Fin360AIForm 5500 signals
Sign in

Questions

Frequently asked questions

Answered in full, including the ones that are awkward for us. If what you need is not here, email admin@infinidatum.net and a person will answer it.

What this is

The short version, for someone who landed here from a search result.

What does Fin360AI actually sell?
Ranked lists of U.S. retirement plans that look like they are about to make a decision — change recordkeeper, renegotiate what they pay for administration, or terminate. Each list ships with the out-of-time backtest that says how well the ranking works, including where it does not work.
What is Form 5500?
The annual return every ERISA-covered employee benefit plan is required to file with the U.S. Department of Labor. It is a mandatory filing, and the Department publishes the structured datasets as a public government record.
Why does a mandatory filing matter so much?
Because it gives the dataset a complete denominator. Every plan is in it, not only the ones that showed up in the trade press or answered a survey. A ranking is only meaningful against the whole population, and most commercial lists in this market are built on a sample nobody can characterise.
Is this an AI product?
No, and calling it one would be a smaller claim than the truth. The models are gradient-boosted classifiers over engineered features from the filings. What is hard here is not the modelling — it is resolving plan and provider identity across years so that a change of identifier does not masquerade as a change of behaviour. The method page explains why.
How often does it refresh?
Monthly. The Department folds amendments into the same download URL, so staleness is decided by the server rather than by our clock: each dataset-year is checked against its published size and last-modified date, and only what has actually been superseded is fetched.

The numbers

How to read the results, and what the four gates are for.

What is a 'lift'?
How much better than random the top decile of the ranking is. A lift of 3x on a 3% base rate means the top decile contains hits at roughly 9% — about one in eleven. Multiply base rate by lift and you get the hit rate you would actually experience working down the list, which is the only number that matters commercially.
What is the absence lift, and why do you keep calling it the moat?
It is the same lift, scored only on plan-years where nothing visible was happening — no provider change, no auditor change, a quiet asset mix. Anyone with the same public filings can rank plans that just moved. Ranking the quiet ones is the part that cannot be reproduced by reading the news, so it is the gate we care about most and the one that is hardest to pass.
Some of your signals fail their own gates. Why publish them?
Because the gates were fixed before anything was measured, and a signal that ships carrying whichever result it earned is worth more than one that ships carrying a threshold chosen afterwards to fit it. A buyer who discovers the failure themselves has no reason to trust anything else here.
What does 'n/d' mean next to a number?
Not determinable. The result straddles its own threshold depending on the random seed, so neither a pass nor a miss is a finding. It is reported as n/d rather than rounded into a verdict, because reporting an unstable result as a pass is the most tempting mistake available to us.
One signal is marked withheld. What happened?
The provider-consolidation signal backtested at an AUC of exactly 1.0000, which is a leak rather than a result: its outcome was defined using a quantity that was also one of its features, so the model recovered a threshold on its own label. It is withheld until it is rebuilt to predict next year from information available this year. The overview page carries the full note.
Is a backtest a forecast?
No. A backtest describes how the model scored a historical period it was not fitted on. It is not a promise that the next period will resemble the last one, and model performance degrades as the environment drifts. One of the signals here has already shown a material shift in its base rate between the fit and validation windows, which is stated on its own page.
How do I check any of this?
Every gate, every standing check and the pass/fail status of the current build are published on the method page, and the numbers on the marketing pages are generated from the same snapshot the product serves. If a figure here disagrees with the product, that is a defect and we want to hear about it.

The data

What is in the archive, what is not, and what the limits are.

Where does the data come from?
The Department of Labor's EBSA Form 5500 structured datasets, downloaded from the Department's own FOIA file service. It is a public government record. Nothing here is licensed from a competitor, scraped from another vendor's product, or subject to a third-party redistribution restriction.
How far back does it go?
Form years 2009 through the most recent published year, across the main return, the short form, and the Schedule C and Schedule H series. The data page lists the exact coverage on the current build.
Why is the newest year incomplete?
Filings arrive nine to eleven months after each plan's year end, so the most recent year is always partial. Rather than present a half-filed year as a shrinking market, every ranked list states what share of a full year has been filed. The last year for which outcomes can be resolved at all is earlier still.
Does this include plan participants' personal data?
No. Form 5500 is a plan-level and sponsor-level return. There are no individual participant records in it and none in this product. If you are a participant with a question about your own retirement account, your plan administrator is the only place that can answer it.
Do you use the Schedule of Assets?
No. Attachments, including the Schedule of Assets, sit behind a separate request and nothing here depends on them. Saying so matters because a competitor's holdings-level claim usually rests on that attachment, and this product does not make that claim.
What is the biggest thing that can go wrong in this data?
Identity. Anywhere an identifier can change, a change of identifier will masquerade as a change of behaviour: a plan renumbering looks like a termination, a provider changing its own tax ID looks like explosive growth, and a single acquisition looks like thousands of plans independently switching. For a product whose whole value is detecting change, the lineage resolver is closer to the core of the build than the modelling is.

Access, pricing and support

How you get in, what it costs, and what state that is in today.

How do I sign in?
Either an email address and a password, or Google. Neither is privileged: the same account, the same watchlist, the same persona. If you set a password we store a scrypt digest of it and never the password itself, so there is none here to lose; if you use Google there is no password at all. You pick a persona and, if you want, a firm domain; that is the whole account.
What does it cost?
Desk is $4,800 a year for one named user. Firm is banded by the size of the plan universe you scope — $9,600, $19,200 or $38,400 a year for the whole team. Market, which is built for acquirers, is not priced yet, because the firm-level signal it exists to serve is withheld and we will not charge for a feature we are holding back. Every tier is annual, and the pricing page shows what each includes and what it does not.
Can I try it against plans I already know?
That is the only honest way to evaluate this. Email us and we will walk through your own book or your own territory, so you can check the output against your judgment before spending anything.
Can I export the lists?
The product today gives you ranked lists in the browser, the movers between refreshes, per-plan history, and watchlists. Bulk export and API access are on the roadmap and are not something you can buy yet; the pricing page marks them as such rather than listing them as features.
Who answers when I email?
The person who built it. There is no ticketing system and no contact form. The contact page lists the subject lines that route a message to the right place first time.

Still deciding?

The fastest way to judge this product is to look at plans you already know. The method page has the gates and the standing checks on the current build; the disclaimer sets out the limits we would rather you read before you sign up than afterwards.

Sign in or create an accountAsk for a walkthrough