Skip to content

Temporal Patterns — TechPort Portfolio

Created: 2026-04-05 (session 3) | Data snapshot: 2026-04-04

Start Year Distribution

Source: techport_portfolio_aggregate, group_by=startYear, n=20,152 projects.

Year Count % Notes
2019 1,243 6.2% Peak year
2020 1,224 6.1%
2015 1,192 5.9%
2011 1,192 5.9%
2021 1,181 5.9%
2017 1,163 5.8%
2016 1,123 5.6%
2018 1,113 5.5%
2014 1,058 5.3%
2022 1,048 5.2%
2012 1,032 5.1%
2023 969 4.8%
2013 914 4.5%
2024 889 4.4%
2009 660 3.3% Ramp-up era
2010 578 2.9%
2004 530 2.6% Early adoption
2005 490 2.4%
2006 465 2.3%
2007 436 2.2%
2003 417 2.1%
2008 398 2.0%
2001 327 1.6%
2002 311 1.5%
2025 197 1.0% Partial year (snapshot Apr 2026)
1992 1 0.0% Single record
1998 1 0.0% Single record

Patterns and Interpretation

Three eras visible

Era 1 — Early adoption (2001–2008): ~400–530 projects/year. SBIR/STTR likely dominates but programs are being added to TechPort gradually. Counts below the steady-state reflect incomplete historical data entry rather than lower R&D activity.

Era 2 — Ramp-up (2009–2011): Jump from 398 (2008) to 660 (2009) to 578 (2010) to 1,192 (2011). The 2011 spike to ~1,200 suggests a large cohort was entered — likely a systematic backfill of earlier SBIR/STTR records or a new program entering TechPort.

Era 3 — Steady state (2011–2022): 914–1,243 projects/year. Reasonably flat at ~1,000–1,200/year. No dramatic program-level events visible. 2019 is the peak (1,243) — no obvious explanation without drilling into which programs drove it.

Era 4 — Declining recent years (2023–2025): 2023: 969; 2024: 889; 2025: 197 (partial year). The 2025 number is undercount — data snapshot is April 2026 so 2025 projects added after the April 2026 snapshot would not appear. More importantly, 2023–2024 show ~10–15% below the 2015–2022 average. Possible explanations: - Budget sequestration or SBIR/STTR program changes - Fewer projects entered into TechPort (coverage issue, not funding decline) - Structural: multi-year grants entered once, not annually

Anomaly: 1992 and 1998 records

Two records with start years before 2001. These are outliers, likely data entry errors or very old program records brought into TechPort. Not representative of the database.

STRG spike noted in Session 1

From session 1: STRG added ~195 active projects with start_date 2026-01-23 — a new grant cycle batch-entered. This kind of batch entry creates temporal clustering that isn't research activity clustering. When analyzing temporal patterns by program, batch-entry artifacts must be distinguished from real activity spikes.

What temporal patterns can and can't tell us

Can tell us: - Approximate program size over time (with caveat about entry timing) - When programs were added to TechPort (new programs show up as first-year spikes) - Major database events (batch backfills, format migrations)

Cannot tell us: - Actual funding levels (entry lag, and some projects span multiple years) - Research output per year (projects don't equate to papers, patents, or milestones) - Program starts vs. continuations (SBIR Phase I vs. Phase II appear as separate projects with different start years)

Open threads

  • Why is 2019 the peak year? Could be SBIR funding level, new programs, or TechPort adoption. Would need program-level breakdown by year to distinguish.
  • 2023–2024 decline: Is this real (budget pressure) or an artifact (projects not yet entered)? Check in 6 months.
  • Program-level temporal analysis: A few programs (STRG new cohort, MCO 40% active) show interesting temporal signals. Worth tracking separately.