What the SSDI Approval-Odds Estimator Actually Calculates (and Its Limits)

General information about how this tool works and about SSA's disability rules — not legal advice about your claim.

The Approval-Odds Estimator asks four questions and reports SSA historical base rates for claims that share those characteristics. It does not predict your outcome, and it is deliberately built so it cannot pretend to — this is what it computes, what it leaves out, and why a percentage is a starting point rather than a case evaluation.

What does the SSDI Approval-Odds Estimator actually calculate?

It calculates base rates, not predictions. From four inputs it looks up the historical allowance rate attached to each — the state's initial-decision rate, the condition category, the age band, and the stage the claim has reached — shows them side by side, and labels the overall picture favorable, mixed, or challenging base rates. The scoring is a fixed additive rule, not a trained model, and it never fuses the components into a single personal probability. The thresholds are plain arithmetic: each component adds or subtracts a set number of points — a Compassionate Allowance match adds the most, a stage rate at or above 50% adds the next-most, a state or condition figure well above its benchmark adds a point, an age figure well above it adds half a point, and any of the three well below its benchmark subtracts a fraction of a point — and a total of 3 or more reads favorable, 1.5 or more reads mixed, and anything below that reads challenging.

That design follows from what it is describing. SSA decides disability through a five-step sequential evaluation, and at the last step, "[i]f you can make an adjustment to other work, we will find you not disabled. If you cannot, we will find you disabled," considering age, education and work experience (20 C.F.R. § 404.1520(b)–(g)). Four dropdowns do not reach most of that.

Which four inputs does it use, and where do the numbers come from?

Four: state of filing (52 jurisdictions), primary condition category (15 categories, plus an optional free-text box for a specific diagnosis), claimant age band (8 bands, "Under 30" through "60-FRA"), and where the claim is now (initial, reconsideration, or hearing). State is first because the first decision is not a federal one: "State agencies make disability and blindness determinations for the Commissioner for most persons living in the State" (20 C.F.R. § 404.1503(a)).

Three provenance details:

• The state comparison is against the median of the 52 jurisdictions' initial favorable rates, not the mean — so outlier states can't drag the benchmark.
• A state's reconsideration rate is suppressed when that jurisdiction decided fewer than 30 reconsiderations in the month. In the May 2026 data Montana had 9 determinations and 0 allowances: literally 0.0%, and reading that as anything about a claimant's odds would be wrong.
• The condition and age figures are older SSA research — one 2013 Social Security Bulletin study of allowance rates by primary diagnosis and by age, covering 1997–2004 application years (Javier Meseguer, "Outcome Variation in the Social Security Disability Insurance Program: The Role of Primary Diagnoses", Soc. Sec. Bull. 73(2), 2013). The tool year-stamps them on screen because they are a single historical study, not a current recurring statistic — shown to compare conditions, not to predict a case.

Any site can embed this estimator free — the widget lives at /embed/estimate, and the snippet is at aissdi.com/embed. Copy-paste citation for the underlying state data: Social Security Administration, "State Agency Workload Data," SSA-SA-MOWL.csv (monthly) and SSA-SA-FYWL.csv (fiscal year).

What can't the estimator know about your case?

Start with the two hearing-level facts it never asks about: who your judge is, and whether anyone represents you. Neither is a footnote in the system — the regulations make an administrative law judge the decision-maker at that stage (20 C.F.R. § 404.929), and separate rules govern appointing a representative to act for you in your dealings with SSA (20 C.F.R. § 404.1705). Whether hearing decisions come out consistently from one judge to the next is itself a standing federal-oversight question: GAO's review of the hearings process is titled Additional Measures and Evaluation Needed to Enhance Accuracy and Consistency of Hearings Decisions (GAO-18-37). The estimator has no field for any of it — not your judge, not your hearing office, not whether anyone represents you — so nothing in its read reflects them.

