NuScore v5.5: Your Moat Now Moves Your Score

## We Audited Our Own Scoring Engine. It Was Too Nice. We scored 1,893 pitch decks. Then we checked which ones actually got funded. The results were uncomfortable. Our green flags — the signals we told founders were strengths — didn't predict funding. And the strongest predictor of all? It wasn't in our scoring system. This is the story of NuScore v5.5 and why we rebuilt our flag system from the ground up. ## The Data That Changed Everything We cross-referenced every scored deck against real funding outcomes: who raised, how much, and at what stage. Three findings stood out. ### Finding 1: Patents Predict Funding at 100% | Signal | Decks | Funded | Rate | |--------|-------|--------|------| | Patents + proprietary data | 17 | 17 | **100%** | | Patents alone | 5 | 5 | **100%** | | Proprietary data only | 32 | 31 | **96.9%** | | No IP | 798 | 509 | 63.8% | Every company with a patent in our dataset got funded. Not most — all of them. And their average score was 8.33 versus 6.24 for companies without IP. But NuScore was only giving patent-holders +0.42 extra points. The engine detected moats but barely rewarded them. **Fixed:** IP is now a first-class green flag. Patent detection, proprietary data, network effects, regulatory capture, and switching costs are surfaced as structural signals on your report. ### Finding 2: Half Our Green Flags Were Wrong Our most common green flag — "Diverse founding team" — appeared on 18% of decks. But in the training data, diverse teams funded at 51.9%. That's *below* the 64% baseline. Same story for "Prior exit." Serial founders funded at 43.7% at seed. First-time founders? 55.1%. The AI era hasn't changed this — serial founders go after harder problems. "Perfect Market Timing" was supposed to indicate opportunity. But decks with timing signals funded at 56.8%. Below baseline. We were rewarding narrative, not substance. **Fixed:** We removed score adjustments from flags that don't predict funding. If a signal doesn't show up in the outcome data, it doesn't move your score. ### Finding 3: Solo Founders Fund at 15% The code literally said: *"Solo founders are NOT a red flag in the AI era."* The data said otherwise: | Team Size | Seed Funding Rate | |-----------|-------------------| | Solo | **15.4%** | | Two | 56.0% | | Three+ | 48.4% | Solo founders are the lowest-funding-rate signal in our entire dataset. Not slightly lower — dramatically lower. **Fixed:** Solo founder is now a medium-priority red flag with specific coaching: "Consider adding a co-founder with complementary skills, or build a strong advisory board." ## What Changed in v5.5 ### New Green Flags (Data-Validated) - **Patent protection** — "100% funding rate in training data" - **Proprietary data advantage** — "96.9% funding rate" - **Network effects** — winner-take-most dynamics - **Clinical/regulatory validation** — hard barrier to entry - **Domain expertise (5–15 years)** — the sweet spot at 73.4% funding rate - **Complementary team (3+)** — teams with diverse skills fund at nearly 100% at growth stage - **Hyper-growth (30%+ MoM)** — upgraded from minor to major signal - **Strong retention (80%+)** — upgraded from minor to major signal ### Removed / Downgraded Green Flags - **Diverse team** — removed from scoring (kept as matching tag) - **Prior exit** — downgraded from major to display-only - **Market timing** — downgraded from major to minor - **"Revenue generating"** — tiered: $1M+ ARR is a signal, $10K is not ### New / Upgraded Red Flags - **Solo founder** — new medium-priority red flag - **No GTM strategy** — now flagged at seed, not just Series A - **No competitive analysis** — now flagged at every stage - **Healthcare without regulatory strategy** — penalty increased 50% - **Fintech compliance at seed** — removed (it's data room material) ## The Defensibility Hierarchy Our data reveals four tiers of defensibility. The higher your tier, the stronger your score: **Tier 1 — Hard IP (100% funding)** Patents, clinical validation, regulatory approval **Tier 2 — Knowledge Moat (67–97%)** Proprietary data, regulatory capture, deep domain expertise **Tier 3 — Structural Moat (47–62%)** Network effects, category creation, 10x product improvement **Tier 4 — Soft Moat (38–53%)** Proprietary system without patent, deep tech R&D If you're building in healthcare, deep tech, or biotech — file that provisional patent. The data says it's the single strongest signal you can send. ## What Didn't Change The core architecture is the same: 6 scored dimensions, LLM judge + rules cross-check, score waterfall, raise probability. NuScore v5.5 isn't a new engine — it's the same engine with better calibrated signals. Your score might shift slightly. If you have real defensibility, it'll go up. If your previous score was propped up by narrative signals that don't predict funding, it'll be more honest. Both are improvements. ## Try It Upload your deck at [nuvc.ai](https://nuvc.ai) and see which defensibility signals your report detects. The new flags appear in your score breakdown alongside the coaching they unlock. If you've already scored — rescore with the latest engine to see your updated flag profile.