Methodology
Plain-English explanation of how CET Compass produces every probability band you see.
What we do
We predict your CAP admission probability for every (college, branch, category) combination using the last 3 years of cutoff data.
The math, in plain English
The same formula every row in your predict result uses.
Probability formula
p = sigmoid((your_percentile - cutoff_percentile) / 1.5)
Then we apply a 3-year trend nudge: -0.05 if cutoffs are rising, +0.05 if falling.
Bands
- Safe — p ≥ 0.85
- Target — 0.40 ≤ p < 0.85
- Reach — p < 0.40
Buffer
buffer = student_pct - cutoff_pct - 0.5 * stddev_3yr
A small or negative buffer means the option is volatile across years even if the headline cutoff looks fine.
The data
We use 9 official MHT CET cutoff PDFs (2022/2023/2024 × Round 1/2/3).
- 119,101 cutoff rows
- 355 colleges
- 27,468 latest-cutoff entries
Source: cetcell.mahacet.org
Backtest accuracy
Backtest pending. Backtest results will appear here after the Sprint A backend backtest script runs. Last computed: pending.
What we don't do
- No sponsored ranking
- No lead-gen
- No fake urgency
- No hidden defaults