Skip to content

One person, one method, every filing.

Everything that decides a SPAC's outcome is already public. Assembling it into something you can compare is the work — so I did that work, once, for the whole market.

Two cycles, one instrument.

How I trade
2021≈ 3× account
2022−70%
2023 – 2025150%+ a year
InstrumentsWarrants, rights
Common sharesNever

I am Sergey Monin, and I trade SPAC warrants. Not the common — the common share is a money-market fund with a ticker, and I have never seen the point of it. The warrant is the trade: leveraged, expiring, and worth either a multiple or nothing at all depending on whether a deal gets signed and closed.

In 2021 that worked spectacularly. Every SPAC found a target, the market paid up for every announcement, and I roughly tripled the account. In 2022 it worked in reverse. The enthusiasm drained out, the SPACapocalypse arrived, and I gave back about seventy per cent as the SPACs I held liquidated one after another and their warrants went to zero.

What bothered me afterwards was not the loss — it was that all of it had been disclosed. Trust coverage, extension terms, redemption history, the auditor’s language: every warning sat in a document filed months earlier, and none of the trackers I used surfaced any of it. They copied each other, went stale, and never showed where a number came from.

So I read the filings myself and watched what happened to companies after a merger actually closed. The ones that fell apart had said so in the terms they set at IPO. That became a classification — investment grade or speculative — and then a fair value model for the warrants. 2023, 2024 and 2025 were triple-digit years.

I build all of it: the pipeline, the maths, the design, the writing, the marketing. The first version was TypeScript and could not keep up with the filing volume, so I rewrote it in Rust for the memory and the throughput. I write about other work at built-in-saratov.com.

If the work has to be done once for every SPAC in the market, it may as well be done once for everybody.

What is covered

525SPACs trackedEvery outcome, since inception
98Live todayPre-IPO through deal announced
$22.8BCapital coveredGross IPO proceeds, all time
89Completed dealsFollowed through to closing
Market data as of Sep 6, 2026, 02:38 UTC

What we believe

Three commitments, each with a mechanism behind it

Primary sources only

Every figure originates in a document filed with the SEC. Nothing is scraped from aggregators, entered by hand, or inferred from a press release.

Each value carries the accession number it was read from.

Show the working

A number you cannot check is a number you should not act on. Where a figure came from is part of the figure.

Source filings are linked from the page, not buried in an export.

The clock is the asset

Most tools model a SPAC as a ticker with a price. It is closer to a bond with an expiry date and an option attached.

Deadlines, extensions and trust runway are first-class data.

How it is built

01IngestEvery new filing from SEC EDGAR, polled continuously.
02ExtractDocuments parsed to text and tables, including scanned exhibits.
03Parse XBRLTagged financial facts read straight from the filer’s own XBRL.
04NormaliseFigures reconciled across periods into one comparable schema.
05ClassifyRisk signals derived, and each SPAC placed on a risk band.
06PublishMarket aggregates recomputed and released every six hours.

And what it cannot tell you

Classification is a signal, not a verdict

Risk bands are derived automatically from disclosure. They summarise what a filing says; they do not predict what a SPAC will do, and they are not a recommendation.

XBRL tagging is uneven

Smaller filers tag inconsistently, and some figures exist only in the narrative. Where a value cannot be read reliably, nothing is shown rather than a guess.

Everything moves at filing speed

Data appears once it is filed. An event the market knows about before it reaches EDGAR arrives here late — usually by a day, occasionally longer.

Restatements happen

Companies revise prior figures. When they do, the record changes with them, which means a number can legitimately differ from one you saw last month.

Nobody pays to look better here

I take no money from sponsors, underwriters or the companies covered here. There is no sponsored placement, no paid removal of a risk signal, and no arrangement under which a SPAC can influence how it appears. Revenue comes from subscriptions, which is the only incentive structure compatible with the product being useful — and, having been on the losing side of the last cycle, the only one I would trust myself with.

I am not an investment adviser, a broker or a dealer, and nothing on this site is a recommendation. This reports what filings say and makes it comparable. What to do about it is your decision, and it should be made against the original documents.

Portrait to follow

Who is behind it

Sergey Monin

Investor, trader and programmer. I traded SPACs through 2021 and 2022, lost a large part of that position, and built this to answer the questions I could not answer at the time. The classification, the warrant model, the pipeline, the design and everything you are reading are mine.

The data side is Rust on AWS: filings ingested from EDGAR, parsed, reconciled across periods, classified, and republished every six hours. I write about that and other work at built-in-saratov.com.

Get in touch

How it got here

2021Traded SPAC warrants into deal announcements and roughly tripled the account. Every SPAC found a target that year, and the market paid up for all of them.
2022Gave back about seventy per cent. The SPACs I held liquidated one after another and their warrants expired worthless — all of it disclosed in advance, none of it flagged by anything I was reading.
After the crashStarted tracking what happened to companies after a merger actually closed, looking for what separated the ones that held up from the ones that did not.
The methodThe difference was visible in the terms at IPO. It became a classification — investment grade or speculative — and then a fair value model for the warrants themselves.
2023 – 2025Triple-digit years, trading the same instruments with the method underneath them instead of a thesis and a hope.
First versionBuilt in TypeScript. It worked, and it could not keep up with the volume of filings.
The rewriteRebuilt in Rust for memory and throughput. Reading every filing every day is an infrastructure problem before it is an analysis problem.
TodayFull lifecycle coverage across the whole market, refreshed every six hours.

Have a question about the method?

If something in the data looks wrong, we want to know — corrections against the source filing are the fastest way to make this better.