Win Probability in 60 Seconds: How Fed-Spend's pWin Verdict Engine Fuses 6 Signals Your Spreadsheet Can't
GovWin shows tabs. Fed-Spend now shows a number. The pWin Verdict Engine fuses incumbent CPARS posture, GAO protest history, recompete vulnerability, NAICS competition density, price-to-win, and profile fit into one explainable win probability score. One click from any recompete in your pipeline.
The Question BD Teams Actually Ask (And Nobody Answers in One Screen)
Every capture call eventually lands on the same question:
"Can we win this?"
Not "is there an opportunity." Not "who is the incumbent." Not "what did they pay last time." Those are table stakes. The question is a probability with a reason you can defend in a Monday pipeline review.
Today most teams answer it one of three ways:
We just shipped the dish.
[pWin Verdict](/dashboard/pwin) is a single calibrated win-probability score that fuses six signals only Fed-Spend already holds, explains every component with the source field it came from, and runs in one shot from any contract in your [Recompete Pipeline](/dashboard/recompete-pipeline).
This is not another dashboard tile. It is the answer to "can we win this?" before you burn a capture week.
What Legacy Platforms Give You vs. What pWin Gives You
| What you get elsewhere | What pWin gives you |
|---|---|
| Incumbent name and award history (USASpending) | Incumbent vulnerability score from CPARS ratings, terminations, exclusions, GAO protest sustain rate, contract growth, and modification churn |
| NAICS award listings | Competition density: unique firms, dollars per firm, competition level for your NAICS |
| Generic company profile fields | Profile fit score against your saved NAICS, agency preferences, and capabilities |
| Manual pricing research | Price-to-win fit: where your target contract value sits vs. NAICS/agency pricing benchmarks |
| Separate CPARS lookup | Past-performance edge for your firm (when performance data is on file) |
| Set-aside flag on the solicitation | Set-aside leverage: whether your certifications match the opportunity |
GovWin IQ, Bloomberg Government, and GovSpend index data well. None of them fuse incumbent protest posture + recompete vulnerability + NAICS density + price-to-win + profile fit into one explainable pWin number with graceful degradation when a source is down.
That fusion is the moat. That is what we built.
The Six Signals (With Weights You Can Quote in a Pipeline Review)
The engine is deterministic. Same inputs, same score. No black-box LLM guess on the number itself.
| Signal | Base weight | What it measures | Primary source |
|---|---|---|---|
| Incumbent Vulnerability | 28% | How beatable is the incumbent on this contract? | `recompete_vulnerability_cache` + `contractor_performance` + `gao_protests` |
| Competition Density | 20% | How crowded is this NAICS? Fewer entrenched primes = better challenger odds. | `naics_competition_cache` |
| Profile & Domain Fit | 18% | Does this opportunity match your saved company profile? | `user_company_profiles` |
| Price-to-Win Fit | 14% | Is the contract value in the realistic award band for this NAICS/agency? | `pricing_benchmarks_cache` |
| Past-Performance Edge | 12% | Does your firm carry CPARS strength in this space? | `contractor_performance` (your UEI) |
| Set-Aside Leverage | 8% | Do your certifications match the set-aside on the opportunity? | Profile + opportunity set-aside type |
Graceful degradation is non-negotiable. If a signal is missing, it drops out and the remaining weights renormalize. A verdict with 4/6 signals resolved still renders a full dial, a confidence label, and recommendations. One source down never blanks the page.
Confidence scales with coverage:
A Worked Example: NAICS 541512 Recompete
Take a common recompete shape: incumbent Booz Allen Hamilton, NAICS 541512 (Computer Systems Design), Department of Veterans Affairs, $12M contract value, full-and-open competition.
Before we warmed the signal caches, a first-run verdict on this shape typically resolved ~2 of 6 signals (profile fit + set-aside only). Useful, but thin.
After pre-warming the top-spend NAICS caches (`541512`, `541330`, `541611`, and seven others), the same cold run resolves 4/6 signals: competition density and price-to-win now contribute real numbers from live benchmark data.
That is the difference between "we think we can compete" and "here is why, with sources."
Run it yourself: [Open pWin Verdict](/dashboard/pwin) and paste those fields, or start from the Recompete Pipeline (below).
One Click From Recompete Pipeline (No Retyping)
The highest-friction moment in federal BD is the handoff from intelligence to decision.
