- What it does: estimates where an international applicant would have placed in the 2027-28 Early Session ranking, from an SAT or Bocconi test score and two school-year GPAs, and the chance of falling inside each program’s last-admitted rank.
- How it was built: fitted on 153 applicant-reported results, then tested on students the model had not seen. Its median rank error was 15%.
- What it is not: an official Bocconi tool, a guarantee, or a forecast of the Winter Session.
Bocconi publishes no cutoffs, admission rates or profiles of admitted students. After Early Session results came out on 11 November 2026, rejected applicants could see their own rank and, they report, the rank of the last student admitted to each program they chose. Enough of them shared those numbers along with their scores to fit a simple model. This page puts that model in your hands, together with an honest account of how often it is wrong.
Estimate your rank
For international applicants to Undergraduate School programs. Italian-school applicants and Law have separate rankings and are not covered.
Enter a test score and both GPAs to see your estimate.
The Winter Session: what this tool cannot tell you
Every cutoff here comes from the Early Session. Winter cutoffs are unknown: they depend on how many places are left after Early enrollments and on who applies in Winter, and Bocconi publishes neither. Bocconi says that about 90 to 95% of places are assigned across the Early and Winter Sessions together and that “early applications can have greater chances to be successful”. It does not say how places split between the two.
That is why the Winter switch above is a scenario, not a forecast. Lowering every cutoff by 10, 20 or 30% shows how much of your chance survives if Winter turns out tighter. If a program only works for you at 0%, plan as if it might not. For deadlines, test dates and how to decide whether reapplying is worth it, read our Winter Session plan after an Early rejection.
How the estimate works
Bocconi ranks applicants on two elements. Its admissions page lists the “Selection test score” at 55% and the “Third-last and second-last year GPA” at 45%, and adds:
Bocconi does not publish how it converts scores to points, and its SAT to Bocconi test conversion uses “comparative tables that analyze candidate percentiles” that are not public. So the estimator does not try to copy Bocconi’s formula. It learns, from applicants’ results, how much each input moved rank in the 2027-28 Early Session:
- One test scale. SAT scores are used as they are. Bocconi test scores are put on the same scale with a conversion fitted from the data. If you enter both, the better one counts, as on Bocconi’s portal, which “will pick the highest for you”.
- One GPA. Your third-last and second-last year GPAs are averaged.
- Log of rank. The model predicts the logarithm of your rank, with a curve for GPA. On a log scale, errors are proportional: being 15% off counts the same at rank 300 as at rank 3,000.
- Chance at a program. The probability that your rank falls at or inside that program’s last-admitted rank, given how far off the model is on students it has not seen.
Show the formula
T = (SAT - 1500) / 100 or T = (Bocconi test - 40) / 4.947 - 0.529
GPA
G = (third-last + second-last) / 2 - 9.5
Predicted log rank
y = 6.894 - 0.631 T - 1.433 G - 0.647 G²
Estimate, 80% range and chance
rank = ey range = ey ± 1.2816 × 0.255
chance = Φ((ln(last-admitted rank) - y) / 0.255)
The cutoffs. The last-admitted ranks are the Early Session figures reported by applicants from their official result screens between 11 and 14 November 2026: Economics 335, Finance 501, Mathematical and Computing Sciences for Artificial Intelligence 646, Management and Computer Science 1,094, Management 1,148, International Politics and Government 1,191, Management for the Arts and Cultures 1,649 and Economia aziendale 3,045. They are unverified, and Bocconi does not publish them. Places are then assigned in rank order. In Bocconi’s words, “points gained in the ranking give admission, not the order of program choices.”
How accurate is it? The data, the tests and the limits
The data
The model was fitted on a dataset of 2027-28 Early Session outcomes reported by applicants themselves. We do not link it or name anyone. After cleaning, 153 international results were usable.
An exact rank is used directly. An admission tells the model that the student’s rank was inside the admitted program’s last-admitted rank and outside those of any program they listed higher. A rejection tells it the rank was outside the last-admitted rank of every program they listed.
Left out: Italian-school applicants, whom Bocconi ranks separately; Law applicants, who have their own ranking; and ACT results, which were too few to fit. Read with care: the data is self-reported and self-selected. Applicants who report their results are not a random sample, and the reports are unverified.
How we tested it
Every accuracy figure below is leave-one-out: we refitted the model 153 times, each time leaving one student out, and predicted that student with a model that had never seen them. This matters because in-sample accuracy flatters any model, which always looks better on the data it was fitted to.
- Rank error. For the 41 students with an exact rank, half of the predictions were within 15% of the true rank; for a true rank of 1,200 that is within about 180 places. 80% were within 23%.
- Outcomes. For 119 of the 153 students (78%), the prediction fell on the same side of the relevant last-admitted rank as what actually happened.
