Planet candidate report · TIC 65043946.01

A 1.8 Earth-radius planet candidate around the red dwarf TIC 65043946, missed by NASA's own search

An anon from the trenches asked an AI to find a planet. It searched nearby stars in public TESS data and found one dimming by 0.16 % every 7.77 days, in every year TESS looked.

TRENCHPLANET ProjectData: NASA TESS via MASTStatus: candidate
Read the paper See the vetting Fly there
$TRENCHPLANETCA coming at launch
Hover the averages for values
Figure 1. TESS brightness of TIC 65043946. Folded: every measurement placed by its position in the 7.77-day orbit, with 30-minute averages and the fitted transit (orange). All data: the same measurements in time order, one block per sector. Switch to watch five years of data become one dip.
1 · SUMMARY

What was found

TESS photographs most of the sky in sectors of about 27 days. NASA's automatic planet search runs on stars recorded every two minutes and stitches their sectors together. TIC 65043946 got two-minute data only once, in 2019. Its other four sectors (2020, 2021, 2024 and 2025) exist only in TESS's full-frame images, so no search ever combined them. Combined, they hold transits of the same depth.

The dip is ± parts per million deep and lasts hours. On this star, a red dwarf 0.42 times the Sun's size, that is a planet about times Earth's radius, receiving 7 times the sunlight Earth gets. It is not on NASA's list of TESS candidates, and NASA's pipeline never flagged a signal on this star.

A candidate is not a confirmed planet. The standard false-positive test puts the chance it is something else at about . TESS observes the star again in December 2026, and the transit times were published before that data exists.

2 · VETTING

Trying to prove it wrong

Most signals like this turn out to be two stars eclipsing each other, a neighbouring star leaking light into the pixels, a spacecraft systematic, or noise. Each row below is a test for one of those. Thousands of other signals from the same search failed at least one of them.

TestResultA false positive would showVerdict
The transit in each TESS sector separately
Figure 2. The transit in each sector on its own, with the same model in every panel. A neighbour leaking light would change depth from sector to sector, because each sector uses a different camera position and aperture.
SectorYearPipelineDepth, corrected flux (ppm)Depth, raw flux (ppm)
Odd and even transits, and the light curve half an orbit later
Figure 3. Left: odd and even transits match. Right: nothing happens half an orbit later, where an eclipsing binary would show its second eclipse.
Difference image and the brightest neighbour's light curve
Figure 4. Left and middle: Sector 88 pixels, and where light is lost during transit. It falls on the target's pixel. Right: the brightest neighbour stays flat at this period.
Every TESS observation of the star
Figure 5. Every TESS observation of the star, with the transit times marked. The 2019 block is the only two-minute data; the rest come from full-frame images.
3 · PREDICTION

The December test, written before the data exists

TESS observes TIC 65043946 again in Sector 111. If the planet is real, the star dims at these times. The pass and fail rules were fixed in advance and are part of the published prediction, so nobody can move the goalposts, including us.

TransitMid-time (UTC)BJD (TDB)Uncertainty
Next predicted transit in
--
4 · THE PATH

The same steps professional planet hunters take

Approval is up to the TESS team. The path is ours, and every step is public.

    5 · FLY THERE

    144 light-years, in Canis Major

    Real sky positions and distances
    143.7 lyfrom the Sun to TIC 65043946, in Canis Major
    TRENCHPLANET mark: the transit dip as a trench with the planet in it

    $TRENCHPLANET

    A memecoin about a real planet search. It does not make the planet real, and December's data will not make the coin worth anything. It is a way to follow along, so size it like a meme.

    CAcoming at launch
    Is TRENCHPLANET a real planet?

    It is a planet candidate: a dip that repeats every 7.77 days and passes every test we could run on public data. The standard false-positive test puts the chance it is something else at about . TESS's December data is the real test.

    Did NASA approve it?

    We filed it with the TESS follow-up program and asked the TESS team for two-minute data in December. Their answer will be posted here word for word, yes or no. TRENCHPLANET is not affiliated with or endorsed by NASA, MIT or the TESS mission.

    Who found it?

    An anon from the trenches working with Claude Code, on public TESS data from NASA's MAST archive. The paper is signed TRENCHPLANET Project.

    Wasn't there another one like this?

    Yes. In October 2026 Pavel Rabtsevich posted TIC 4206066, also found with Claude Code. Our search found both of his signals on its own, and a confirmed Earth-sized planet nobody told it about. That is how we knew it worked before we trusted this one.

    Can I check it myself?

    Yes: the paper, the light curve, the code and the prediction are public.

    What if December says it is fake?

    Then we say so, here and on X. The rules were fixed before the data existed.