GratisAIlearning · field guide · local-first · no account
guessing threshold matrix

Check Expected Marks Before Using a Guessing Strategy — Free

Negative marking changes the arithmetic of an uncertain answer. Compare an estimated chance of being correct with a stated penalty while remembering that confidence is not a calibrated probability.

The proof surface

Grade-weight targets work backwards from course totals. This computes expected value for an attempt-versus-blank decision under one explicit marking rule.

Inputquestion group | estimated correct probability percent | questions to attempt; wrong-answer penalty when a correct answer earns one mark
Rare deviceGrade-weight targets work backwards from course totals. This computes expected value for an attempt-versus-blank decision under one explicit marking rule.
Output artifactexpected marks per row with transparent workings
Cost$0 · local-first · one free run · samples labelled
Sample, not your facts: Illustrative sample: Confident | 80 | 10; Unsure | 20 | 5. With wrong-answer penalty when a correct answer earns one mark set to 0.25, total expected marks is 7.50 expected marks.

Why the flat version breaks

Confusing percent with odds

Enter 20 for twenty percent, not 0.2 or one-in-five text.

Ignoring a different blank penalty

The model assumes a blank earns exactly zero.

Treating expectation as a score promise

A small set of actual answers can finish far above or below its expectation.

How to work the guessing threshold matrix

Read the actual marking rule

This model gives one mark for correct, zero for blank and the entered penalty for wrong. Partial credit, multiple selections and adaptive exams need a different model.

Estimate confidence cautiously

Enter whole question counts and a probability between 0 and 100. Use past calibration if available; an optimistic feeling is not evidence of an 80 percent success rate.

Calculate expected marks

For probability p, expected marks per question are p minus (1 − p) × penalty. Multiply by question count. At penalty 0.25, a 20 percent chance breaks even with leaving blank.

Compare strategies, not guarantees

Negative rows flag a lower expectation than a blank. Expected marks describe repeated similar decisions, not the result of your next question; follow exam rules and manage time separately.

What the guessing threshold matrix replaces

QuestionBeforeVisible working
Confusing percent with oddsEnter 20 for twenty percent, not 0.2 or one-in-five text.Estimate confidence cautiously
Ignoring a different blank penaltyThe model assumes a blank earns exactly zero.Calculate expected marks
Treating expectation as a score promiseA small set of actual answers can finish far above or below its expectation.Compare strategies, not guarantees

A local arithmetic aid replaces hand calculation, not expert review. No paid AI service is needed.

Run it on the samples, right here

FIRST-LOAD

Research prototype — checkout unavailable. Input is processed locally and a draft is saved automatically in this browser profile when storage is available. Reset to sample clears that draft; Pro history has its own clear button. A state link encodes your inputs in its URL: share only non-sensitive rows. Browser history, clipboard and anyone receiving the link may retain it. Optional AI use below leaves this device; it is not required.

Educational probability arithmetic only. No exam outcome is predicted, and the tool cannot estimate your true knowledge or choose answers for you.

Data note: Everything runs in this browser tab on the guessing threshold matrix: your question group | estimated correct probability percent | questions to attempt stays on this device, nothing is uploaded, and the reading is rebuilt only when you press run.

Go deeper: the companion app files the same reading as a scored comparison matrix

The article demo above runs without limits. The companion app keeps a local history, exports the rows as CSV, prints the guessing threshold matrix reading, and holds your drafts on this device — one complete free run; the proposed $4 one-time filing layer is not currently for sale.

The guessing threshold matrix reading is complete for free. The optional $4 layer adds print, row CSV and the last five local reading summaries; it does not add hidden answers. Checkout is not configured yet; the article demo remains unlimited.

Open the check expected marks before using a guessing strategy companion

Boundary

Educational probability arithmetic only. No exam outcome is predicted, and the tool cannot estimate your true knowledge or choose answers for you.

What this is built on

Before: Enter 20 for twenty percent, not 0.2 or one-in-five text. After: the guessing threshold matrix shows the working beside each named row so the reader can change the assumption and inspect the consequence.

Optional AI formatting, not calculation

For this guessing threshold matrix, use the free AI Studio interface only if available to you, with no paid API key. Manual entry completes the same workflow for free. Supply only fictional or non-sensitive notes. Review its output against the source; never paste unresolved questions into the numeric rows.

Format my non-sensitive notes for a guessing threshold matrix. Return plain rows only: question group | estimated correct probability percent | questions to attempt. Preserve supplied quantities exactly. Do not guess missing values; list questions separately. Do not calculate or add advice. I will check every row before pasting into the local tool.

Accepted schema: question group | estimated correct probability percent | questions to attempt. No AI response is executed as code.