quota changes remain visible in “What changed?”; a historical rate still describes the years shown, rather than a quota-adjusted future forecast.

Scope: the reviewed general rules cover party limits, whole-party fit, youth age and hunter education, mobility certification, military eligibility and resident-only antlerless elk. Source-page footnotes are available in the expanded comparisons. All 6,847 extracted rows and area headings match their cited source text. Bighorn area seasons, dates, and license counts can be inspected by year in the expanded comparisons. Matching text does not establish that every hunt definition stayed unchanged. Live closures, landowner permission, and individual eligibility documents are not certified by this review.

Rulebook comparisons · Model input corrections and excluded years · Official GMU boundary rule · Application and eligibility rule

Second- and third-choice historical draw rates

A historical draw rate is the number of hunters awarded a hunt at that position divided by everyone who listed it there. The denominator includes applicants who received an earlier choice. It is not the chance of drawing this hunt conditional on your earlier choices failing.

The public reports do not link each applicant's choices, so that conditional chance cannot be recovered for a particular hunt. Later-choice rankings also reflect the earlier choices made by those applicants. Use those results as a shortlist rather than a ranking of conditional fallback opportunities.

For the included 2024–2026 resident records, second-choice awards total 37,182 among 556,947 entries (6.68%); third-choice awards total 28,411 among 523,235 entries (5.43%). Suggested lifts of 16.6% and 25.9% assume that earlier-choice winners appear in optional later-choice listings at the same proportions as the whole pool. Those assumptions are unverified; we do not apply those lifts or display them as conditional odds.

Choice strategy review

Each choice’s displayed percentage is its contribution to the current application estimate: the share of simulated applications awarded that hunt at that position. A contribution below 0.5% is marked “Very low.” A low first-choice contribution can still reflect an intentional priority. The second and third contributions depend on the hunts selected before them; they are not standalone or conditional fallback odds.

Eligibility and party-capacity concerns take precedence. “No wins observed in the simulation” means this selection received no awards in the modeled draws, not that drawing it is impossible. A repeated hunt shows “No added chance.” Unsupported species and unavailable simulations show no numeric estimate; historical records stay available in the expanded details.

What the warning system checks

Small positive draw figures display as <0.1% at one-decimal precision, or <0.01% at two-decimal precision, so rounding never turns a positive value into zero. “No recorded awards” describes the historical sample. “Estimate unavailable” means the party approximation reaches zero despite historical awards; it does not mean the hunt is impossible. Capacity failures are labeled separately.

• Party size exceeds a hunt-specific limit or modeled capacity.
• No simulated applications received a selected hunt at that position.
• A choice has a very low modeled contribution to the application.
• The same hunt appears more than once.

How we estimate this

The application range estimates the chance of receiving any of your selected hunts. Each simulated application gets one random draw position. We check your first choice, then second, then third, stopping when a hunt has enough remaining tags for your entire party. Blank choices are skipped. Repeating a hunt does not create another chance.

For each supported species and pool, we generated 2,048 draw positions under each of eight scenarios: two capacity assumptions combined with four competing-applicant scenarios— solo or mixed parties, with independently linked choices or a preference for the same weapon. All selections are evaluated at the same position within a draw, preserving dependence between choices. These are additional target applications competing against fixed synthetic populations.

The populations use the 2026 report's applicant-person counts at each hunt and choice position. A simulated application cannot repeat a hunt, and hunt-specific party limits apply to competitors. Assignment repairs preserve counts where feasible; remaining differences must be no more than 0.5% of first-choice people. Exact residuals and duplicate counts are published in the simulation manifest. Mixed parties use nominal size weights of 55%, 25%, 15%, and 5% for sizes one through four, restricted and renormalized to the species party limit. The same-weapon scenario uses an eightfold assignment weight. These assumptions are not observations of actual parties or linked choices.

The initial range runs from the lowest scenario's 95% Wilson simulation lower bound to the highest upper bound. We widen it by a reconstruction sensitivity allowance, then round outward in half-percentage-point steps below 10% or whole points at 10% and above, using the lower endpoint to choose the step. This is not a calibrated 95% confidence interval for real-world odds. A zero-success sample retains a positive upper bound. Exactly 0% is shown when the party exceeds every selected hunt's modeled pool capacity.

