Pitch Behaviour - using AI to develop more Charts

Over the past few days I have been experimenting with AI to see how it might help create a number of new Pitch Behaviour Cards for Minden Cricket. The game itself comes with some examples of these, but I wanted to see if AI could provide detailed and credible Pitch Behaviour Cards for Test grounds around the world.

The following is the result of my "chat" with Chat GPT where I asked it the following questions:

1. Describe the Pitch Behaviour System in Minden Cricket. What is it and how does it work?
2. Research and create ground-specific Pitch Behaviour Charts for major Test venues in Australia, India and England.
3. Simulate 10,000 Test matches for each ground to produce probability distribution charts showing how Pitches might change during a game.

I shared with ChatGPT a PDF of the Minden Rules, the various tables used in the game and each of the included Pitch Cards (there are 11 of them).

As is usual with AI, nothing is straight forward. There were a lot of tweaks to my prompts in order to get the engine to understand what I wanted and to correct the usual basic mistakes AI makes when one first starts interacting with it on a new subject. However, over time, it came back with some excellent analysis and, I am hopeful, the basis of a large range of new Pitch Behaviour Charts for the game.

What is the Minden Cricket Pitch Behaviour System?

First up ChatGPT summarised the system. Here's what it came back with.

In Minden Playing Card Cricket (MPCC), pitch deterioration is governed by the Pitch Behaviour system, designed to simulate the progressive changes in playing surfaces during a five-day Test match. Each Test begins with a selected Pitch Card, such as a “Fair Wicket” or “Grassy Pitch,” reflecting the expected base conditions of a ground. As the match progresses, the pitch can change by transitioning to other cards, representing wear, bounce variation, or increasing assistance to certain types of bowlers.

Pitch deterioration is tracked using the Pitch Deterioration Number (PDN). After every session, a 2d6 roll determines whether the PDN increases:

• 2–6 → +1 to PDN (41.7% chance)

• 7–11 → No change (55.6% chance)

• 12 → +2 to PDN (2.8% chance)

In a full five-day Test, there are 14 rolls total: 3 per day on Days 1–4 (morning, afternoon, end-of-day) and 2 rolls on Day 5 (no end-of-day roll). Statistically, the pitch will increase by ~6.6 PDN points over a full match. This deterioration causes scheduled pitch transitions at certain PDN thresholds, reflecting the natural arc of many real-world Test matches where surfaces deteriorate gradually but can vary based on luck and conditions. Customising these transitions to each ground allows players to replicate famous venue behaviours—whether it’s the Gabba’s consistent pace or Chennai’s dramatic late-match turn.

The analysis took several hours to complete and is quite detailed. I will share the full analysis soon, but for now, below is the MCG and how ChatGPT sees it.

The great Tony Greig giving one of his pitch reports at the MCG during the 1982-83 Ashes.

MCG. Australia. Melbourne - Pitch Behaviour and Analysis

Real-World Summary of MCG (Melbourne Cricket Ground):

The MCG is known for its true, hard surface early, with good pace and bounce on Day 1. Historically, it flattens out on Days 2–3, leading to high-scoring innings. Spin starts to feature late, especially Day 4–5, as footmarks develop. The pitch doesn’t typically turn into a “minefield,” but there’s variable bounce and deterioration if matches go into the fifth day. Batting collapses happen occasionally on Day 5, but more from pressure than extreme pitch damage.

Key Characteristics:

• Day 1–2: Hard and true, rewards stroke play.
• Day 3: Flattens, less for fast bowlers.
• Day 4–5: Spin comes in, occasional variable bounce.

MCG Pitch Behaviour Chart (Custom PDN Thresholds):

This PDN spread reflects MCG history: batsmen dominate early, deterioration starts from Day 3 afternoon, Day 5 brings challenges but rarely unplayable conditions. Crazy Wicket is included as an outlier for rare collapses.

• 0–4 - Hard & Bouncy (Pitch Card - 002) - Consistent pace, rewards batsmen with higher SR; balanced Day 1–2 conditions.
• 5–8 - Good Batting (Pitch Card - 003) - Flattens out by Day 3, minimal help for bowlers except top-class skill.
• 9–11 - Taking Spin (Pitch Card - 011) - Pitch starts wearing, spinners come into play, but batting still possible.
• 12–13 - Erratic Bounce (Pitch Card - 010) - Day 5 variable bounce, tougher for lower-order and lower-rated batsmen.
• 14+ - Crazy Wicket (Pitch Card - 007) - Rare, but extreme deterioration can happen by late Day 5 under certain conditions.

MCG Probability Distribution Chart from 10,000 simulated Tests

• Hard & Bouncy (0–4 PDN): ~22.8% of matches finish before the pitch changes from its Day 1 state.

• Good Batting (5–8 PDN): ~38.3% of matches spend significant time on the flatter, bat-dominant surface.

• Taking Spin (9–11 PDN): ~26.7% of matches reach the phase where spin starts to bite by Days 4–5.

• Erratic Bounce (12–13 PDN): ~9.5% experience late variable bounce challenges.

• Crazy Wicket (14+ PDN): ~2.7% end in extreme deterioration, typical of rare final-day collapses.

ChatGPT's Conclusion

Most MCG matches follow a classic Australian arc — strong batting phase mid-match, with gradual wear but rare total breakdown, in line with historical Test trends.

---

The above shows some interesting results. I expect other Minden Cricket players will be able to tweak these results, and perhaps will have a few scenarios for different pitches at the MCG. For example, if it's been raining in Melbourne prior to the Boxing Day Test, one might expect the pitch to favour bowlers on Day One, etc.

The design goal of the Minden Cricket pitch system was to allow the player freedom to either use pre-canned Pitch Behaviour Charts or to simply decide their own progression. I hope the AI work I am doing provides more of the later for players keen to test out different scenarios.

I will share the other Test venue information as it comes in, and perhaps you dear reader might like to do your own analysis -- either by yourself or with the help of AI.

Either way I'd be interested to know the outcomes and your thoughts on the above.

A postscript on AI

By the way, I did the same exercise with Grok (X's AI engine) and the responses were nowhere near as good or insightful as ChatGPT. Why this is the case, I am not sure. At any rate I stuck with ChatGPT after initially trying it on both systems. Also, for the record, I pay a subscription for each, so am getting access to better tools and deeper analysis than one gets with the free versions.