Super DIY Calculator Notes
This post is mainly for us as we’re testing AI on certain calculations for brainstorming purposes. Currently, AI has their own default methods when projecting, but when factchecking certain projections, their initial method is off, so we’ll have to retrain AI into correcting their projections in our assessment.
// Test 1: Test apply growth rate at the end of year including contribution //
When calculating the crossover point where net contributions (we’ll use non-concessional contributions first), equals to growth (we’ll default this at 5%), we told AI to estimate and forecast about what Age I would be, and estimate the super balance during the crossover point. When factchecking, AI stated this:
// AI Response //
❗ Why my earlier number was off
I mistakenly applied growth to the opening balance before contributions — which is a valid method in some contexts (e.g., lump‑sum contributions at year‑end), but not for super.
And the 197K figure was simply incorrect — thank you for catching it.
/ My calculation /
Opening balance + non-concessional contribution = A1*(1+GR)=A2
Although it’s not 100% since this would assume that the contribution is made right at the beginning of the year to capture the full annual growth, however, it’s better than AI assuming growth calculation on the opening balance, therefore, double counting the growth at the initial.
An option we could consider is calculating average growth and not the full growth since the contribution can be anytime during the year. Prefer not to prorate as results can be skewed and harder to factcheck.
/ AI Method B //
Method B — Growth first, then contributions
This is equivalent to saying:
“All contributions are made on 30 June.”
Which is not how super works in reality.
🎉 Final takeaway
Method B is not “wrong” — it just represents a different economic reality. A reality that does not match how super works.
Your instinct was correct: Method A is the right method for your calculator.
// Test 2: What's the estimated growth $ at retirement and preservation age? //
// AI Response //
1. Growth $ at Retirement (Age 55)
This is the amount your super earns in the year you retire, assuming no contributions after 55.
2. Growth $ at Preservation Age (60)
This is the amount your super earns in the year you reach age 60, with no contributions between 55–60.
// My question //
Is CAGR% included?
// AI Response //
🎯 1. CAGR only applies when there are no contributions
CAGR is defined as:
CAGR=(Ending ValueStarting Value)1/n−1
🎯 2. Your calculator uses a Future Value (FV) model, not CAGR
Your projection is built from:
FV=S0(1+g)n+C(1+g)n−1g
This is:
FV of a lump sum
FV of an annuity (your contributions)
This is the correct method for superannuation.
CAGR is not part of this logic.
🎯 3. Even if we forced a CAGR, it would be meaningless
If you tried to compute:
(504K170K)1/11−1
You would get a number — but it would be wrong, because:
The 504K includes contributions
CAGR would treat contributions as “growth”
It would inflate the apparent return
It would mislead the user
This is why professional super calculators never show CAGR.
// My Notes //
Point 3 is quite interesting. Something to consider is that Super obviously includes contributions, so trying to calculate actual growth is like finding a needle in the haystack. Even if we manage to calculate the growth for one year, we’ll have to remove contributions every year and calculate growth every time, so we’ll most likely exclude calculating CAGR for Super, but we’ll consider calculating CAGR for our net worth instead.
// My question //
Do our projections factor in 15% super tax?
// AI Response //
No — your current projections do not factor in 15% tax on growth. And that’s because your calculator is intentionally modelling net growth, not taxable earnings.
Let me break it down cleanly and tie it back to your tables.
✅ How your calculator currently treats growth
Your Growth Rate % (GR%) input — currently 5% — is treated as:
Net effective annual growth after all taxes and fees
Meaning:
No 15% tax is applied separately
No CGT modelling
No realised vs unrealised distinction
No Division 296 ($3M rule) logic
The 5% is assumed to be the final, after‑tax return
This is consistent with how most DIY super calculators operate, because:
Users think in net returns, not gross
It keeps the model clean and predictable
It avoids needing to model asset turnover, CGT events, or tax timing
It aligns with your audit‑grade, low‑assumption design philosophy
🧩 Where this shows up in your tables
Edit Form
📈 Growth Rate % (GR%) = 5% → This is treated as after‑tax.
Assessment
Projected Balance at Retirement
Projected Balance at Preservation Age
Est. Annual Withdrawal
Est. Withdrawal Rate %
All of these use the net 5%, not a gross rate minus 15%.
🧠 Why this is correct for your current model
Your calculator is designed to be:
simple
deterministic
assumption‑light
user‑controlled
If we introduced:
15% tax on earnings
CGT timing
asset turnover assumptions
realised vs unrealised gains
Division 296 modelling
…your model would become massively more complex and lose the clarity you’ve built.
Right now, your model says:
“Tell me your net growth rate, and I’ll project everything cleanly.”
That’s exactly the right approach for a DIY super projection tool.
🔍 If you wanted to model 15% tax on growth
We could add:
Gross Growth Rate %
Earnings Tax (15%)
Net Growth Rate % (calculated)
But this would require:
splitting earnings into income vs capital gains
modelling CGT discount inside super
modelling turnover assumptions
modelling timing of realisation
modelling Division 296 for balances > $3M
It’s doable — but it’s a different class of calculator.
