The 5 best prompts for analyzing NBA games with AI

• Each prompt covers a single question: form and pace, injuries, fatigue, totals and handicap, and a checklist.
• AI makes up statistics with total confidence: ask for a source and check every number before using it.
• A prompt organizes your analysis, it does not decide the bet and it does not guarantee a result.

Which prompts work for analyzing an NBA game?

An NBA game has many variables.

Recent form, pace, injured players, rest and the price of the odds.

If you throw it all at an AI at once, the answer comes out mixed up.

That is why this guide splits the analysis into five prompts.

Each one asks a single question.

You put the answers together and draw your own conclusions.

Every prompt has parameters in [BRACKETS].

Replace each one with your own data before pasting.

If you leave a bracket empty, the AI will fill it with whatever it comes up with.

Before you paste: three rules so the AI answers better

1. Turn on web search: without it, the AI answers with what it learned and may be out of date.
2. Give it the data you already have: the odds, the line and the exact date keep it from making them up.
3. Ask for one thing per message: a short, verifiable answer is worth more than a long report.

If you want to build prompts tailored to you, try the prompt builder.

For a more general foundation, read the Top 5 ChatGPT prompts for betting.

Prompt 1: recent form and game pace

This is the opening prompt.

It gives you the picture of each team before you look at the details.

Analyze the [HOME TEAM] vs [AWAY TEAM] game on [DATE] in the NBA.
Tell me, for each team: 1) results of the last [N] games, 2) points scored and allowed per game over that stretch, 3) pace (possessions per game) and 4) whether it plays better at home or on the road.
Cite the source for every number.
If you cannot verify a figure, write “I don’t know”.

What it returns

A comparison sheet for each team.

It should include streaks, scoring averages and an idea of the pace.

It is the foundation for the other prompts.

How to verify it

Open the source it cites and compare two numbers at random.

Check that the games are from the stretch you asked for and from the current season.

If it mixes seasons, repeat the prompt and state the year.

Prompt 2: injuries, doubts and how they change the game

In the NBA, one absence can move the line more than any streak.

This prompt sorts out who plays and who does not.

For the [HOME TEAM] vs [AWAY TEAM] game on [DATE]: list the players confirmed out and those listed as questionable on each team.
For each one, tell me: minutes per game, role on the team and who replaces them in the rotation.
Give me the date and time of the injury report’s source.
If the report is old or you cannot find it, let me know.

What it returns

A list of absences with each player’s role.

Also an idea of how the rotation reshuffles.

The date of the source is the most important part of the answer.

How to verify it

Cross-check the list against the league’s official injury report.

Look at it again a little before tip-off, because it changes until the last minute.

A “questionable” tag is not an absence: do not take it for granted.

If player markets are your thing, see the guide to NBA props with AI.

There we look at how minutes and absences break a line.

Prompt 3: back-to-back, travel and fatigue

The NBA schedule is tough.

There are games on consecutive nights and long trips.

This prompt checks each team’s rest.

For [HOME TEAM] and [AWAY TEAM], tell me how many games each one played in the last [N] days and how many days of rest each has before [DATE].
Indicate whether either one is playing the second night of a back-to-back and whether it traveled between cities.
Also tell me how each team has done in that situation this season, with the source and the number of games.
If the sample is small, say so.

What it returns

A summary of each team’s rest.

Also whether there is a back-to-back and what happened in similar situations.

The warning about a small sample is key.

How to verify it

Build the schedule yourself from each team’s official calendar.

Count the days of rest by hand.

Be wary of any percentage based on five or ten games.

The topic has its own article: back-to-backs in the NBA and betting.

There we explain the effect and its limits with an example.

Prompt 4: totals and handicap, the market against your read

This prompt does not look for a winner.

It looks to understand whether the points line or the handicap line makes sense.

For that it needs you to give it the line and the odds.

The market offers for [HOME TEAM] vs [AWAY TEAM]: a points total of [TOTAL LINE] (over at [OVER ODDS], under at [UNDER ODDS]) and a handicap of [HANDICAP LINE] at odds [HANDICAP ODDS].
Using the form and pace data I gave you, explain whether the total looks high or low to you and why.
Calculate the implied probability of each price and the difference with your estimate.
Show the math step by step and flag which part is opinion.

What it returns

A reasoned read of the line and the implied probability math.

