5-minute guide

How to judge a dip

When a planet passes in front of its star, the star dims slightly for a few hours. Our search finds every dip like that in years of data. Most are not planets. Your job is to spot the fakes. Computers are still bad at this; people are good at it.

Reading a candidate plot

Top strip: the star's brightness over every sector TESS observed (years), squeezed together. Red lines mark the dips the search found; orange lines are dips it already flagged as suspicious.

Middle panels: a zoom of about two days around each dip. Blue is the cleaned light curve and green is the raw one, shifted down. The shaded band is where the search thinks the dip is.

Bottom panels: where the star sits on the camera. If it jumps during the dip, the light probably comes from a neighbouring star.

ppm = parts per million. 10,000 ppm is a 1% dip, about what a Jupiter causes on a Sun-like star. Earth makes about 84 ppm.

Five real examples from our search

A real planet: TOI-4562 b

Verdict: Looks like a transit

A real planet: TOI-4562 b

  • Several dips of the same depth and shape, years apart (red lines on the top strip).
  • Each dip is flat-bottomed with sharp edges, the classic transit shape.
  • Blue (processed) and green (raw) agree, and the star position (bottom) stays steady.
Variable star: the "dip" is just part of the wobble

Verdict: Variable star

Variable star: the "dip" is just part of the wobble

  • The whole light curve swings up and down by thousands of ppm.
  • The flagged dip is no deeper than the normal ups and downs around it.
  • Spotted or pulsating stars fool the search all the time. Vote "Variable star".
Glitch: a dip at the edge of a data gap

Verdict: Glitch / data problem

Glitch: a dip at the edge of a data gap

  • Look at the middle zoom: the dip sits right where the data starts after a gap.
  • TESS drifts slightly as it restarts after each data download, making fake ramps.
  • Anything starting or ending at a gap edge deserves "Glitch" unless it repeats elsewhere.
Glitch: flux drops to zero

Verdict: Glitch / data problem

Glitch: flux drops to zero

  • A planet blocks a percent or two of the light, never 100%.
  • A drop to (almost) zero is a data problem: a dropout, a bad frame or a saturated pixel.
Hard call: a short, V-shaped blip

Verdict: Can't tell

Hard call: a short, V-shaped blip

  • Only an hour or so long and pointed, not flat-bottomed.
  • Could be a grazing binary, a cosmic-ray artefact or noise. Real long-period planets usually take several hours to cross.
  • Unless something else convinces you, "Can't tell" or "Glitch" is fine. Honest uncertainty is useful data.

Signs of a real transit

  • Flat or rounded bottom, clean edges
  • Several hours long (long-period planets move slowly)
  • The same in blue and green
  • The star is quiet before and after
  • If there are several dips, they match in depth and shape

Red flags

  • Touches a data gap or sector edge
  • A sudden step up or down, or a single-point spike
  • The star is wobbling all the time anyway
  • Very deep (over ~3%) or sharply V-shaped: often a binary star
  • Dips of two different depths, or the star position moves during the dip

How votes work

Every candidate needs at least three votes. When two thirds agree, the candidate takes that label. You can change your vote any time, and you can't vote on dips you found yourself.

You see other people's votes only after you vote, so nobody is nudged by the crowd. Candidates the community rates transit-like get a closer look from the project team (Gaia, ExoFOP, pixel-level checks), and the strongest are submitted to NASA's ExoFOP as community candidates, crediting the volunteers who found and vetted them.