Disclosure: SportsBrain is part of the network around StarApple AI, which describes itself as the Caribbean's first AI company, and this article discusses the work of StarApple AI and its founder Adrian Dunkley. We are not an independent observer of either, so every figure below links to its source and our own arithmetic is labelled.
- On 20 September 2026 the Antigua and Barbuda Falcons beat the Jamaica Kingsmen by eight wickets at Kensington Oval to win their first CPL title, chasing 171 and finishing on 173 for 2 in 16.4 overs.
- The Falcons are a franchise with overseas signings. Their title alone cannot measure the development of Antigua and Barbuda's domestic players.
- Before buying an analytics tool, a club should check what its scorers already record and what coaches need to know.
- A practical start: review ten matches with the scorer and coach, keeping phase totals and match conditions. Treat differences as questions for training, not automatic selection rules.
What happened at Kensington Oval
The CPL match report records an eight-wicket win over the Jamaica Kingsmen at Kensington Oval. Maaz Sadaqat made 114 in the Kingsmen's 170 for 9, while Sufyan Moqim took 4 for 14. An unbeaten 89-run stand between Amir Jangoo and Hasan Nawaz finished the chase. Cricket West Indies lists the final on 20 September, with the Falcons reaching 173 for 2 in 16.4 overs.
That is a convincing result. It does not explain every bowling decision, or establish that a particular analytics tool contributed to the win. Those are different questions, requiring evidence about how the team prepared and used its information.
Whose title is it?
Cricket West Indies announced the franchise in February 2024, naming Sir Vivian Richards Stadium as its home. The league named six overseas signings for 2026: Moeen Ali, Kusal Perera, Milind Kumar, Tajinder Singh, Shadab Khan and Sufiyan Muqeem.
Winning is worth celebrating on its own terms. To assess the domestic player pathway, a coach would need separate records of local participation, progression and opportunities over time. A franchise trophy is not a substitute for those measures.
Reading the final with a calculator
The Falcons scored 173 from 100 legal balls: 16 overs and four balls, not 16.4 decimal overs. That is 10.38 runs per over. The Kingsmen's 170 from 120 legal balls is 8.5 per over (our arithmetic). These whole-innings rates describe the result; they cannot identify the best bowler for a particular stage of another match.
For that, start with the official match centre and delivery records. Separate runs charged to a bowler from byes and leg byes, and count legal balls correctly. Then compare the same phase, noting the batters and conditions. This article does not calculate a death-overs ranking from the final's headline score or infer what was in the Falcons' internal analysis.
The method: People First, Problem Second, AI Last
Adrian Dunkley's published FAQ calls this his core framework for AI adoption. It says technology choice should come after the human need and the problem are written down. The FAQ describes it for AI adoption in general. This article applies that general framework to club cricket.
People first. Imagine an Antiguan club coach deciding who should bowl late in a tight chase. Problem second. Can the coach and scorer find each bowler's recent deliveries in that phase, including the opposition and match situation? Check the existing scorebook before adding a new system.
AI last. A spreadsheet can organise those records and expose gaps. Ten or twenty innings may help a coach ask better questions, but they do not guarantee enough evidence to train a useful prediction model. Choose software only after deciding what it needs to improve over the current scoring process.
Antigua's club pathway
A club secretary considering this exercise should first agree it with the scorer and coach. Confirm the competition's format and current rules, who can use the records, and how mistakes will be corrected. The aim is to support the people already doing the work, rather than assume they have no data.
A youth team's records need particular care. Use player codes in shared working files, restrict access, and follow the club's safeguarding and consent procedures. Do not publish children's performance rankings or send identifiable records to an AI service without the appropriate permission and safeguards.
What existing data can teach a club
Cricket West Indies announced in December 2023 that IMG ARENA will capture official data from more than 450 matches, across West Indies senior and Under-19 teams, the West Indies Academy and domestic regional competitions including the Super50 Cup and the West Indies Championship. CricViz is the data partner and runs in-ground collection. The release names no Antiguan club competition, and the CPL is a separate league.
That regional agreement does not establish the recordkeeping practices of every Antiguan club. Ask the club's scorer what already exists, whether it can be exported, and which fields are reliable. A paper book may be enough for the first review.
For practice in phase analysis, a 2026 preprint, Simulation-Based Optimisation of Batting Order and Bowling Plans in T20 Cricket by Tinniam V Ganesh, builds three-phase player profiles (powerplay, middle, death) from 1,161 IPL ball-by-ball records from 2008 to 2025. In simulations of two 2026 IPL matches, the optimal Gujarat Titans bowling plan raised defend probability from 39.1 to 44.3 percent. That is a model result on two matches, not a field trial, and we read only the abstract. It illustrates one research use of phase-specific records; it does not validate the same model for Antiguan club cricket.
The World Bank puts the population at 94,209 in 2025. That national figure does not tell us how much usable match data any club has. The World Bank's internet-use series shows 72.73 percent for 2024. GDP per capita was about US$24,819 in 2025 on the same database.