It does not tell you what to bet.

It shows you where your read and the market’s read part ways.

How to verify it

Redo the math with the odds converter and the expected value calculator.

AIs get arithmetic wrong more often than it seems.

If your math and its math do not match, send yours.

The numbers in the table are an example to show the math.

A gap in your favor depends on your estimate being good.

That is why it has to be tested across many bets, not one.

If handicap gives you trouble, review how it works in the article on handicap betting.

And to understand the margin, read the one on implied probability.

Prompt 5: the checklist before you bet

This is the closing prompt.

It does not analyze the game: it reviews your analysis.

You ask it to look for gaps before you put money down.

This is my analysis of [HOME TEAM] vs [AWAY TEAM] on [DATE]: [PASTE YOUR CONCLUSIONS AND DATA].
Act as a critical reviewer.
Tell me: 1) which figure is not verified, 2) what important information is missing (injuries, rest, pace), 3) what bias I may have and 4) what would happen to my bet if my estimate were [LOWER PROB.]%.
Do not give me a betting verdict.
Only flag risks.

What it returns

A list of gaps and risks.

It is the most useful prompt for stopping an impulsive bet.

If the list is long, the best move may be not to bet.

How to verify it

Tick off every risk you resolved.

The ones you cannot resolve become a reason to skip the game.

A good analysis can also end in not betting.

How to use the 5 prompts in order: step by step

1. Pick a game and write down the basic data

   Teams, date and time.

   Add the total line, the handicap and the odds.

   Without this data, the AI improvises.

2. Run prompts 1, 2 and 3

   Form and pace, injuries and fatigue.

   Copy each answer into a document of your own.

   Mark what comes with a source and what does not.

3. Verify three figures from each answer

   Open the source and check the number.

   If one fails, distrust the rest of that answer.

   Repeat the prompt if needed.

4. Run prompt 4 with the real line

   Paste the line and the odds of the moment.

   Redo the math with your calculators.

   Write down your estimate before looking at its own.

5. Close with prompt 5

   Paste your conclusions and let it look for gaps.

   Resolve the risks you can.

   If many remain, skip the game.

6. Decide how much and record everything

   If you decide to bet, work out the amount with the Kelly calculator or the bankroll one.

   Save the analysis and the result.

   Without a record there is no learning.

The limits of these prompts

• AI does not see live odds unless you paste them in.
• It can mix up seasons or players who are no longer on the team.
• Pace and scoring numbers change every night.
• It does not know whether an injury was confirmed ten minutes ago.
• A well-written answer is not a true answer.

Prompts help you think in an orderly way.

They do not replace official data or a model with a track record.

Treat them as a notebook, not as an oracle.

What Gambeta does with the NBA, no promises

Gambeta publishes picks from its own model.

Every pick is recorded with its odds before the game.

You can review it afterwards in the public record, win or lose.

Every day the Betting Journal comes out, an article with the day’s recap.

You also get a free Kelly calculator and a bankroll one.

To see the basketball section, head to the NBA predictions with AI.

A pick is not a promise.

Even if the model publishes its record, no bet is safe.

What to read next

If you want more, continue here.

For the player market, the guide to NBA props with AI.

To understand the markets, how to bet on the NBA.

And to get back to the basics, the complete guide to AI for betting.

Frequently asked questions

Which prompt should I use to analyze an NBA game?

Start with the form and pace prompt, which gives you the big picture.

Then add injuries and doubts, and the fatigue check.

Finish with the checklist before you bet.

Does ChatGPT know today’s injuries?

Only if web search is turned on, and even then it can be wrong.

An injury status can change several times a day.

Always confirm in the league’s official report before you bet.

How do I know if the AI made up a statistic?

Ask for the source and open it.

If the link does not exist or the number does not appear, discard it.

A figure without a source you can open is no use for betting.

Do these prompts work with Gemini, Claude or Grok?

Yes.

They are plain text and work in any chat AI.

The quality of the answer changes depending on whether it has web search or not.

Does a good prompt guarantee I will win?

No.

A prompt organizes information, it does not predict the result.

A pick is not a promise, and betting always means a risk of losing money.

Where can I see NBA picks with a verifiable record?

At Gambeta, every pick is published with its odds before the game.

You can review the public record and read the daily Betting Journal.

It is meant to inform you, not to bet without judgment.