You're locked into ATP-D2, this is Signal & Noise, and I'm Clark Devereaux. This week the whole damn category split itself clean in two for me: on one side, a regulated app that actually files the paperwork and tells you what the AI is not allowed to do; on the other, an options backtesting funnel with five different prices doing laps around the same $209 anchor like it's an offshore powerboat chase at golden hour. And when the saxophone kicks in, put your beer down, because the pricing page told me more than the backtest ever could. First, let's go to weather.
The Weather Report
The tape is friendly enough right now to make a lot of bad marketing look smart. The S&P 500 is brushing its all-time high, the Nasdaq is within striking distance of its own peak, and that means every AI-trading pitch in your inbox gets to borrow credibility from a market tailwind it did not build. In an updraft like this, a tuned backtest can look like prophecy. That's exactly when you need to separate product mechanics from market mood.
The Fed is still parked in the mid-4% range after two modest cuts, and the signal from policy is basically: we'll go lower if inflation behaves near 3% and labor doesn't crack. That's not a green light for heroics. That's cautious moderation. It rewards patience, balance-sheet reality, and not confusing a decent macro backdrop with proof that your black-box signal engine has discovered the secrets of the universe.
Schwab and Fidelity both used their mid-year outlooks to hammer the same word: selective. AI hyperscaler dominance is real, capex concentration is real, and the next leg of this cycle may not carry every AI-adjacent story on its shoulders. That's the exact environment where a backtest optimized for a broad bull run turns into grade-A, top-shelf bullshit the minute leadership narrows and the narrative cracks.
This Week in AI Investment Technology
This week in AI investment technology, I have one product that's basically honest about being a model-portfolio advisory service, one that's a pricing labyrinth wrapped around secret-sauce options signals, and one broader institutional silence that says more than a launch announcement ever could. Then we'll hit the live court fight where prediction markets are still trying not to get called a casino.
TradeMachine's Five-Price Problem: When 'Proprietary Generative AI' Costs Anywhere From $69 to $209 a Month, the Real Signal Is in the Funnel
You know that Andy Sidaris moment: boats, champagne, somebody smiling too hard, and then the sax kicks in and you realize the real action is not on the deck, it's in the setup? That's this pricing architecture. Across TradeMachine landing pages, TradeMachine Pro is quoted at at least five different prices: $149/month on Home v3, $69/month after a $9 nine-day trial on one page, $99/month after that same trial on another, $99/month flat on a separate funnel, and $179/month on the MTS discount page. Nearly every one of those pages waves the same 'normally $209/month' anchor around. At that point we're not discussing price discovery. We're reading a funnel map.
Now, technically, options backtesting is real work. Return distributions are real analytical objects. If you've built something novel, charge for it. Fine. But CML's own marketing for the AI-branded layer says the algorithms behind its return distribution signals 'will never be made public,' while also claiming '100–200% historically higher returns' without specifying the benchmark, period, or out-of-sample framework. That is not methodology disclosure. That's a dare with a checkout button attached.
And the fine print gets more useful the deeper you go. The MTS discount page explicitly says TradeMachine Standard at $69/month, 'normally $129,' does not include the proprietary return distribution data. So the AI story is not the base product. It's an upsell tier. TradeMachine Platinum sits at $995 per year across funnels, described as non-refundable on multiple pages. Meanwhile CML is not presented in available public disclosures as a registered broker-dealer, registered investment adviser, or FINRA member; what it's selling is a subscription software and analytics service, which means the execution and judgment risk stays entirely with the user.
This lands squarely on Ledger Prediction 4: very few genuinely autonomous, no-human-in-the-loop platforms emerge, and the ones that do are expensive. This week's reading: TradeMachine's proprietary AI signals are locked behind paid tiers, the Standard plan explicitly excludes them, and the pricing is scattered across multiple landing pages with a recurring $209 anchor. Receipt: TradeMachine's own Home v3, get-trademachine-11, blog-get-trademachine, get-trademachine-5, and MTS discount pages all quote different Pro prices, while the MTS page confirms Standard does not include proprietary return distribution data.