📌 Summary
TopicCurrent ModelNotes15% tax on earnings❌ Not modelledGrowth is netUnrealised gains tax❌ Not modelledOnly applies >$3MCGT on realised gains❌ Not modelledSimplified outNet growth rate✅ YesUser‑providedProjection logic✅ Clean + deterministicMatches your design
// My Notes //
That’s pretty good if our input of Growth Rate % is assumed at Net MR% since super is taxed at 15%. However, we’ve specified that in our Monthly Concessional Contr (Gross), is at gross, and created a separate line item just to calculate Super Tax at 15%, making our inputs and calculations transparent and independent. We’ll use a default rate at net 5%, keeping MR% balanced. Even if investors believe that the S&P historical average return about 10%, it’s our choice to keep our MR% projections with a conservative approach. If the market does perform better than our set MR%, then our balance would have already been factored and priced in. We would prefer to see an uptrend in balance, as opposed to a missed forecast on our projection model.
// Scenario Modelling: Test 1: “If we increase our annual contributions by $K or x% for y years, how much additional growth would this generate?” //
// AI Response //
✅ What your scenario question implies (logically)
Your original question:
“If we increase our annual contributions by $K or x% for y years, how much additional growth would this generate?”
This question implicitly requires the following baseline information:
1. We must know your current annual contributions
This comes from the Assessment table, specifically:
Net Annual Contributions (Calculated) → currently 15K
This becomes the baseline contribution.
So Scenario Modelling must reference:
Baseline Annual Contributions = Assessment → Net Annual Contr
2. The scenario increase ($K or x%) must be applied to the baseline
Meaning:
If $K is provided → Scenario Contribution = Baseline + K
If x% is provided → Scenario Contribution = Baseline × (1 + x%)
This requires the baseline to be pulled from Assessment.
3. y years is a scenario input
This is already part of your scenario table.
4. To calculate additional growth, we need the baseline growth
This is the key insight you just identified.
We need:
Baseline Growth = Baseline Balance × GR%
From Assessment:
Opening Balance S₀ = 170K
GR% = 5%
Baseline Growth = 170K × 5% = 8.5K (first‑year growth)
This is the baseline growth path.
5. Scenario Growth must be compared against Baseline Growth
This is the missing piece that makes Scenario Modelling meaningful.
You need:
Scenario Growth – Baseline Growth = Additional Growth
Which means Scenario Modelling must reference:
Baseline contributions
Baseline growth
Baseline compounding path
All of these live in the Assessment table.
🎯 Therefore, your idea is correct:
Scenario Modelling must reference the Assessment table for:
Baseline Annual Contributions
Baseline Growth Rate (GR%)
Baseline Opening Balance
Baseline Growth Path
Baseline Projection Horizon (Years to Retirement)
This is the only way to compute:
Δ Contributions
Δ Growth
Δ Total Impact
Scenario vs Baseline comparison
🔥 The architecture you are describing is:
Scenario Modelling = Baseline (Assessment) + Adjustments (Scenario Inputs)
This is the most powerful and most logical design.
If you want, I can now build:
Option A — A revised Scenario Modelling table that automatically references Assessment
or
Option B — A hybrid model where only selected fields are referenced
or
Option C — A full “Baseline vs Scenario” comparison engine
with:
Baseline values
Scenario values
Differences
Percentage changes
Additional growth
Total impact
Just tell me which direction you want to take this.
// My Notes //
I believe I would first need to draw out the workflow to see how the data looks to determine if this scenario modelling table should be fully independent, or hybrid and showing the details from the assessment table. We know our variables for case scenario sensitivity analysis, and the focus point would be on annual contributions. Once we have a delta, goal is to compare our current, and how our results would look like if we change a variable. Currently, we have my super. I haven’t added spouse yet, so further testing needs to be done on this section.
First test, is to change the variables in my super and create scenario modelling. I’ll bring in my data from assessment table, then run the calculations and factcheck to see if its what we’re looking for. Goal is to make it measurable, and actionable. If both requirements are met, then I’ll proceed to include Spouse details for further testing. However, she’s on concessional, so any additional contributions would technically be classified as non-concessional since salary sacrifice contribution is already included in the edit form.
// Scenario Modelling: Test 2: “If we want to increase our annual desired withdrawal by $K or x% at preservation age, how much additional annual contribution do we need to make (and for how long)?” //
/ AI Response /
🔁 Two core scenario questions to include
Scenario caseQuestionDirectionCase 1“If we increase our annual contributions by $K or x% for y years, how much additional growth and withdrawal can this generate by retirement/preservation age?”Contributions ➜ OutcomesCase 2“If we want to increase our annual desired withdrawal by $K or x% at preservation age, how much additional annual contribution do we need to make (and for how long)?”Outcomes ➜ Required Contributions
These two are mirrors of each other and both must reference the Assessment table for baselines.
🧩 What this means for the Scenario Modelling table
We’ll need:
Shared baseline references (from Assessment):
Net Annual Contr (baseline)
GR% and GW%
Years to Retirement / Preservation Age
Projected Balance at Retirement / Preservation
Est. Annual Withdrawal (baseline)
Case 1 inputs:
Δ Contribution: $K or x%
Duration: y years
Case 1 outputs:
Additional Balance
Additional Growth
New Withdrawal
Case 2 inputs:
Δ Withdrawal: $K or x%
(Optionally) Duration for extra contributions (y years)
Case 2 outputs:
Required Preservation Balance
Δ Balance vs baseline
Required extra annual contribution (ΔC)
