How I Ace Midterms at a Top CS School by Studying 1-3 Hours (and Skipping Class)

I have a midterm tomorrow (actually)[0] and instead of studying I decided to write this piece on exactly how I spend as little time as possible on school while achieving decent results (top ~5-10% on tests).

This protocol is heavily tailored to me and my background specifically. I do think some people can get a few ideas from this though.

tl;dr:

Step 0: Optimize health and mental state (sleep, nutrition, meditation).
Step 1: Use Claude to quickly analyze lecture slides and practice tests.
Step 2: Memorize slides by creating interconnected mental stories.
Step 3: Brain dump notes, write code, and practice active recall.
Step 4: Take practice test, rapidly iterate on weak areas.
Step 5: Identify easy wins—predict and memorize likely test content.
Step 6: Final review, active recall, and Claude-powered reinforcement right before the exam.

background

I’m a decent test-taker and code a lot, so most concepts aren’t new. I rarely attend lectures and do the bare minimum homework required to pass. My broad curiosity gives me wide, shallow knowledge. Interestingly, I enjoy taking tests, especially multiple-choice.

studying process

The zeroth thing I do is the night before the exam I will get a perfect sleep and the morning of I will minimize the amount of brain power I am spending until the exam. I will eat properly and cleanly and meditate (which I generally try to do daily anyways). This is maybe the biggest power law in terms of my score on exams. Around 2-4 hours immediately before the exam I will start the process.

The first thing I will do is collect all of the lecture slides + practice tests and upload them to a Claude project and ask it what’s on the exam. I’ll ask for the types of questions, a structured overview of the materials, and what I have to know/memorize. What’s especially important at this stage is getting a feel for the types of questions the test likes to ask while not revealing any specific questions on the practice test. This should take no longer than 15 minutes. If I feel like I already know 100% everything on the test I will skip to taking the practice test.

I will then read over all the lecture slides once and commit them to memory. I’ll frame it like a story, X works with Y to do Z and Z is used to do A, and you need A to do B, and so on. I’ll also relate them to concepts I already know. Just build a mental map. After reading the slides for one lecture I’ll see if I can recall them before moving on to the next.

After I memorize all the lecture slides I’ll brain dump everything I know into written notes. I’ll try to be as accurate and all encompassing as possible. I will also try to write as much code as possible while taking these notes to get familiar with writing code on paper with a pencil. I’ll then review all the notes and see what I missed. Generally I miss a few details for each concept. I will rewrite whatever I missed and on particularly tricky concepts make a mental note to recall them later on in my studying/reviewing (active recall).

I then will take the practice test. I will go over my test taking strategy in the next section. Generally I just take the test as fast as possible and answer every question in this stage. I’ll review and rewrite all my questions that I got wrong. Referring only to my written notes.

After this I go for easy wins, trying to figure out exactly what’s on the test. I’ll give an example. If you know only topics from the lectures + homework are covered, and that the practice tests and real tests are similar you can infer and just memorize stuff that will probably be on the test. If there is a practice question about implementing an algorithm, and you only covered 2 algorithms in your homework in that specific section, chances are you can just memorize the other algorithms and be fine. Some test questions are so specific that all permutations of them are obvious and you can just memorize the trick or entire solution to them.

To keep my mind warm I’ll ask claude for questions simmilar to the ones on the practice test and try and recall the lecture slides (particularly the tricky parts) while I am walking to take the test. If there are like complex/technical things I’ll make Claude explain them to me again and review my physical notes.

The best part about the process is that you can basically start at any step depending on how confident (or lazy) you are. Sometimes I just talk to claude frantically while walking. You can also do this days before the exam and multiple times to ensure you know everything 100%.[1]

Generally what I notice is that if you actually do and 100% know the homework and go to lecture you just don’t have to study (shocker).

test taking strategy

The way I take tests is I skim the whole test front to back first. Then if I am very lost I’ll start on the easiest questions (I often do this when I haven’t studied at all), but generally I just go front to back and circle harder questions to come back to later. I am trying to go as fast as possible and rely on my intuition as much as possible. The asterisk is that if I attempt a problem (write something down) I will see it through until I get an answer or am at a complete blocking point (like I just don’t know something and I can clearly articulate what piece of knowledge I am missing). I will never actively stop my train of thought because I am taking too long (my CS tests generally give loads of time, so I don’t have to worry about this). If the question is multiple choice I will eliminate every other solution instead of just finding the one that’s correct. Once I am done I’ll go back to the hard, circled questions. I am generally completely in flow and I’ll slow down my rush and cadence to my test taking once I get to the circled questions.

