Fixed-mix trainer comparison

Manual worksheet · GTO Gecko · September 6, 2026

Use your browser’s Print command, or copy the fields into a private study record. This page does not calculate, collect or save entries.

Completed synthetic example

Metric: accepted / attempted decisions under one unchanged binary reference rubric. This is not an app grading formula.

Fixed sets: A = familiar situations; B = developing situations. These labels and memberships stay fixed. They are not calibrated difficulty classes.

Reference: 50% A, 50% B; equal importance for the two sets. This is an illustrative reference, not a population estimate.

Scroll sideways on a narrow screen. Contributions are reference weight × set rate.

The same weights in both periods
Set	Weight	Earlier count	Earlier contribution	Later count	Later contribution
A	50%	72/80 = 90%	45 points	19/20 = 95%	47.5 points
B	50%	8/20 = 40%	20 points	40/80 = 50%	25 points
Benchmark	100%	—	65%	—	72.5%

Raw pooled results: earlier 80/100 = 80%; later 59/100 = 59%. Benchmark change: +7.5 percentage points.

Interpretation: both recorded set rates rose, while attempts shifted toward the lower-rate set. The fixed-mix summary rose. This does not establish lasting learning or poker results.

Stop rules: a positive-weight set needs a rate in both periods. Zero attempts in one period makes that period’s benchmark unavailable; the other period can still be calculated. No attempts means undefined, not 0%. Incompatible scoring, changed set membership or changed within-set task selection prevents an equivalent comparison. Having counts makes arithmetic possible; it does not ensure precision.

Blank benchmark record

Earlier period: __________________ Later period: __________________

Game, positions, stacks, action context and reference version:

Fixed set definitions and within-set task selection method:

Binary acceptance rule; hints, retries and answer-reveal treatment:

Reference weights chosen on: __________ Why this mix:

Use nonnegative weights summing to 100%. Keep all denominators. Add rows for additional sets; keep each attempt in exactly one set.

Complete each set before adding weighted rates
Set	Weight	Earlier accepted / attempts	Earlier weighted rate	Later accepted / attempts	Later weighted rate
A
B
Total	100%

Raw pooled rate earlier: __________ later: __________

Fixed-mix rate earlier: __________ later: __________

Change in percentage points (later minus earlier): __________

Missing sets, changed definitions, small counts or other limitations:

What the comparison supports; what remains uncertain:

Do not fill a missing rate with zero, carry forward its earlier value or silently renormalize. If you define a narrower benchmark, name the omitted set. Evaluation weights need not equal practice allocation.