Original mock contests
Three-problem sets in USACO style, one per division. Each set is calibrated to the typical difficulty range of that division. Self-grade with the rubric on each problem page.
How to take a mock. Set a 4-hour timer (5 hours for Platinum). Close every tab except the
problem page and your editor. No editorials, no AI assistant, no chat. When the timer ends, score yourself
using the rubric, then read the solution sketch.
🥉 Mock Bronze · "Springtime"
Three problems, 4 hours total. Focus: complete search, simulation, ad-hoc.
| # | Title | Skills | Target time |
|---|---|---|---|
| B1 | Flower Sorting | Simulation, sorting | 30–60 min |
| B2 | Cow Bingo | Complete search over bitmasks | 60–90 min |
| B3 | Fence Painter | Ad-hoc + case analysis | 60–90 min |
🥈 Mock Silver · "Pasture Tour"
Three problems, 4 hours total. Focus: BFS/DFS, binary search, prefix sums.
| # | Title | Skills | Target time |
|---|---|---|---|
| S1 | Lost Sheep Count | BFS on a grid | 30–60 min |
| S2 | Hay Bale Stacker | Binary search on the answer | 60–90 min |
| S3 | Sunrise Schedule | 2-pointer + prefix sums | 60–90 min |
Detail pages coming as you reach Silver. Skeletons land first; full statements get added during August in the study plan.
🥇 Mock Gold · "Migration"
Three problems, 4 hours total. Focus: DP, Dijkstra, segment / Fenwick tree.
| # | Title | Skills | Target time |
|---|---|---|---|
| G1 | Highway Toll | Dijkstra on layered graph | 60–90 min |
| G2 | Pasture Permutations | 1-D DP with combinatorics | 60–90 min |
| G3 | Friendship Index | Fenwick tree + offline queries | 60–90 min |
Self-grading rubric
USACO scoring is per-test-case out of 1000. For mocks (where you don't have a judge), grade each problem out of 1000 like this:
| Outcome | Score |
|---|---|
| Full correct solution, fits constraints, you tested it on a hand-crafted edge case | 1000 |
| Correct algorithm, works on all samples, you ran it mentally on one edge case | 800 |
| Brute force only, correct on small n, would TLE at full constraint | 400 |
| Wrote something but didn't finish / known WA on some cases | 100–200 |
| Didn't attempt or didn't compile | 0 |
Total possible: 3000. Target for promotion-equivalent: ≥ 750.
After the mock
- Don't open the editorial yet. For each problem, spend 15 minutes writing what you tried, what worked, what didn't, and the moment you got stuck.
- Now read the editorial. For each unsolved problem, identify the one piece of knowledge or trick you were missing. Write it down.
- Re-implement any problem you didn't solve, without looking at editorial code.
- Add the technique you missed to next week's drill schedule.