The short of it

  • The SEC EDGAR database holds every required filing, with no estimation or scraping standing in for the original record
  • Four managers behind Traderbot's strategies reported a combined $444,876,568,456.00 across 4,197 positions in filings from 2026-08-13 and 2026-08-14
  • The two strategies built on form 13F outperformed the two built on faster-reporting congressional and form 4 filings
  • Form 4 gives an officer or director two working days to report a trade, versus 30 to 45 days for a congressional periodic transaction report

What a sec edgar database search is built on

A sec edgar database search draws on filings the regulator requires by law, not on anything scraped or estimated after the fact. Traderbot's own mirror holds 2,609 fund lines, 9,153 congressional lines and 335 officer lines, pulled from 13 managers and 143 members of the House, each line traceable to a single filing in that same database. Three separate kinds of filer feed the same underlying system, and none of the three lines is estimated or filled in where the record is silent.

The newest filing Traderbot has read was filed 2026-08-14, which sets the edge of how current a database-driven mirror can be at any given moment. Traderbot also stores 465,335 daily closing prices, the oldest from 2014-01-01, alongside the filing data, so a strategy built on a filing can still be measured against ordinary price history from the same stretch.

The edgar database in numbers

The edgar database itself is large enough that no person reads it filing by filing. Form 13F becomes mandatory within 45 days of the quarter end once a manager's US shares exceed $100 million, and four of the largest managers behind Traderbot's own institutional strategies, Berkshire Hathaway, Renaissance Technologies, Coatue Management and Bridgewater Associates, reported a combined $444,876,568,456.00 across 4,197 positions in their filings from 2026-08-13 and 2026-08-14 alone.

An edgar database search on any one of those names returns its own filing history, not the combined total. Adding the four together is arithmetic a reader has to do themselves, which is exactly what a raw database search leaves for the searcher.

Renaissance Technologies alone accounted for 3,116 of those 4,197 positions, spreading its $72,617,871,974.00 across far more individual holdings than the other three managers combined. A database search returns that count directly; explaining why the manager holds so many positions is a separate question the filing itself does not answer.

Running an edgar database search against a benchmark

Four of Traderbot's strategies are built directly on records an edgar database search would surface. All four launched with the same $10,000.00 stake on 2017-01-03, and all four are judged against the same plain S&P 500 fund, which returned 13.7% a year over the identical window.

Strategy Filing type Annual return Ending value
Institutional Consensus Form 13F 13.0% $32,381.38
Fund Mirror Form 13F 6.2% $17,820.18
Capitol Flow Periodic transaction report 2.0% $12,051.65
Insider Cluster Form 4 0.8% $10,771.27

The two strategies built on form 13F outperformed the two built on faster-reporting filings, which runs against the instinct that a quicker report should carry a stronger signal. Institutional Consensus alone carried a 46.9% deepest fall along the way, a clear reminder that a stronger average return did not come without its own rough stretch.

Reporting windows inside the database

Not every filing in the database moves at the same speed. The STOCK Act gives a member of Congress 30 to 45 days to report a trade on a periodic transaction report, while an officer, a director or a holder of more than 10 percent of a company files form 4 within two working days of a trade, a gap of weeks between the two kinds of disclosure.

A form 13F sits closer to the slower end of the two, filed within 45 days of the quarter end rather than of the trade itself, since it reports a full portfolio rather than a single transaction.

A database search cannot shorten any of these windows. The fastest signal it can surface is only as fast as the filing deadline the law sets, which is part of why Traderbot tests each rule against its own historical filing dates rather than assuming a same-day reaction was ever possible.

A worked example from the database

Nine years and roughly seven months separate 2017-01-03 from 2026-08-14. Compounding $10,000.00 at Fund Mirror's 6.2% a year across that same stretch, by hand, lands close to $17,600, within a couple hundred dollars of the strategy's actual $17,820.18 ending value.

Compounding the same $10,000.00 at Capitol Flow's 2.0% a year over the identical stretch gives roughly $11,900, close to its real $12,051.65, a small enough gap to trust both published rates.

Run the same arithmetic on the benchmark itself: $10,000.00 compounded at 13.7% a year across the same nine years and roughly seven months comes out close to $34,300, more than double either filing-based result above despite both strategies drawing on real, verifiable records rather than guesswork. A verifiable signal is not automatically a strong one, and the database search that produced these filings says nothing about which outcome a reader should expect in advance.

Where the raw data lives

The underlying records behind these strategies are public. The SEC quarterly form 13F data sets publish the same filings in bulk, structured for someone building a database search of their own rather than clicking through one filing at a time, and the individual form 4 itself lays out exactly what an officer or a large holder has to report within two working days.

Traderbot runs these four database-built strategies on the mirror page, where the balance is simulated rather than real and a one percent charge applies to every operation that moves money through it, whether a deposit, a withdrawal, a buy or a sell.

Questions

What does an EDGAR database search actually pull from?
It pulls from the same filings companies, managers and insiders are required by law to submit, indexed by filer and date. Traderbot's mirror reads that same database rather than any secondary or estimated source.
Which filing type produced the best backtested return?
Institutional Consensus, built on form 13F filings from institutional managers, returned 13.0% a year, the strongest of the four filing-based strategies tested. It still finished behind the 13.7% a year the S&P 500 benchmark returned.
How fast does a form 13F have to be filed?
A manager holding more than $100 million of US shares files form 13F within 45 days of the quarter end. That window is measured from the end of the quarter, not from any single trade inside it.
Why did faster-reporting filings not produce better results?
Capitol Flow and Insider Cluster, built on faster filings, returned only 2.0% and 0.8% a year despite shorter reporting windows. Speed of disclosure did not translate into a stronger trading signal in this test.
Is any of this built on real trading?
No. A Traderbot account holds simulated money, the service holds no client funds and sends no order to an exchange. It charges one percent of every operation that moves money on the account instead of trading live capital.

Sources