Admit-probability regression (per school, per round)
Trained on top-30 cycle outcomes, returns RD, ED, and EA probabilities calibrated to CDS.
§ Suite · Services
What Illuma actually does.
▸Live previewClick to collapse
College list
Your balanced list · CS
▸Reach5 schools
Reach
MIT
Cambridge, MA
Admit %
Overall 4%→You 11%
READue Nov 1
Reach
Stanford
Stanford, CA
Admit %
Overall 4%→You 9%
READue Nov 1
Reach
Carnegie Mellon
Pittsburgh, PA
Admit %
Overall 6%→You 14%
RDDue Jan 3
Reach
Harvard
Cambridge, MA
Admit %
Overall 3%→You 8%
READue Nov 1
Reach
Yale
New Haven, CT
Admit %
Overall 5%→You 10%
SCEADue Nov 1
▸Match4 schools
Match
Georgia Tech
Atlanta, GA
Admit %
Overall 17%→You 34%
EADue Oct 15
Match
UT Austin
Austin, TX
Admit %
Overall 12%→You 22%
RDDue Dec 1
Match
Michigan
Ann Arbor, MI
Admit %
Overall 18%→You 29%
EADue Nov 1
Match
UIUC CS
Urbana, IL
Admit %
Overall 21%→You 32%
RDDue Jan 5
▸Likely3 schools
Likely
Purdue
West Lafayette, IN
Admit %
Overall 42%→You 58%
EADue Nov 1
Likely
UMD
College Park, MD
Admit %
Overall 38%→You 52%
EADue Nov 1
Likely
Wisconsin
Madison, WI
Admit %
Overall 44%→You 61%
EADue Nov 1
Output includes
- Tier assignment using top-30 ML prediction
- Per-school fit + risk notes
- ED / EA strategy recommendation
- Autofill admit %, SAT policy, and supplements
Under the hood
ML models + calibration layers, not a ChatGPT wrapper.
Profile-score rubric
Initiative, EC impact, and spike domain scored 0 to 100, fed back into the regression.
Mechanical tier assignment
Reach, Match, and Safety derived from ML probability plus the school's 25th and 75th percentile bands.
Counselor-curated archetype guidance
Pool-pattern guidance matched by archetype and injected as context, not improvised.
ED-leverage scoring
Ratio of ED to RD probability flags ED candidates above 2x and ED-redundant below 1.5x.