Once I have answered every question I’ll go back and review all the multiple choice ones, just go back through my chain of thought and verify I didn’t do anything stupid. I’ll often try to solve the problem in different ways to verify also. The questions where I write code I’ll briefly go over how I thought about it but I mainly look for syntax errors. I don’t continue checking my work until the time runs out mainly because I just don’t find that to be the most valuable use of my time. But on the SAT and other tests I did do this, and this will maximize your score. When you are checking your answer, you should have a clear and defensible chain of thought. If you are not sure don’t change your answer just go with your initial instinct. Only change if you made a silly mistake or there was a clear and distinct error in your previous chain of thought.

After the test I rarely have any clue how I did because I was just in flow state. It’s like I just blacked out and took the test. I immediately forget everything that is on the test too. If I am asked a question about it the next day I won’t even remember what you are talking about and I’ll probably search it up.

conclusion

This is my process, my mind works in very mysterious ways. The main reason I am efficient with my time is that I have the (learned) ability to focus for many hours at a time and there is not “waiting” time or “relaxing” time when I am studying. I don’t take breaks and my mind is always firing on all cylinders, 100% occupied.

I have a decent memory but the pieces of information need to be highly interconnected. My memory is mostly a result of being able to find these connections, not just raw memorization. I also have never not studied like this and this is the only way I study currently.[2] I never did much meta creating this workflow this is just a descriptions of what happens when I need to take tests. (although I had prior knowledge of “active recall” and memory stuff because I find that stuff interesting)

There are risks to this approach, as if you bite off more than you can chew you have no time to self-correct. If you do this short/intense thing a lot you can burn out. I don’t know how well this would work on humanities tests. I just don’t care enough about school to spend anymore time than what I give.

Hopefully this doesn’t come off like I am bragging. Although my test scores are decent my GPA is very subpar and I am far from the picturesque college student. I am involed in almost no university affiliated activities. I have never scored the highest in the class. I don’t reccomend my exact approach, though I think implementing the elements into something more sustainable/personal is valuable.

why? (postscript)

Why would anyone not choose to attend the school they pay for? Why would they try and game the system?

For me, this is because I feel like, particularly in computer science, I have all the necessary skills and knowledge to do what I want. I want to build stuff, make stuff. I spend most of my time trying to build and make stuff. Somehow I thought in high school that my CS curriculum would be things I couldn’t learn online or was significantly different from what I already know. It turns out this was not the case.

Going to school for me is a hedge. If I can’t or don’t build something of significance, it’ll be easier for me to get a job with a degree. As soon as I build something that has clear evidence of significance I will stop doing school. The other option I have is trying to drop out and try and find work now, but I don’t know if I really want to work for someone else or if I could even find a job. It would most likely be at a startup, which is a huge career risk. Although I would 100% drop out if I could work on an interesting problem with smart people.

[0]

[1] It’s also scalable. I did this process when I was in algebra 2 (or maybe precalc?) to self-study for the AP Calc BC exam. I think the extent of my calc knowledge was limits + derivative is slope of the tangent line + integral is area under the curve. I pulled an all-nighter the day before the exam and just took it (this was apparently “fun” to me at this point in my life). I of course, only got a 3, which is a passing score but not one that helped me. Actually I used that score to take Calc 3 + linear algebra at community college while in HS but in college, even though I had Calc 3 credit I had to somehow make up Calc 1 and Calc 2 credit.

[2] I can’t do study groups. I actively bring down the productivity of study groups. I also notice I can’t code (work) when I am around my friends. Like even them just being in the corner of my vision prevents me from locking in. I am too drawn to social interaction. Different story if we are working together though.