Pinned post of "31 matches, all hit" and an AI match-prediction app with only 12 installs a day: Bravo

Bravo is a sports scores and AI win-probability app built by a small Brazilian marketing company: football, NBA, NFL and F1 in one subscription. In six weeks it set up a full funnel: a scores site, social accounts, free prediction games, public hit rates, a cheap subscription and a 50% lifetime affiliate commission. The app went live two and a half weeks ago and has about 1,300 Google Play installs in total, now roughly 12 a day. What's worth studying is how it builds the funnel. The product itself has not yet shown anyone pays for it.

A pinned post

Open @bravoscores on X and the pinned post is a screenshot with a Portuguese caption: "O BRAVO É UMA LOUCURA 🤯 Our model got 31 of 31 right, 0 wrong... sir, you are not Bravo PRO, and frankly I'm truly sorry for you", followed by the subscription link.

Scroll down and it's the daily routine: "O que o Bravo prevê para Operário-PR x Sport, hoje às 19:30" (what Bravo predicts for tonight's 19:30 match), then three probability lines, "both teams score 55%" and "under 2.5 goals 51%", always ending with "Previsão é chance, não promessa" (a prediction is a chance, not a promise). It shouts 31 of 31 correct while saying predictions are not promises. The approach: copy the flexing language of tipster circles, and sell the product through statistics and transparency. The account has posted 2,336 tweets, yet in the last 30 days almost nobody else on X has mentioned it.

More than a month, three identities

From Portuguese football scores, to community prediction games, to "AI prediction subscriptions" for US sports, all in under two months. The Play store title and description are in English, opening with "The most complete sports app. Every game, predicted."

How much it actually grew

A small peak in late October to early November: over 600 installs in four days, then down to roughly 12 a day. No ratings, no reviews, not charted in any country.

Getting 255 installs on day one is hard for a new app without outside traffic. The likely explanation (my inference): the 46k Instagram followers and visitors from the scores site were pushed over in one go on launch day. Those 46k followers also deserve a question mark. The account follows only 13 people and has 102 posts, its earliest post is from late September at the earliest, so it reached 46k in about six weeks. It could be a clip that really blew up, bought followers, or a renamed old account. We can't tell yet. The same clip on TikTok got 39 followers, with each video at tens to over seven hundred views (11-06).

What kind of funnel it built

Broken down, the six parts are all established playbooks. The rare thing is a small team installing all of them in six weeks:

• Programmatic scores pages: 217 competitions, one page per match, per team and per league, in Portuguese and English, with internal links across the footer. The store links carry utm_source=google&utm_medium=organic, which shows people really do arrive from Google search. But the domain is less than two months old, so how much search traffic it gets is an open question.
• Social clip accounts: Instagram reposts football memes and classic clips, and X posts daily probabilities and prediction leaderboards by match. The brand line: "Enquanto os outros chutam, o Bravo calcula" (while others guess, Bravo calculates).
• Free prediction games as the hook: the monthly "Palpitão" leaderboard and private leagues with friends. It published its November board itself: "86 pessoas na disputa", 86 people competing.
• Public hit rate: this is the piece it put the most work into. The PRO page reads "Every hit. Every miss. Logged before the game. Checked after. In public." Backtested on 4,014 NBA games (2023-24 to 2026-27 seasons), Bravo hits 67%, while "always pick the home team" hits 55%. Across the 1,412 games where Bravo gave 70%+ confidence, its win rate was 80% (official site, 11-06). Tip-style products rarely compare against a baseline like this.
• Cheap subscription, web and app paths: the web version uses Stripe, $7.99/month or $29.99/year (about $2.50/month), with the line "Other AI sports apps charge up to $99.99 a month", directly benchmarking US competitors. The first FAQ item: "Do I need the app? No, it works in your browser." Play in-app purchases are the other route, at $1.99 to $139.99 per item.
• 50% lifetime affiliate commission: the partners page says you get half of what referred PRO users pay, including renewals, and "anyone who signs up through your link is yours forever". Payouts go out monthly via Pix, with an extra R$1,000 bonus once you reach 1,000 active accounts. I haven't found any influencer promoting it yet.

The company behind it explains why it looks like this. The operating entity, Rice Agência LTDA, is registered in Campos dos Goytacazes, Rio de Janeiro state, with "promotional services" as its registered business, and a separate account handles social media management. The scores data is sourced externally. The prediction model, judging from the description, is statistics built on margin, home advantage and back-to-back games. The company is strong at content and distribution; technology is not its strength.

Where the money comes from

No public revenue, and the app hasn't charted in any store's top grossing. Structurally: web Stripe subscriptions avoid store fees, and the money saved goes toward paying affiliates 50%, the same model common in betting-tip circles. Play in-app purchases are just a fallback. It also deliberately avoids the word "betting", describing itself as "Bravoscores provides sports statistics and predictions", which makes passing store and ad review easier. Whether anyone actually pays is not visible from outside.

Neither it nor the company shows up in the Meta, Google or TikTok ad libraries. But those libraries are incomplete, so not finding it doesn't prove it hasn't advertised.

Who it's competing with

Everyone is competing for this. At every layer it has rivals orders of magnitude bigger (Play installs, November):

Bravo has nothing other players lack. Its distinctiveness is in the combination: one subscription covering four sports, a public hit rate with a baseline, and very low pricing. That makes good social material, but on the install curve it hasn't turned into organic growth yet.

If it were you

C: Study its funnel, but don't enter its category. Bravo shows a small team can set up "scores site + clip accounts + free games + public hit rate + cheap subscription + 50% commission" in six weeks. It hasn't shown that this stack drives growth, let alone that anyone pays. AI match prediction is already crowded.

Who it suits: people who already have an audience in a particular minority-language sports market, such as a football clip account with tens of thousands of followers, a scores site with search traffic, or a team with access to creators in betting-affiliate circles. The most expensive part of the funnel is the audience. If you already have that, the other pieces can be added week by week, following Bravo.

Who it doesn't suit: indie developers with no audience who hope to break out in the store on "more accurate AI predictions". 180Score alone is near 3M installs, there's a large pack of near-identical apps, and the scores layer belongs to Sofascore and Flashscore.

How to do it: stop building yet another all-in-one scores app. Two things are worth taking. First, a public hit rate with a baseline and locked before the game, like "versus always picking the home team". It is far more credible than "80% accuracy", and it produces shareable material every day. Second, the combo of web subscriptions plus affiliate commissions: web subscriptions dodge store fees, and that money pays creators half their commission. Pick a language and league that prediction products don't cover well yet, such as Spanish, Turkish, Arabic or Indonesian local leagues.

How to get users: build the account first, then the site, and the app last. Clip accounts grow the following, scores pages catch search traffic, free games keep people around, and affiliates with local sports creators drive subscriptions. Bravo's 255 installs on launch day suggest "build the audience before launching the app" works. Only 12 a day afterward suggests one wave isn't enough, and you need a steady acquisition channel after it.

When to move: follow the season. European football and the NBA are mid-season, so there is content to post every day until 2026-27 ends. To hit a moment, build the audience ahead of the next major tournament or new season.

What would mean I got it wrong: ① If Bravo's iOS version brings a second wave, or Play installs pass 50k within a few months, then "multi-sport + transparent hit rate + low price" really can break out in English, and this rating should go up. ② If partner-page affiliates start appearing and outside users post their picks on X and Instagram, the commission machine is running. ③ The reverse: if by 2027 early it's still below a few dozen installs a day and the prediction games still have dozens of players, then no matter how complete the funnel looks, without original content and a real audience it won't turn.