Methodology
Benchmarking Methodology
SyndTrack places your vintage cohorts against illustrative sample quartile ranges. This page says exactly where those numbers come from, and what the product does not compare you against.
Where the quartile ranges come from
The quartile ranges SyndTrack shows are an illustrative sample: representative figures written for this application. They are not licensed index data, they are not attributable to any index provider, and they are not derived from other investors' portfolios. The newest vintage in the set is 2022.
Treat a quartile placement as a rough orientation point, not as a measured ranking against the market and not as investment advice. If you need a licensed benchmark to allocate against, use one.
Cross-account comparison is not built
SyndTrack does not aggregate portfolios across accounts, and no figure anywhere in the product is computed from another investor's data. Nothing described in the rest of this page has run.
We publish the design anyway, in the conditional, because the privacy thresholds below are a commitment we want on the record before anyone is asked to consent to anything. Read the sections that follow as a specification, not as a description of a running system.
Opt-in Consent
Any cross-account aggregation would be strictly opt-in. Your portfolio data is never included in any aggregation unless you explicitly choose to participate. You will be shown a clear explanation of what data is contributed and how it is anonymized before you consent.
Withdrawal flags your data as excluded, and the next aggregation run would not include your portfolio. No historical removal is required because any benchmark would be recomputed from scratch on each cache refresh. Note the current gap: there is no self-serve withdrawal control in the app yet. That is one of the reasons cross-account benchmarking is not switched on, and it will ship before any aggregation runs.
Consent is versioned. If the scope of data collection changes materially, we will ask for your consent again under the new version.
Privacy Protection: k-Anonymity & Minimum Thresholds
To prevent individual deal leakage, two thresholds are enforced in code before any cross-account segment could be displayed. They are also the reason nothing is displayed today: no cohort comes close to either number.
- Minimum 10 distinct investors in the cohort (k-anonymity parameter k = 10).
- Minimum 50 data points (individual deals) across all contributing investors in the segment.
When either threshold is not met the statistics for that segment are withheld and the screen says so instead of showing a number. This is what makes it impossible to reconstruct an individual portfolio from an aggregate, and it is enforced server-side rather than by hiding a value in the browser.
Aggregation Approach
For each segment, a per-investor summary statistic (e.g. average IRR) would be computed first, and percentiles and means taken across those per-investor summaries. The two-step order matters: it stops investors with large portfolios from dominating a statistic relative to investors with fewer deals.
Aggregated statistics are written only to the pre-computed cache table, which holds no user IDs, deal IDs, or any other personally-identifiable information. That table is empty.
Planned Segments
None of the following is computed or shown today. Each is written in the conditional deliberately.
Portfolio IRR by vintage year
Would take each participating investor's average deal IRR within a vintage-year cohort, then compute the 25th percentile, median, and 75th percentile across those per-investor averages. Grouping by vintage year controls for market-cycle effects.
Inputs used
- Deal IRR (annualized)
- Vintage year of first investment
Privacy note
IRR would be averaged per investor before any aggregation, so no individual deal IRR is surfaced.
Distribution yield by asset class
Would calculate distribution yield (DPI — distributions paid / called capital) per investor per asset class, then report the median and interquartile range across investors for each asset class (e.g. multifamily, industrial, private credit).
Inputs used
- Total distributions received
- Total called / invested capital
- Asset class of each deal
Privacy note
Yield would be aggregated at the investor level before any cross-investor statistic is computed.
Sponsor concentration against a recommended threshold
Would compute the fraction of called capital each participating investor has allocated to their single largest sponsor, and report the spread of that ratio. The 25% recommended threshold is a general guideline, not a measured figure: allocating more than a quarter of a portfolio to one operator is widely treated as a concentration risk.
Inputs used
- Called capital per deal
- Sponsor identifier per deal
Privacy note
Only the top-sponsor ratio, a single number per investor, would be used. No sponsor names are shared.
Asset-class diversification against the cohort average
Would report, for each asset class, the mean allocation share across participating investors — what fraction of portfolio capital the typical participant holds in that asset class.
Inputs used
- Called capital per deal
- Asset class per deal
Privacy note
Per-investor allocation percentages would be aggregated across investors. No individual allocation breakdown is exposed.
Data Freshness
The illustrative quartile ranges are static reference figures. They carry an as-of date, and the newest vintage in the set is 2022. They do not refresh, because there is no upstream feed to refresh them from.
Cross-account segments would be pre-computed and cached for up to 1 hour, and withdrawal invalidates the cache immediately so the next request recomputes without that portfolio.
What We Never Share
- Individual deal names, values, or returns
- Sponsor names or identifiers in any aggregate output
- User identities or profile information
- Any segment with fewer than 10 investors or 50 data points
- Raw data — only percentiles and means would ever be reported
- Your portfolio with any sponsor, and your portfolio with any other account, which is the case today because no cross-account aggregation runs at all
Questions?
If you have questions about where the benchmark figures come from or how your data is used, please contact us via our help page.