DSA TrainerHow to know which LeetCode pattern to use
You understand each pattern on its own. Then a fresh problem loads and you go blank. Which one is this? That gap is the whole game. The skill isn't knowing what a sliding window is; it's noticing the problem in front of you is a sliding window problem. This page maps the signals: read the clue words in the problem, find the row, click through to learn the pattern.
Arrays & strings
| Pattern | Clue words in the problem | Reach for it when… |
|---|---|---|
| Hash Map | sums to a targethave I seen this valuecount occurrencescomplementlook it up by valuegroup by a shared key | You need to find the complement or pair of a current value |
| Set | duplicateall uniqueseen beforealready visiteddoes it existdistinct values | You need to detect duplicates in an array or string |
| Frequency Map | how many times eachanagramrearrange the letterssame charactersappears most oftencount of each character | You need to check if two collections have the same elements regardless of order |
| Two Pointers | sorted arrayfind a pair that sums tocompare both endspalindromemove inwardin place, no extra space | You need to compare elements at two ends of an array or string |
| Sliding Window | contiguous subarraysubstringlongest or shortestat most kwithout repeatingwindow of size k | You're looking at a subarray or substring of the input |
| Prefix Sum | sum of a rangesubarray sums to kcumulative / running totalsum between i and jnumber of subarrays with sum | You need the sum of elements between two indexes, potentially many times |
| Kadane's Algorithm | maximum sum subarraycontiguouslargest sumrunning totalmust pick at least one | You need the maximum (or minimum) sum of a contiguous subarray |
| Tracking Minimum | maximum profitlowest seen so farbuy low, sell highlargest gap where later beats earliersingle pass | You need the best value seen before the current position |
| Intervals | intervalsoverlapping rangesmerge themstart and end timesscheduling conflictnon-overlapping | You're working with [start, end] pairs that may overlap |
| Cyclic Sort | numbers 1 to nvalues in a fixed rangefind the missing numberfind the duplicatesmallest missing positivewithout extra space | The input contains numbers in the range 1 to N where N is the array length |
| Sorting Algorithms | sort the inputascending or descending orderarrange bykth largest after sortingrank / orderrearrange | The problem becomes much simpler if the input is in order |
Stacks & queues
| Pattern | Clue words in the problem | Reach for it when… |
|---|---|---|
| Stack | matching parenthesesbalanced bracketsvalidnestedmost recently seeninnermost first | You need to match opening and closing pairs like brackets or tags |
| Monotonic Stack | next greater elementnext smaller elementnearest warmer day to the rightspan until a bigger valueprevious smaller element | You need the next element that is larger (or smaller) than the current one |
| Monotonic Queue | maximum in every windowminimum in every windowsliding window of size kbest in the current windowdouble-ended queue | You need the maximum or minimum of a sliding window of fixed size k |
Searching
| Pattern | Clue words in the problem | Reach for it when… |
|---|---|---|
| Binary Search | sorted inputfind target in O(log n)first or last positionrotated sorted arrayminimum value that workspeak element | The input is sorted (or the answer space is ordered monotonically) |
| Divide and Conquer | split in halfsolve each halfcombine the resultsmergekth element by partitioningcount across halves | The problem can be split into independent subproblems of the same type |
Linked lists
| Pattern | Clue words in the problem | Reach for it when… |
|---|---|---|
| Linked List | linked listnode with a next pointerreverse the listmerge two listsdetect a cyclefind the middle | You're working with nodes connected by next pointers |
| Fast & Slow Pointers | cycleloopcircularfind the middle in one passwhere the cycle startswithout extra space | You need to detect a cycle in a linked list or number sequence |
Trees & graphs
| Pattern | Clue words in the problem | Reach for it when… |
|---|---|---|
| Tree Traversal | binary treeroot and childrendepth or heightlevel by levelinorder / preorder / postordersubtree | You need to visit every node in a binary tree |
| Graph Traversal | nodes and edgesconnectedis there a pathnumber of islands / regionsshortest pathvisit neighbors | You're exploring connected regions in a grid or graph |
| Matrix Traversal | grid2D matrixadjacent cells4-directionalregions in a gridrows and columns | The input is a 2D grid and you need to explore connected regions |
| Topological Sort | prerequisitesmust come beforevalid order to finishdependency orderingdirected acyclic graphbuild / compile order | Tasks have prerequisites, some must be completed before others can start |
| Union Find | connected componentsare these in the same groupnumber of groupsmerge two groupsredundant edgedynamic connectivity | You need to group elements into connected components dynamically |
| Trie | prefixstarts withautocompletesearch words in a dictionarylongest common prefixword search | You need to search for strings by prefix, autocomplete, word suggestions |
Recursion, backtracking & DP
| Pattern | Clue words in the problem | Reach for it when… |
|---|---|---|
| Recursion | defined in terms of itselfsubproblemexplore every branchnested structurenth termbreak into smaller pieces | The problem naturally breaks into a smaller version of itself |
| Memoization | overlapping subproblemsthe same call repeatscount the waysminimum or maximumnth valuecache the results | A recursive solution is correct but too slow because it recomputes the same inputs |
| Dynamic Programming | how many waysminimum or maximum to reachcan you reach the targetcount the pathseach step is a choiceoverlapping subproblems | The brute-force recursion recomputes the same subproblem multiple times |
| Backtracking | all possibleevery combinationevery permutationgenerate allfind all validsatisfies the constraints | You need to generate all subsets, combinations, or permutations of an input |
| Greedy | minimum number ofmaximum you cancan you reach the endfewest stepspick the best at each stepafter sorting | Each step has a clear locally optimal choice that doesn't undermine future steps |
Heaps & merging
| Pattern | Clue words in the problem | Reach for it when… |
|---|---|---|
| Heap / Priority Queue | k largest or smallestkth largesttop krunning medianappears most frequentlyfrom a stream | You need the k largest or k smallest elements from a collection |
| Two Heaps | running medianbalance two halvesfrom a stream of numbersmax of the low half, min of the high halfschedule by two priorities | You need the median of a dynamically growing stream of numbers |
| K-way Merge | k sorted listsk sorted arraysmerge sorted inputskth smallest acrosssorted matrixcombine sorted streams | You have K sorted lists or arrays and need to merge them into one sorted output |
Bit manipulation
| Pattern | Clue words in the problem | Reach for it when… |
|---|---|---|
| Bit Manipulation | appears once while others repeatno extra memorypower of twocount the set bitsXORbinary representation | You need to find a unique element in a sea of duplicates |
Stuck between two that feel similar?
The hardest part is telling apart patterns that look alike. These break down the exact difference:
Train the recognition, not just the patterns
150guided problems. Five are free, no account needed. Each one walks you from “what is this asking?” to the pattern to the code, with hints that nudge, not spoil.