01 Headline outcome — Balanced Approach
Pre-tax by default; corrected for the identified spreadsheet errors, evaluated at fixed monitoring checkpoints rather than assumed perfect intraday timing, and shown alongside dispersion and benchmark context so the average is never read alone.
This is the sell-strategy matrix behind the number above (Table 2 of the methodology). Edit any threshold or sell percentage to test a rule of your own — every chart on this page recomputes live. The final tranche always liquidates whatever is left, regardless of its own row, so its threshold isn't editable.
| Tranche | Window | PP threshold | PP sell % | IPO threshold | IPO sell % |
|---|
02 Every company, not just the average
Seventeen historical placements is a small, non-random sample: MDB reports that essentially all of its private placements have reached an IPO, so there is no "hidden graveyard" of unlisted failures — but that says nothing about what happened after listing. Two of seventeen were total losses; two others (AREN, CUE) required real, cited reconstruction after we traced multiple ticker/name changes and a reverse split that had broken the original price feed — see Section 06 for exactly what was found and how each gap was filled.
Share of total realized profit contributed by each company. A strategy that "works on average" can still be a bet on one or two names.
Each placement joined the portfolio on its own real calendar date, not a shared start, so this walks the full 1996–2036 span as one blended position: for every dollar committed by that point, how many times over the portfolio was worth it — cash already realized from sales, plus the mark-to-market value of anything still held, divided by capital committed so far. Shown as a multiple rather than a dollar figure since the amount invested itself keeps growing as new placements join; a dollar value would make an early, small portfolio look misleadingly flat next to the ending balance. Starts at 1.00× by definition, the moment the first dollar was invested, and can swing sharply in the earliest years when only one or two placements had been funded. Pre-tax, this strategy's exit rules and haircut. The dashed segment after today is the model's scheduled completion, not yet-observed history — a few of the newest placements aren't due to fully resolve until the 2030s.
03 Every company's actual price path
The IRR bars above compress an entire multi-year price history into one number. Here is the underlying path for all 17 companies — value of $1 invested at the IPO price, on a shared logarithmic scale, so a 30× winner and a total loss can sit on the same page without one flattening the other.
04 Does the strategy add anything — or is this just being invested?
The gain thresholds in Table 2 were set with the benefit of already knowing what happened to these 17 stocks. That is a real risk of overfitting a rule to its own answer key. To bound it, every strategy is shown against a threshold-free baseline: buy at PP/IPO, do nothing, sell everything in year 10.
05 Compared to doing nothing clever at all
Each company's actual cash-flow dates and amounts are replayed against a named, publicly-sourced passive benchmark (selectable above) as a simplified public-market-equivalent check — not a substitute for a true index-level PME using the exact daily series, which MDB should run internally with licensed data before external use.
Bootstrap resample (2,000 draws) of the 15 companies with usable data. This describes how much the cross-sectional average could plausibly have differed by chance in this specific historical sample — it is not a forecast interval.
06 Full company-level ledger
Every one of the 17 placements, with status, realized outcome under the selected strategy, and any data-quality flag identified while rebuilding this tool.
| Company | Status | PP price | IPO price | Proceeds | Return (x) | IRR | Notes |
|---|
07 What this version changes
The prior methodology document already disclosed several limitations candidly. This build goes further: it changes the underlying calculation, not just the caveats.
08 Methodology & disclosures
How the simulation works
Each of the 17 companies receives an equal, hypothetical investment (default $1,000) in its Private Placement (PP) round, its IPO round, or both, matching what MDB Capital reports investors were actually offered in each deal. Shares are tracked separately by round because PP and IPO shares carry different cost bases, different gain thresholds under MDB's own strategy definitions, and different QSBS holding-period start dates.
Prices are sampled at fixed monitoring checkpoints relative to the IPO date — 30, 60 and 180 days, and 1, 2, 3, 5 and 10 years — rather than continuously. A tranche's exit condition is evaluated at each checkpoint inside its window, in order; the position sells at the first checkpoint where cumulative gain meets the strategy's threshold for that round. Unsold allocations carry forward and are swept into the next tranche, or fully liquidated in year 10, exactly as described in the original methodology document's Table 2 sell-strategy matrix.
Checking price at discrete calendar checkpoints, rather than assuming the investor could sell at the exact instant a threshold was crossed intraday, is a deliberately more conservative execution assumption than the original tool used.
Internal rate of return is solved per company (mixing PP and IPO cash flows) and at the portfolio level using a standard Newton–Raphson XIRR on actual cash-flow dates, matching the method described in the original document.
Sample composition & survivorship
MDB Capital represents that all of its historical private-placement "Big Idea" companies have gone on to complete an IPO or public listing — i.e., there is no denominator of unlisted, written-off private deals excluded from this sample. Taken at face value, this sample is the full population of MDB placements through 2024, not a survivor-selected subset of them, which addresses the specific pre-IPO selection-bias concern raised about the prior version of this report.
