Methodology, data and analysis by Logan Wong.
Long-horizon, rules-based models for the Australian share market: market size versus the economy, earnings multiples, interest rates, trend deviation and the equity-bond trade-off. Each is scored in standard deviations from its own history, and the five scored models combine into a single aggregate reading. Built on 45+ years of monthly data.
The aggregate reading is +0.42σ from the average of the five scored models, down from +0.48σ at the previous update as the market fell further. The individual model readings are set out below and in the table.
On the data, the S&P/ASX 200 closed at 8,665 on Friday 25 September, about 6% below the all-time high of 9,203 set in February 2026, having slipped back into negative territory for the calendar year the day before. The S&P/ASX 300 closed at 8,600, which is within a tenth of a point of where it sat in April. Because the P/E and dividend yield readings are scaled from the last published April figures by the index move, both have returned almost exactly to those published values. The trailing P/E is 22.9x, which is 44% above its 15.9x long-run average and higher than 93% of monthly readings since 1980 (z +1.20). The dividend yield is 3.09%, about 24% below its 4.08% long-run average and lower than all but 7% of monthly readings (z +1.20).
Pulling the other way, the Southern Cross Indicator (market capitalisation to GDP) is about 120%, roughly 19 points below its long-run linear trend (z −0.81); the ratio has trended structurally higher across four decades of superannuation inflows, so the model measures deviation from trend rather than the raw level. Total ASX market capitalisation has fallen to about A$3.58 trillion against annualised nominal GDP of about A$2.97 trillion. Mean reversion places the S&P/ASX 300 about 5% above its long-run price trend (z +0.36), well inside one standard deviation, and the earnings yield gap is a little above its long-run average (z +0.13).
Rates backdrop, as published by the data sources: the RBA cash rate is 4.35%, with the next decision due on 29 September 2026, and the 10-year government bond yield closed at 5.37% on 25 September, the highest monthly reading in this series since 2011. The rise in long-dated yields is the main reason the interest rate model has moved further above its own long-run average.
These are model outputs calculated from published data, not a prediction. This is factual information only, not financial product advice, a recommendation or a forecast.
Every model on one line: today's reading, the model's own fair value, how far the two differ, and where today sits in the model's own history. Select any row to jump to that model.
| Model | Current | Fair value | Deviation | History percentile | z-score |
|---|
How to read this. "Fair value" is each model's own long-run trend or average. "Deviation" is how far today's reading sits from it. "History percentile" shows where today's reading sits within the model's own history. "z-score" expresses that deviation in standard deviations. All figures are calculated, not estimated.
Total ASX market capitalisation as a percentage of nominal GDP, annually 1979–2020 (World Bank) with a live 2026 reading (ASX market cap A$3.58T ÷ annualised GDP A$2.97T ≈ 120%). Scored against a linear trend fitted to the full history — the ratio drifted structurally higher through the 1980s–90s as superannuation deepened Australia's capital markets, so the model measures deviation from trend rather than the raw level.
Market-cap-weighted trailing P/E of the Australian market (All Ordinaries), monthly since 1980. Scored on LN(P/E) versus its long-run average, which damps the distortion of recession-era earnings collapses (1992, 2009, 2020–21 spikes are earnings artefacts, not euphoric pricing).
Australian 10-year government bond yield, monthly since 1980 (OECD via FRED). Scored against the modern inflation-targeting era (1995–present): a positive z-score means the current yield is above its 1995-present average. This series is shown for context and is not included in the aggregate.
S&P/ASX 300 monthly closes since 2001 (the index launched April 2000), log scale, versus an exponential trend fitted on
continuously compounded returns — r = LN(Pₐ/Pₐ₋₁) — with ±1σ/±2σ bands. Price index (excludes dividends);
an accumulation-index version is in development.
Equity earnings yield (100 ÷ P/E) minus the 10-year bond yield, monthly since 1980. A positive gap means the equity earnings yield exceeds the 10-year bond yield; a negative gap means the bond yield exceeds the equity earnings yield. The z-score is calculated inversely, so a lower gap gives a higher z-score.
Market-cap-weighted dividend yield of the Australian market, monthly since 1980 (excluding franking credits). Yield moves inversely to price. The z-score is calculated inversely, so a lower yield gives a higher z-score.
Calendar-year total returns (price + dividends reinvested) of the All Ordinaries Accumulation Index, 1980–2024. Most years are positive, the average is double-digit, and the worst years cluster around 1981–82, 1990, 2008 and 2022.
A backward-looking check on the mean-reversion model. For every month since 2001 it measures how far the index sat from its long-run trend, then the price change that actually followed over the next 12 months. The chart plots every month; the table groups the same history by starting valuation. It reports what has happened, not what will happen.
| Starting valuation band | Months | Avg next-12m price change | Share positive |
|---|
Read with care. This is the price index only and excludes dividends, so total returns were higher. The record covers 2001–2026, one market's history, and the bands furthest from trend hold few observations. It is an objective summary of the past, not a forecast, and past performance is not a reliable indicator of future performance. Factual information only, not advice.
Objective by design. Nothing here is a personal opinion. Each reading is produced by a fixed formula applied to published market data, so the same inputs always give the same output. This page is factual information only. It is not financial product advice, not a recommendation to buy, hold or sell anything, and it does not consider anyone's circumstances.
Scoring. Every model is reduced to a z-score: how many standard deviations today's reading sits from its own long-run trend or average. A larger positive z-score means the reading sits further above the model's own long-run average or trend; a larger negative z-score means further below. The aggregate is the simple average of the five scored models' z-scores (Southern Cross, P/E, Mean Reversion, Earnings Yield Gap, Dividend Yield). Interest rates and the subsequent-returns section are context and are not included in the aggregate.
Continuously compounded returns. Trend fitting uses natural-log returns LN(Pₐ/Pₐ₋₁), which are additive across time and
the correct basis for exponential trend regression.
Sources. Market Index (PE, dividend yield, market statistics) · FRED/OECD (10-year yields) · FRED/World Bank (market cap ÷ GDP) · FRED/IMF (nominal GDP) · TradingView (S&P/ASX 300 monthly closes) · All Ordinaries Accumulation data per Vanguard and ASX research. Data retrieved .
Freshness. Index prices and bond yields are current to the latest monthly update; P/E and dividend yield after April 2026 are price-scaled nowcasts pending publication. The aggregate is refreshed monthly.
This page presents factual, objectively calculated market information; every figure is produced by a fixed rule from published market data. It is not financial product advice (general or personal), not a recommendation, not an opinion on any security, and not a forecast of future returns, and it does not consider anyone's objectives, financial situation or needs. Past performance is not a reliable indicator of future performance. Boston Global Wealth Pty Ltd ABN 99 691 504 506 is a Corporate Authorised Representative (No. 1318023) of Advisory Circle Pty Ltd AFSL No. 513052. © 2026 Boston Global Wealth Pty Ltd.