- Spread. On the log scale, the residual standard deviation is 0.204 in-sample and 0.255 out-of-sample. The ranges and chances on this page use the larger, out-of-sample figure, so an 80% range runs from about 28% below to about 39% above the central estimate.
What the data shows
- 0.1 of average GPA is worth about 23 SAT points in this data, measured at an average GPA of 9.5. The GPA curve is steeper above that and flatter below it.
- A Bocconi test score of 40 lines up with about SAT 1447, and each Bocconi test point is worth about 20 SAT points. This comes from how applicants ranked, not from Bocconi’s tables, which are not published.
- Test scores bunch near the top. Bocconi weights the test at 55% and GPA at 45%, but half of the SAT scores in this data sit between 1460 and 1520. When most applicants’ test scores are that close, the test part differs little between them and GPA does much of the separating.
- Region: inconclusive. Applicants from Asia ranked about 20% further down than the model predicted for the same scores. But adding region to the model made its rank predictions worse on students it had not seen (median error 18.8% against 15.3% without it), which points to noise in a small sample rather than a stable effect, even though it matched slightly more admit or reject outcomes (122 against 119). So the model leaves region out. Bocconi publishes no country weighting; it says only that in a round with an extraordinarily high number of applications it will consider balancing offers by applicants’ profiles, backgrounds and nationality.
Limits
Requirements and next steps
This page estimates rank; it does not repeat the entry rules. For the SAT minimums (1040 overall, 520 per section), the no-superscore rule, score dates and Bocconi’s code, see Bocconi SAT requirements: SAT vs the Bocconi test. If you were not admitted in the Early Session, our Winter Session plan covers the 26 January 2027 deadline, the attempts you have left and when reapplying is worth it.
Frequently asked questions
How accurate is the Bocconi rank estimator?
Tested on students it had not seen, its median rank error was 15%, and 80% of its rank predictions were within 23% of the true rank. Its admit or reject call matched what happened for 119 of 153 applicants. These figures come from leave-one-out testing, where each student is predicted by a model fitted without them. It is less reliable at the extremes and for profiles the data barely covers.
Is this an official Bocconi tool?
No. PrepDrills is not affiliated with or endorsed by Bocconi University. Bocconi publishes no cutoffs, admission rates or last-admitted ranks. The estimator is fitted on outcomes reported by applicants after the 2027-28 Early Session, and its numbers are estimates, not guarantees.
Does the order of my program choices change my chances?
Not your chance of reaching a program. Bocconi says that points gained in the ranking give admission, not the order of program choices: applicants are taken in rank order and each gets the highest listed program that still has places. The order decides which program you get among those your rank reaches, so list programs in your true order of preference.
Why does GPA matter so much when the test is 55% of the score?
Because most applicants' test scores sit close together near the top. Half of the SAT scores in the data are between 1460 and 1520, so the test part differs little between applicants and GPA does much of the sorting. In this data, 0.1 of average GPA was worth about 23 SAT points around an average GPA of 9.5.
How does a Bocconi test score compare with the SAT here?
In this data, a Bocconi test score of 40 ranked about the same as an SAT of 1447, and each Bocconi test point was worth about 20 SAT points. Bocconi says the tests are considered equivalent and converts them with internal percentile tables it does not publish, so this is a fitted estimate, not Bocconi's table.
Can I use the estimator for the Winter Session?
Only as a what-if. All cutoffs on this page come from the 2027-28 Early Session. Winter cutoffs depend on the places left after Early enrollments and on the Winter applicant pool, and neither is published. The optional Winter switch lowers every cutoff by 0, 10, 20 or 30% so you can see how sensitive your odds are, but it is a scenario, not a forecast.
Does my country or region change the estimate?
No. Applicants from Asia ranked about 20% further down than the model predicted for the same scores, but adding region made rank predictions worse on unseen students (median error 18.8% against 15.3%). The evidence is inconclusive, so the model leaves region out. Bocconi publishes no country weighting.
I took both the SAT and the Bocconi test. Which one counts?
The better one. Bocconi's portal picks the highest of the results you upload, and the estimator does the same: it puts both on one scale and uses the higher. For the SAT, Bocconi takes your best single sitting and does not superscore.
Sources and method
Bocconi rules on this page were read on Bocconi’s own pages on 16 November 2026. The model and its tests were run by PrepDrills on 17 November 2026 on a dataset of 2027-28 Early Session outcomes reported by applicants; individual records are not published, to protect the students in it.
- Bocconi: Admissions 2027-28 (55% and 45% weights, best score counts, up to 4 choices, sessions, nationality balancing)
- Bocconi: Results and Enrollment 27/28 (assignment by rank, separate international ranking)
- Bocconi: SAT and ACT (tests considered equivalent, percentile-based conversion, no superscore)
- Bocconi: GPA and high school curriculum (years counted, grades excluded)
- PrepDrills: Bocconi Winter Session plan (last-admitted ranks reported by applicants, Winter dates)