That does not mean every listed company remained a going concern or a good investment afterward. Of the 17: three were acquired for cash by strategic buyers (Pfizer/Medivation, Murata/Resonant, Sanofi/Provention Bio) at unambiguous, arm's-length prices; two (Lipid Sciences, Uni-Pixel) resulted in a total loss of capital; one (Hollis-Eden) is effectively worthless; several show severe share-price impairment despite remaining technically listed (Energous, after two reverse splits; eXoZymes; Second Sight/Vivani; and the entity now known as Paradium.AI, down roughly 90% from its original placement price across three successive corporate identities — theMaven, Arena Group, and Paradium.AI). This report treats "reached an IPO" and "produced a positive investor outcome" as two separate claims and does not conflate them.
This report cannot independently verify MDB's representation that the 17-company list is exhaustive; that representation is stated here as provided by MDB Capital and should be confirmed against the firm's deal records before external distribution.
Benchmark sources
Each benchmark in Section 05 is a real, published, publicly-sourced figure — not an arbitrarily chosen rate — applied as a flat annualized comparison rate over each company's own holding period:
- Russell 2000 Total Return Index — 10.7%/yr, trailing 20 years, per FTSE Russell factsheet data as reported in industry coverage (Feb. 2026). Broad U.S. small-cap benchmark; the closest standard market-cap-matched comparison for this portfolio.
- S&P 500 Total Return Index — 12.46%/yr, trailing 10 years, per published index performance data (2026). The standard broad-market baseline.
- Nasdaq Biotechnology Index (NBI) — 4.92%/yr, trailing 10 years, per published index performance data (2026). Sector-matched: roughly a third of this sample's 17 companies are biotech/pharma.
- Renaissance IPO ETF (IPO) — 8.35%/yr, since inception (Oct. 2013), per fund performance data. The most thematically precise match (a basket of recently-public companies) but only relevant to the more recent placements in this sample, since the fund did not exist before 2013.
These are trailing-period figures as publicly reported, not a bespoke calculation over this sample's exact 1996–2024 span, and different sources show measurable disagreement on some of these indices depending on the measurement window (we observed this directly for the Russell Microcap Index, which is why it is not included above). Treat each as a directional reference point, and confirm the precise figure against a licensed data terminal before external use. A "Custom rate" option remains available for sensitivity testing against any other assumption.
Taxes & QSBS
After-tax figures are an illustrative estimate only. They assume Section 1202 Qualified Small Business Stock treatment applies at the exclusion rate in effect on each round's issuance date (50% for stock issued before 2/18/2009; 75% between 2/18/2009 and 9/27/2010; 100% after 9/27/2010), subject to a five-year holding period, plus the selected state's long-term capital gains treatment and an assumed 23.8% federal rate (20% top long-term capital gains rate plus 3.8% net investment income tax) on any non-excluded gain. Actual QSBS eligibility depends on company-level facts (gross-asset limits, active-business requirements, and others) that were not independently verified security-by-security for this tool. Investors should consult their own tax advisor; nothing here is tax advice.
Limitations that remain
- N = 17, all with usable data. Three of those seventeen (AREN, CUE, EXOZ — marked "est." where applicable) rely in part on interpolated or proxy price points rather than a directly observed quote at every checkpoint; see Section 06 for exactly which figures are estimated. Every statistic here, including the bootstrap interval, describes this specific historical sample — it is not a forward-looking probability statement about future MDB deals.
- Checkpoint-based pricing is coarser than daily data; a threshold crossed and reversed between checkpoints would not be captured.
- IRR annualizes whatever holding period actually occurred, so a large price move realized within weeks of an IPO (e.g., CLIR) can generate a triple- or quadruple-digit IRR that is mechanically correct but overstates that position's economic significance. Per-company IRR is capped at 500% for display and averaging so one very-short-holding-period observation cannot dominate every downstream statistic; the (uncapped) total-return multiple and dollar-contribution chart in Section 02 are the more reliable read for those names.
- Equal $-per-company allocation does not reflect how any real investor actually sized these positions.
- No reinvestment of proceeds is modeled; realized cash is not assumed to compound into new deals.
- The benchmark comparison uses a flat assumed CAGR, not an actual daily small-cap or IPO index series.
- Past performance — MDB's or this sample's — is not a guarantee or reliable indicator of future results.
09 Full calculation audit
Every number above is computed from the strategy, investment amount, tax state, and haircut you’ve selected. This section shows the exact per-company calculation behind that selection — every checkpoint examined, which one triggered a sale, at what price, and the resulting cash flow — so the model can be checked line by line rather than taken on faith. Collapsed by default; click any company to expand it.