Imagine two venues with the same number of poker machines.

Would you expect those machines to generate about the same amount of player losses?

Data published by the Victorian Gambling and Casino Control Commission (VGCCC, 2026) let us test that assumption. Follow five interactive views: start with machine counts, then compare venue types, hold location and size roughly constant, and finally check whether the difference persists through time.

How to explore: Hover over marks for details. Use chart legends to toggle groups; drag to zoom where supported and double-click to reset. Dollar figures are Australian dollars.

1. More machines, more losses?

START WITH THE OBVIOUS EXPLANATION

Explore the evidence: Select 2023–24, 2024–25 or 2025–26 using the dropdown above the plot. Then hover over individual venues to compare machines and losses. Bigger venues often record larger total losses, but venues with similar machine counts do not necessarily report the same losses.

Chart 1 · Individual hotels and clubs, selected financial year. Annual losses and reported machine counts. Source: VGCCC (2026), venue-level expenditure dataset.

What to notice: Venues with comparable machine counts can sit far apart on the losses axis.

2. What if the venue type matters?

THE FIRST DIVIDE

To compare venues more fairly, divide each venue’s reported annual losses by its machine count. The result is an approximate annual loss per machine. Hotels and clubs do not have identical distributions—even within metropolitan and country Victoria.

Chart 2 · Losses per machine by venue type and region, 2025–26. Source: VGCCC (2026), venue-level expenditure dataset.

What to notice: Compare the diamonds (group medians) as well as the spread of individual venues.

3. What happens in the same area?

THE CENTRAL TEST

Perhaps hotels simply operate in different places. Here we compare median hotel and club losses per machine within the same local government area (LGA). Only LGAs with at least two qualifying hotels and two qualifying clubs are included.

Before looking: do you think hotels lead in fewer than half, about half, or nearly all comparable LGAs? Click to reveal the count.

THE BIG REVEAL

33 out of 37

Comparable LGAs recorded higher hotel medians per machine. The remaining LGAs matter too: this is a widespread pattern, not a universal rule.

Chart 3 · Within-LGA comparisons of venue-type medians, 2025–26. LGAs are geographic groups, not matched individual venues. Source: VGCCC (2026), venue-level expenditure dataset.

What to notice: Most connecting lines place the hotel marker to the right of the club marker; a few LGAs break the pattern.

4. Could it just be venue size?

CHALLENGE THE EXPLANATION AGAIN

This time, compare hotels and clubs within similar machine-count bands, separately for metro and country areas. The colour scale represents median losses per machine, not the number of machines. Hover to see how many venues contribute to each cell.

Chart 4 · Venue size is grouped into 20–39, 40–59, 60–79 and 80+ machines. One common colour scale is used for both panels. Country hotel cells with 60–79 and 80+ machines contain only five and six venues, respectively; treat these medians cautiously. Source: VGCCC (2026), venue-level expenditure dataset.

What to notice: Compare hotels and clubs within each column, not just the darkest cell in each region.

5. Was this just one unusual year?

THE LONGER VIEW

Across three financial years, the difference persists at the regional level. Compare the two clearly marked series and hover over their points to inspect the annual values.

● Hotel — solid line ▲ Club – – dashed line

Chart 5 · Median annual losses per machine, 2023–24 through 2025–26; nominal dollars, not adjusted for inflation. Source: VGCCC (2026), venue-level expenditure dataset.

What to notice: The two venue types remain separated across all three years in both regions.

The discovery

Machine numbers alone cannot account for the differences in reported losses. Hotels record higher median losses per machine than clubs in 33 of 37 comparable LGAs in 2025–26, across all eight region-by-size groups shown, and at the regional level in all three observed years.

What we cannot say: These records do not establish why the gap exists. Customer behaviour, operating hours, machine characteristics, exact location and other venue features may contribute. The data describe venues; they do not show that a particular kind of person gambles more.

The house’s secret? The difference is visible in public data; the reasons behind it remain unresolved.

Method and limitations

Reproducible preparation check

The table below is calculated directly from the imported worksheets, not typed in manually. “Other rows” include totals, footnotes or records outside the selected Hotel/Club categories; “quality exclusions” fail at least one machine-count or operating-expenditure check.

Import and filtering audit (calculated in R)
Year Rows read Other rows Hotel/club rows Quality exclusions Final venues
2023-24 500 4 496 34 462
2024-25 520 35 485 19 466
2025-26 499 15 484 23 461
  • Source and scope: Victorian Gambling and Casino Control Commission (VGCCC), Current Gaming Expenditure by Venue; hotel and club EGMs only. Casino gaming is outside this analysis.
  • Import: Three financial-year detail worksheets were read from the downloaded official Excel workbook with readxl and combined in R; values were not hard-coded.
  • Exclusions: The analysis retains venue records categorised Hotel or Club with at least 20 machines and positive expenditure in both halves of the financial year. Excluded records remain available in venues_all. These filters favour comparisons of active full-year venues but may omit some legitimate venues.
  • Measure: Annual venue expenditure divided by the source’s June average EGM count, which is a proxy rather than an exact annual-average machine count.
  • Aggregation: The charts generally report venue-level values or medians, not player-level outcomes. Comparing LGA medians does not control for all other differences between venues.
  • Time and prices: 2023–24 to 2025–26 in nominal AUD; changes should not be interpreted as inflation-adjusted growth.
  • Interpretation: The observed associations do not prove venue type causes higher losses. Differences in sample size, especially within some size bands, limit generalisation.

Data source and references

All five interactive visualisations use data published by the Victorian Gambling and Casino Control Commission (VGCCC, 2026a). The analysis uses the Current Gaming Expenditure by Venue Excel dataset, covering the 2023–24, 2024–25 and 2025–26 financial years. References Victorian Gambling and Casino Control Commission. (2026a). Current gaming expenditure by venue [Data set]. https://www.vgccc.vic.gov.au/sites/default/files/2026-07/current_gaming_expenditure_venue_0.xlsx

Victorian Gambling and Casino Control Commission. (2026b, October 23). Gambling data. https://www.vgccc.vic.gov.au/for-community/gambling-victoria/gambling-data

Acknowledgement

I used ChatGPT as supporting tools to polish R code for readability and improve the clarity of written explanations. All coding, data analysis, visualisation decisions, and interpretation of results were completed independently by me. I reviewed and verified the final work and used publicly available VGCCC data in my submission.