You find a vulnerable incumbent in the [Recompete Pipeline](/dashboard/recompete-pipeline). You open the contract detail modal. You see vulnerability factors (CPARS, protests, growth, modifications). Then you used to open a spreadsheet and retype everything.
Now you click Run pWin Verdict.
Fed-Spend carries incumbent name, UEI, NAICS, agency, contract value, set-aside type, and contract ID into `/dashboard/pwin?...&run=1`. The form prefills. The verdict runs. You land on a score, a label (Strong Pursue, Conditional, Long Shot), a headline stat, component breakdown, and actionable recommendations.
Zero retyping. That is HyperCognate in practice: the answer, not the ingredients list.
Outcome Logging: The Score Gets Smarter When You Tell It the Truth
Every verdict is a prediction. Predictions without feedback stay wrong.
After you bid, log the outcome on the pWin page:
Outcomes feed `pwin_outcomes` and roll into a calibration delta per NAICS sector once you have enough decided outcomes (5+ in a sector). The engine applies a bounded adjustment (+/- 15 points max) so your historical win/loss pattern nudges future scores toward reality.
This is how pWin compounds. Month one is generic weights. Month six is your firm's calibration curve by NAICS.
Free Teaser vs. Professional Full Verdict
We gate depth, not the first revelation (HyperCognate law #2).
| Tier | What you get |
|---|---|
| Free | Score, verdict label, headline stat, top component preview. Enough to know if the pursuit is worth upgrading for. |
| Researcher ($49/mo) | 5 pWin verdicts/month + full breakdown when within quota |
| Professional ($199/mo) | 25 pWin verdicts/month + full component breakdown, recommendations, outcome logging, calibration summary |
| Enterprise | 5,000/month for teams running pipeline at scale |
The free view still delivers a real number from real signals. It is not a fake "upgrade to see your score" dark pattern.
[Start the 14-day Professional trial](/pricing) if you are running an active recompete pipeline this quarter.
How to Run Your First Verdict in 90 Seconds
Path A: From Recompete Pipeline (fastest)
Path B: Manual entry
Path C: API (Enterprise / internal automation)
`POST /api/internal/pwin` with session auth or `x-api-key`. Same engine, same cache layer, same quota rules. Documented in [llms.txt](/llms.txt) for AI crawler citation.
Frequently Asked Questions
Is pWin an AI hallucination?
No. The score is deterministic math on fetched signals. AI quota gates access and logs usage, but the number itself is weighted fusion, not GPT guessing. Recommendations are templated from component gaps.
What if CPARS data is missing for the incumbent?
Incumbent vulnerability falls back to protest history, contract growth, and modification patterns from the recompete vulnerability model. The component marks itself unavailable only when no incumbent signals resolve. The rest of the verdict still renders.
Does pWin replace a capture plan?
No. It replaces the 45-minute spreadsheet before the capture plan meeting. You still need teaming, pricing, and technical strategy. pWin tells you whether that work is worth starting.
How is this different from GovWin's "Opportunity Intelligence"?
GovWin scores opportunities against your profile in their ecosystem. pWin fuses Fed-Spend-native signals (recompete vulnerability cache, NAICS competition density, pricing benchmarks, GAO protest posture) that GovWin does not combine into one explainable 0-100 pWin with per-field citations.
Can I run pWin on an active SAM.gov solicitation?
Yes. Enter NAICS, agency, set-aside, and your target price. Contract ID is optional. Profile fit and competition density still resolve.
The So What
Federal BD does not have a data problem. It has a synthesis problem.
You already have CPARS, protests, recompete timing, NAICS density, and pricing bands. You just never had one number that fused them with sources you could cite in a pipeline review.
[Run your first pWin verdict](/dashboard/pwin) on the recompete you are already watching. Or open the [Recompete Pipeline](/dashboard/recompete-pipeline), pick the highest vulnerability score on the board, and click Run pWin Verdict.
The spreadsheet can wait. Your Monday meeting cannot.
*Signal sources: USASpending.gov, FPDS, GAO protest decisions, FAPIIS/CPARS contractor performance, Fed-Spend `recompete_vulnerability_cache`, `naics_competition_cache`, `pricing_benchmarks_cache`, and `user_company_profiles`. Weights and degradation rules defined in Fed-Spend pWin Verdict Engine v1 (June 2026).*
Related Guides
More from the How to Win Government Contracts (2026) series