DEV Community

videocompress
videocompress

Posted on Edited on

Guess or Skip? Akuna 60 in 8 vs Optiver 80 in 8: a Random Pick Lost 1.56 Points per Question, a Skip Lost 2

Originally published on DEV.to — tested with QuantSprint.

Two trading-firm style math tests, same eight minutes on the clock. One gives you 60 typed questions and charges nothing for a miss. The other gives you 80 multiple-choice questions and takes a point back for every miss. The question I kept getting asked was simple: when a fraction shows up and you're stuck, do you guess, or do you skip?

So on Friday, November 20, 2026, I ran both simulators on QuantSprint seven times and changed one thing per run: what happens on fraction questions.

Disclosure: I build QuantSprint. A Playwright script in headless Chrome typed and clicked the answers, so the run times (26–37 seconds) are machine times, not human ones. The point here is the scoring, not the speed. Both pages are free, with no account needed. QuantSprint isn't affiliated with Akuna Capital or Optiver; these are practice presets modeled on the public formats.

The two formats, side by side

Akuna-style OA Optiver 80 in 8
Questions 60 80
Clock 8 min (8.0 s per question) 8 min (6.0 s per question)
Answer style typed 4 options (keys 1–4 or A S D F)
Wrong answer 0 −1
Skip button Skip (0) Skip (−1)

Here is what the Akuna Capital math test simulator looks like before you start. Note the "Wrong answers score 0" line:

Akuna-style arena ready screen: 08:00, 60 questions, wrong answers score 0

And the Optiver 80 in 8 practice test, same clock, but with a −1:

Optiver 80 in 8 arena ready screen: 08:00, 80 questions, wrong answers −1

How I set up the seven runs

Each run started fresh and drew its own question set. The script solved everything exactly with Python fractions. Then a "policy" took over on any question that contained a fraction (like 5/6 − 5/9):

  1. Careful: answer every question correctly (one run per test).
  2. Wrong on purpose: type or click a wrong answer on every fraction.
  3. Skip: press Skip on every fraction.
  4. Random (Optiver only): pick one of the four options at random on every fraction, seeded so it can be repeated.

One honest caveat: my solver couldn't handle the a ÷ ? = b shape, so it pressed Skip on four of those across all runs. I counted them in the tables, and they don't change the conclusions.

Step 1: what a fraction looks like on each test

On the Akuna-style test you type the answer. There is no list to eliminate from, and Skip is labeled (0):

Akuna-style typed question 0.2 ÷ 7/15 with a text box and Skip (0)

On Optiver you get four options, often near neighbors (5/18 vs 5/19), and Skip is labeled (−1):

Optiver question 5/6 − 5/9 with options 2/9, 1/3, 5/18, 5/19 and Skip (−1)

That label is the whole story. On this simulator, skipping costs exactly what a wrong answer costs.

Step 2: the Akuna-style runs (miss = 0)

Run Fractions in set Correct Wrong Skips Score
Careful 7 60 0 0 60
Wrong on purpose 7 52 7 1 (solver) 52
Skip every fraction 20 40 0 20 40

Score equals correct answers, every time. A wrong entry and a skip both cost the point you didn't earn, nothing more. So on a no-penalty test, never leave a box empty. Type your best estimate and move on, because an estimate can only add points. The page's own prep notes say the same thing: every reachable item deserves an entry, but don't linger.

Step 3: the Optiver runs (miss = −1, skip = −1)

Run Fractions in set Correct Wrong Skips Score
Careful 9 79 0 1 (solver) 78
Wrong on purpose 15 64 14 2 (solver) 48
Skip every fraction 15 65 0 15 50
Random pick on fractions 9 73 7 0 66

The arithmetic is score = correct − wrong − skips: 65 − 15 = 50, 64 − 16 = 48, 73 − 7 = 66.

Here is a deliberate wrong pick. The script chose 5/13 for 1/3 + 1/12 (the answer is 5/12, and you can see the near-neighbor trap):

Optiver question 1/3 + 1/12 with options 1/3, 5/12, 5/13, 0.5

And here is one of the random picks that got lucky: 1/3 − 0.25, where the random option was 1/12, which is correct. Two of the nine random picks landed:

Optiver question 1/3 − 0.25 with options 1/13, 1/6, 0, 1/12

The four end screens put it plainly: 60 vs 40 on Akuna, 50 vs 66 on Optiver.

Four QuantSprint result screens: Akuna careful 60, Akuna skip 40, Optiver skip 50, Optiver random 66

Step 4: normalize per question, because the sets differ

Each run drew its own set (7 to 20 fraction questions), so raw scores aren't directly comparable. Points lost per fraction question, against answering it correctly, are:

Bar chart: points lost per fraction question. Akuna wrong 1.00, Akuna skip 1.00, Optiver wrong 2.00, Optiver skip 2.00, Optiver random measured 1.56, expected 1.50

  • Optiver skip: 2.00 points lost (the +1 you missed plus the −1).
  • Optiver random 1-of-4 pick: 1.56 measured (14 points over 9 questions). The expected value is 1.50: a 25% chance of +1 and a 75% chance of −1 is −0.5, against +1 for a correct answer.
  • Akuna, wrong or skip: 1.00 either way.

What I'd actually do on test day

  1. No-penalty, typed (the Akuna-style preset): always type something. A rounded estimate costs you nothing extra. Spend your seconds on speed, not on deciding whether to answer.
  2. Optiver-style with Skip (−1): a skip is the most expensive move in the game. Even a blind 1-of-4 guess beats it on average (−0.5 vs −1). Cut one obviously wrong option and it gets better: 1 of 3 gives −0.33, and 1 of 2 gives 0.
  3. The real fix is to stop meeting fractions cold. Most of the near-neighbor traps (5/12 vs 5/13, 1/12 vs 1/13) go away if you know the common fraction-decimal pairs by heart. I keep a fraction-to-decimal table for that, plus an unreduced fractions drill and a missing-operand drill for the a ÷ ? = b shape that even tripped my script.
  4. Know your pass line. If you want a target number for the −1 format, I wrote up what score counts as good on 80 in 8.

Reproduce it

Open both simulators in two tabs, do one honest run each, and write down the fraction questions you hesitated on. Then do a second run where you force yourself to answer every one of them. Compare the two scores, not your accuracy. On a −1 test, that comparison is the one that tells you whether your skip habit is costing you.

If you're studying something other than trading math, MorsePass is my free ham-radio exam and Morse trainer, built the same way.

Related: the first version of this firm-format comparison (Maven 50 in 5 vs Flow Traders 75 in 8) is on Hashnode, and the Optiver 80 in 8 timed run is too.

— mengyuxuan

Top comments (0)