Capacities are held fixed within each scenario and pool, using the two capacity assumptions described above. Mixed-residency parties, cross-pool transfers, and special allocation adjustments are not modeled. Individual eligibility and actual choice linkages cannot be reconstructed from aggregate reports. Bighorn's nested hunt-area selections and the dominant once-in-a-lifetime linkage in the ibex pool are not adequately represented, so combined estimates for both species are withheld. Ibex historical choice records remain available. Bighorn guidance is in Q&A instead of the hunt dropdowns, odds, or rankings. A range is unavailable when required simulation or reconstruction data cannot be loaded or matched.

The engine passed exact mathematical allocation benchmarks. Historical reconstruction exposed differences in awards by choice, and out-of-sample predictive validation has not been completed. These are experimental model estimates, not official odds or validated forecasts. More simulations reduce random noise but do not resolve missing applicant information.

The choice cards show contributions from the same simulation as the application estimate. Each simulated application receives at most one hunt, so these mutually exclusive contributions add to the overall estimate before rounding. The headline and choice figures are displayed to two decimal places; displayed numbers may therefore differ slightly when added. Changing party size reruns the same draw-position samples with the new number of tags required. A larger party cannot increase the total in those fixed samples, but equal results are possible if no sampled draw position changes from a win to a loss. Historical counts stay unchanged when party size changes. The historical rates in the expanded details are separate statistics and must not be added to calculate the application estimate. The combined range uses 2026 application volumes and modeled 2026 capacities. It is a scenario under those conditions, not an applicant-volume forecast for the next draw. Future demand changes can move the result; historical-volume scenarios and held-out predictive validation are future work.

One harvest estimate across the site

Draw Strategy, Hunt Explorer, and Ask Hunt Smart use the same reported-hunter-weighted calculation for matched 2023–2025 harvest reports. Years with more reported hunters carry more weight. Published annual percentages are rounded, so the combined rate is an estimate.

Years with missing, invalid, or zero reported-hunter counts are omitted. If no usable year remains, the site shows n.a., not 0%. Each hunt shows its included years and reporting sample. Low harvest success is flagged consistently across all three screens using the comparison below.

Low harvest success for this hunt type

We compare each hunt with the full site catalog of hunts for the same species and weapon category: archery, muzzleloader, or rifle. Your selected unit, residency, party size, and other search filters do not change the comparison group. Private-land-only deer hunts are excluded from the site catalog. Youth and mobility-impaired hunts are included in their species-and-weapon group.

Each hunt contributes one reported-hunter-weighted harvest rate for the available 2023–2025 years. Missing or unusable rates are excluded. A group needs at least 20 hunts with usable rates before we flag its bottom 10%. Unknown weapon categories receive no percentile flag.

We rank from lowest to highest using the calculated rates before display rounding. A hunt is flagged only if the number of hunts with an equal or lower rate is no more than 10% of the group. We keep ties together: if a tied group crosses the cutoff, none of that tied group is flagged. This can flag fewer than 10%, and an all-tied group receives no flags.

The indicator means relatively low success compared with this hunt type, not a universal poor-success threshold. An archery hunt is not flagged simply because its success is below 20%. Sample sizes and included years remain available in the expanded harvest details.

How Ask Hunt Smart works

Ask a question in plain English. We’ll match hunts to your preferences and compare them using New Mexico’s published draw and harvest data.

Results use historical draw rates, party-size approximations and harvest reporting. Draw Strategy uses a separate simulation of your complete application. Each method is explained here. If a question needs data we don’t have, we’ll tell you.

Official sources

The 2024–2026 draw counts have been checked against named cells in the downloaded department workbooks. Modeled pool capacities are derived assumptions, not source-certified legal allocations. Two inconsistent source pool totals for 2024 DER-2-152 are preserved and documented.

Source verification results · Workbook hashes and named cells

Proclamation labels and elk management context are dated separately from the draw and harvest years.