It also never sees the evidence. Residual functional capacity is "the most you can still do despite your limitations," assessed on "all of the relevant medical and other evidence" in the record (20 C.F.R. § 404.1545(a)) — and that record is yours to build: "you have to prove to us that you are blind or disabled," and you must "inform us about or submit all evidence known to you" that relates to it (20 C.F.R. § 404.1512(a)). No RFC, medical records, work history, earnings, insured status, onset date or education goes into the estimate.

Why does changing the age band move the read?

Because SSA's own rules change with age. The regulations define a "younger person (under age 50)," a person "closely approaching advanced age (age 50-54)," and a person of "advanced age (age 55 or older)," with further rules for those "closely approaching retirement age (age 60 or older)" (20 C.F.R. § 404.1563(c)–(e)). Age then enters the medical-vocational "grid," which combines "the various vocational factors (i.e., age, education, and work experience) in combination with the individual's residual functional capacity" (20 C.F.R. Part 404, Subpart P, App. 2 § 200.00(a)).

So the age signal is an echo of a real rule, measured after the fact — not that rule applied to your file. Two of the grid's three vocational factors, education and work experience, are not inputs at all.

Does a Compassionate Allowance change the estimate?

Only an exact match to a listed condition does. SSA's fast-track works off what you report, not what is proven: "CAL selection software identifies cases for CAL processing based solely on the claimant's alleged impairments listed on the SSA-3368 Disability Report – Adult or SSA-3820 Disability Report – Child," and those cases are "processed as expedited cases" by designated examiners (POMS DI 23022.010). SSA's list names more than 300 specific conditions (POMS DI 23022.000).

The matching is word-bounded on purpose. Generic words alone — "cancer," "disease" — never match, because the listings are specific qualifying variants rather than diagnoses. On a partial overlap the tool says so: the condition "is not itself a Compassionate Allowance, but the list includes" the closest entry, and the base rates do not assume it. SSA runs a predictive model of its own and uses it only to triage — "[a] QDD case is an initial EDCS case identified electronically using a predictive model (PM)" with "a high degree of probability that the claimant is disabled" (POMS DI 23022.010). Even SSA's model routes work; it does not decide the claim.

Why is a percentage a starting point rather than an answer?

Because a rule that matches you almost exactly decides nothing. SSA's own grid directs a conclusion only "[w]here the findings of fact… coincide with all of the criteria of a particular rule"; where any one finding "does not coincide with the corresponding criterion of a rule, the rule does not apply in that particular case and, accordingly, does not direct a conclusion of disabled or not disabled" (App. 2 § 200.00(a)). If the agency's own lookup table refuses to conclude on a partial match, a four-input base-rate lookup certainly cannot.

Three hard gates also sit entirely outside the estimate: the 12-month duration requirement (§ 404.1509), work that is both substantial and "for pay or profit" ending a claim at step one (§ 404.1572), and insured status — generally "at least 20 QCs in the 40-quarter period" for a worker 31 or older (§ 404.130). A favorable base rate for a claim that fails one of those is still a losing claim.

Aside, for firms reading this from the other side of intake: the same component signals sit behind this site's Lead Scorer, with the same limits described above.

Questions people ask about the estimator's methodology

Why does the number change when I switch stages? Because the record changes with the stage, not just the odds. Reconsideration "consists of a case review" with "an opportunity to present additional evidence" (§ 404.913), requested within 60 days of receiving the initial notice (§ 404.909(a)(1)). At a hearing you "may submit new evidence" and the ALJ decides "on the preponderance of the evidence in the hearing record" (§ 404.929) — a different record from the one the state agency saw.

Should a "challenging" read stop me from filing? No. It describes historical aggregates and is blind to the strength of your evidence, which is the variable you can change — see how that distinction plays out for depression and anxiety claims.

What should I do with the estimate? Treat it as a question-generator: if the read leans on your age band, the grid and your work history are worth understanding; if it leans on a condition category, your medical documentation is where the case is won. More guides sit on the claimant hub, and re-running the estimator above with different inputs shows which component is moving the read.