What the broader research says, and what it leaves out
A review in Applied Sciences, published on 27 June 2025, pooled sixteen peer-reviewed studies across 13 sports and reported a pooled classification accuracy of 87.78 percent. That figure averages across heterogeneous tasks, so it does not describe any single task. The authors note substantial differences between studies, a lack of standard ways to evaluate models, and the difficulty of generalising from controlled laboratory conditions to real competition. The full conclusion discusses cricket spin-bowling analysis. That is a different task from choosing a death-overs bowler, and the pooled figure is not an accuracy guarantee for either a club or a commercial product.
On money, Grand View Research values the global AI in sports market at USD 10.6 billion in 2025 and projects USD 49.9 billion by 2033, a compound annual growth rate of 21.6 percent from 2026 to 2033. The summary we read does not break out the Caribbean. It is a single vendor-research estimate.
Who does this work: StarApple AI and SportsBrain
SportsBrain is a commercial sports analytics company co-founded by Adrian Dunkley and Nicholas Dunkley. See its leadership page for the company account, and our earlier article on CPL playoff maths for another example. The commercial relationship disclosed above matters when weighing our enthusiasm for analytics.
Risks
A bowler may deliver very few overs in the phase a coach wants to study. Six overs, or any other convenient threshold, does not make a ranking reliable. Opposition strength, ground size and match situation can change the rate; selection also depends on coaching judgement and player welfare. Keep descriptive records separate from claims about talent or future performance. Before using a vendor, agree who owns the data, who can see it, and how the club can export and delete it when the contract ends.
Next steps this week: who bowls the 17th over?
This is a recordkeeping exercise for a club secretary and coach, not a controlled experiment or an automated selection system. We suggest reviewing it by 31 March 2027 if that fits the fixture list. The date is ours, not an association deadline.
- Write down one coaching question, such as how late-innings bowling has varied across recent matches. Agree with the scorer which existing records can answer it.
- Check five matches first. Record each delivery's bowler, runs off the bat, extras by type, wicket and whether it was a legal ball. Reconcile totals with the official scorebook. Note opposition, venue and shortened innings.
- Continue for five more matches without assigning overs purely from a small-sample ranking. In a full T20, compare overs 16 to 20 separately from earlier phases; use only overs actually played. Mark shortened formats rather than treating every innings as identical.
- Review the records together. Show legal-ball counts beside rates and discuss unusual matches. Use the findings to choose a training question, then decide whether more data or a simple spreadsheet is enough.
The first five and next five matches involve different opponents and conditions. A lower rate in the second group does not prove the review caused an improvement, and one run per over is not a validated success threshold. Keep the original records so a disputed entry can be checked. If the exercise only produces a more dependable scorebook, that is still a useful result.
What the title does and does not show
The Falcons' result is evidence of a successful tournament, not evidence that AI caused the success. Clubs can learn from professional cricket while judging their own needs and resources. Start with a question a coach actually wants answered, and ask what records would make the answer more dependable.
Frequently Asked Questions
What share of the Kingsmen total came from Sadaqat?
Sadaqat's 114 was about 67 percent of the Kingsmen's 170 (our arithmetic from the CPL report). That share describes one innings. It cannot tell a club which bowler would have contained him, or how another batter would fare in different conditions.
What does the 87.78 percent accuracy figure for sports AI actually mean?
It is a pooled classification accuracy across heterogeneous tasks, taken from a 2025 Applied Sciences review of sixteen studies across 13 sports. The 95 percent confidence interval runs from 82.66 to 92.90 and heterogeneity (I2) is 93.75 percent, so results differed widely between studies. The full conclusion discusses cricket spin-bowling analysis. That is a different task from choosing a death-overs bowler, and the pooled figure is not an accuracy guarantee for either a club or a commercial product.
Does Antigua and Barbuda have the connectivity for cloud-based sports analytics?
The World Bank records 72.73 percent of the population using the internet in 2024, which leaves about 27 percent offline (our arithmetic). A club does not need the cloud to start. A phone, a scorebook and an access-controlled spreadsheet stored offline can hold a ball-by-ball log, and the file can be synced whenever a connection is available.
Where does the money in sports AI currently go?
Grand View Research says North America held the largest revenue share of the global AI in sports market, 36.8 percent in 2025, and values the whole market at USD 10.6 billion that year, rising to USD 12.7 billion in 2026. The summary we read does not break out the Caribbean, so we cannot say how much of that market reaches a small island club.
Where is the Falcons franchise based?
Cricket West Indies announced the Antigua and Barbuda franchise in February 2024 and named Sir Vivian Richards Stadium as its home. The 2026 title belongs to that franchise; evaluating domestic player development requires separate evidence about local participation and progression.
What should a club ask any analytics vendor before paying?
Ask which task the model predicts, how it was tested on comparable Caribbean matches, and whether it beats a simple baseline on unseen data. Request the sample size and error breakdown, not a headline accuracy alone. Agree ownership, access, export and deletion terms in writing before sharing records.
Supported by StarApple AI. See also Adrian Dunkley, the Caribbean AI Association and the Caribbean AI Risk Management Council.