Skip to content

Instantly share code, notes, and snippets.

@jacoblyles
Created February 9, 2026 17:52
Show Gist options
  • Select an option

  • Save jacoblyles/def47d7da6141f3448936e2b584933da to your computer and use it in GitHub Desktop.

Select an option

Save jacoblyles/def47d7da6141f3448936e2b584933da to your computer and use it in GitHub Desktop.
Missing Bowl Events: Equipment Regression Analysis

Missing Bowl Events: Equipment Regression Analysis

Summary

We analyzed the relationship between center equipment configuration and the rate of missing bowl events across 155 centers over the last month. Scoring hardware generation is the dominant predictor of missing bowl events, explaining ~92% of the variance. Centers running legacy (pre-BES X) scoring hardware experience missing event rates of 25-72%, while centers on current-generation BES X hardware consistently show rates below 1%.

Methodology

  • Dependent variable: Percent of completed games with missing bowl events per center (logit-transformed to handle the bounded 0-100% range)
  • Independent variables: Conqueror version (bucketed), scoring hardware, pinsetter manufacturer/model
  • Model: Weighted least squares (weighted by total games per center) with HC3 robust standard errors
  • Data window: Last 30 days, centers with 50+ completed games (n=155)
  • Source tables: FACT_BOWL (event data), DIM_CENTER (equipment metadata)

Management was excluded as a predictor (all 155 centers share the same value). Scoring software was excluded due to near-perfect collinearity with scoring hardware; a separate model confirmed it carries redundant information.

Key Findings

1. Legacy scoring hardware drives missing events

All five legacy hardware categories are significant predictors of elevated missing bowl rates (p < 0.001 for each):

Scoring Hardware Centers Median Missing % Logit Coefficient
BES (Classic +2 environments) 1 72.0% +7.3
Bowland-X 18 33.7% +5.1
BES Premium 22 27.8% +4.9
BES (RDB hardware) 3 14.3% +4.1
BES Standard 1 9.7% +3.6

The reference category is 4HD (the most common BES X hardware). All BES X variants (4HD, 5HD, 6HD, and combinations) show missing rates consistently under 0.5%.

2. Pinsetter model has a smaller but real effect

After controlling for scoring hardware, three pinsetter effects are significant:

Pinsetter Effect p-value
AMF 8270 OmegaTec +1.0 pct pts higher 0.004
AMF EDGE - String -1.0 pct pts lower 0.002
Brunswick A -0.8 pct pts lower 0.007

These effects are much smaller than the hardware effect, suggesting pinsetter choice is a secondary factor.

3. Conqueror version is not independently significant

After controlling for hardware and pinsetter, conqueror software version does not significantly predict missing event rates. The apparent version effect in raw data is driven by older versions co-occurring with older hardware.

Interpretation

The missing bowl events problem is fundamentally a hardware generation issue. Legacy scoring systems (BES Premium, Bowland-X, BES Classic, BES RDB, BES Standard) batch multiple bowl updates into single event messages at much higher rates than current-generation BES X systems. This is likely due to differences in how the scoring hardware firmware reports events to the Conqueror software, rather than a software configuration issue.

The 45 centers currently on legacy hardware account for the vast majority of missing bowl events in the fleet. Centers that have been upgraded to BES X hardware (any HD variant) see near-zero missing event rates regardless of their conqueror version or pinsetter configuration.

Model Diagnostics

Metric Value
R-squared 0.919
Adjusted R-squared 0.904
Condition number 21.1
Residual normality (Omnibus p) 0.178 (passes)
Residual kurtosis 3.6 (near-normal)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment