All Reports

The 80-point arb scanner is a trap, and I'd still build the notch widgets

Nov 9, 2026

Daily dashboard screenshot

The 80-point arb scanner is a trap, and I'd still build the notch widgets

Seven new cards today, 47 signals, and every single platform signal came from Product Hunt. Zero from anywhere else. That tells me the radar is reading the same firehose as everyone else this week, so treat the scores as "what's loud" rather than "what's real."

The loudest one is Sports prediction market arbitrage scanner at 80. The pitch is genuinely good: Kalshi, Polymarket, Novig, odds drift apart, opportunities vanish in seconds, the guy is manually tabbing between three sites like a chump. I felt that. But here's the thing — the complaint tells you what the product is. "Opportunities vanish fast" means this is a latency game, and you do not win a latency game from a solo dev's Hetzner box against firms colocating next to the exchange. Also: fees. Also: thin order books on Novig. Also: whether you can even legally offer this to US users on Polymarket without becoming a regulated entity.

So I'd build it, but not as a trading tool. I'd sell the alerts. $49/mo, real-time push, historical spread data, an API tier. The customer isn't me — it's the guy who's convinced he can arb and wants to be told when to click. That's a shovels business, and shovels businesses print money because the customer's P&L doesn't have to be positive for yours to be. That said, I'd feel a little gross about it. That's the honest version. Tooling for people who are probably going to lose money is a real business and I'm allowed to be ambivalent about it.

Custom Mac notch widgets at 72 is the one that actually made me open Xcode in my head. NotchNook proved people will pay a subscription for this, which still feels insane to me — it's a small strip of dead pixels — but the demand is real and the install base is self-selecting: you only have a notch if you bought a recent MacBook, which means you have money. I'd do it in SwiftUI, ship a $12 one-time unlock, no subscription, and win on "I'm not charging you $25/yr for a calendar widget." The risk is Apple. They could absorb it into macOS as a system feature and you're done. But that risk is two years out and the build is two weekends. That's a good trade. The 72 is fair.

I'd skip Simple API key and secret manager at 66. The pain is real — my keys are in three .env files and a pinned Slack message to myself — but the graveyard here is full. 1Password does this, Doppler does this, Infisical is free and open source. "Local encrypted CLI with paid team sync" means your monetization is the one part that's crowded and your free tier is the part that's a weekend of work. No thanks.

The Billing and entitlements layer card at 64 is the sleeper. "Stripe is just a payment rail" is exactly right — tiers, trials, proration, feature gating, all hand-rolled, and everyone hand-rolls it badly. Their pricing is smart too: $19–49/mo tiered by active subs, plus a $99 one-time source package as a low-friction door. That $99 package is the actual growth engine — devs won't subscribe to a billing layer but they'll absolutely pay $99 to not write proration logic again. I'd build this before the arb scanner, honestly. Boring, unsexy, recurring.

And the furniture preview card at 61? AR furniture has been "about to happen" since 2017. Apple already shipped the measurement app. Pass.

What I keep coming back to: 6 of 7 cards came from Product Hunt, and the only non-PH card — the billing layer — is the one I'd actually put money into. Maybe that's a signal about where the good ideas live. Or maybe I'm just tired of launch-day upvote threads. Either way I'm cloning the billing repo tonight and seeing how much of it is already on GitHub.

Back to all daily reports