And here's the structural threat under all the sax and chrome: the backtest-to-live gap. The independent YouTube breakdown in the brief lays out exactly why historically profitable backtests fail in live markets: fragile parameters, dependence on specific windows, and optimization that amplifies curve-fit risk. That's not some side issue here. It's the whole damn show. Tastytrade, by contrast, offers backtesting with visible user-defined rules and no secret-AI haze. You can disagree with the strategy, but you can actually see the machine. With TradeMachine, the strongest signal on the page may be the fact that the price changes depending on which door you walked through.
Read TradeMachine Home v3 pricing and compare it with the alternate funnel pages. →📡 The Ledger · Prediction 4
On the record: Very few genuinely autonomous, no-human-in-the-loop platforms emerge, and the ones that do are expensive.
This week’s reading: TradeMachine's own pages show AI-branded signals are tier-gated and priced through a five-page funnel from $69 to $179 for Pro, usually anchored to 'normally $209,' while Standard explicitly excludes the proprietary return distribution data.
Alinea's 'Invest Like Politicians and Hedge Funds' Is Actually a Registered RIA With a $120/Year Price Tag — and That's the Honest Contrast
I don't get to say this often enough on this broadcast, so let me enjoy it: this one mostly does what it says on the tin, and the receipts are clean. Alinea Advisory Services LLC shows up in the SEC's IAPD as an approved registered investment adviser, CRD 311782, SEC number 801-120032. Not exempt. Registered. That means Form ADV, Form CRS, fiduciary obligations, the whole grown-up package. The Form CRS and ADV brochure were both updated in March 2026, and the platform's own FAQ puts the advisory fee right there in daylight at $120 per year, auto-renewing, non-refundable for unused time. Custody sits with DriveWealth, a FINRA and SIPC member.
Now let's separate the glitter from the engine. The tagline says you can 'invest like politicians, hedge funds, and AI.' The actual product, per the available materials, is a curated playlist and model-portfolio advisory structure that users can mirror. Long-only, mostly listed equities, some cash equivalents, not some autonomous machine ripping through intraday order flow. And crucially, Allie — the AI chat assistant — says in its own terms of use that it provides educational chat content, does not provide personalized recommendations, does not act as a fiduciary, and has no authority over user accounts. That is the fine print doing its job. The AI is contractually separated from the regulated advice function.
This is Ledger Prediction 5 in a clean suit: most 'AI trading' apps are stock tips, not algorithms; the real product is usually recommendations or copy-style execution of what the user already approved. This week's reading: Alinea's own structure is a registered RIA offering curated portfolios for $120/year, while the AI assistant explicitly disclaims personalized advice and trading authority. Receipt: SEC IAPD shows Alinea Advisory Services LLC as approved, and Alinea's own terms and FAQ describe the fee and the limited role of Allie.
And that gap matters. The difference between 'AI investing app' in the ad and 'regulated model-portfolio adviser with a chat interface' in the documents is not zero, but it's narrow enough that I can live with it. In fact, I want more of this. If you're going to use AI in retail investing, I would much rather see it as explanation, onboarding, and interface wrapped around a disclosed advisory program than as a fake robot oracle whispering alpha into a payment funnel. That's not a sexy answer. It's the right one.
Verify Alinea's SEC registration and advisory status in the IAPD record. →📡 The Ledger · Prediction 5
On the record: Most 'AI trading' apps are stock tips, not algorithms; the dominant real product is a recommendation or copy-trade feed.
This week’s reading: Alinea presents as an AI investing app, but the documented core is a $120/year SEC-registered RIA offering curated model portfolios, while its AI assistant explicitly says it is educational only and has no authority over accounts.
Big Houses Stay in Their Lane: Schwab and Fidelity Drop Mid-Year Outlooks on AI Stocks While Saying Nothing About AI Trading Tools — The Silence Is the Prediction
Every week I check whether one of the big houses has finally lost its mind and launched an AI trading copilot for retail. Every week, nothing. This week Schwab published its '2026 Mid-Year Outlook: U.S. Stocks and Economy' on July 24, and Fidelity published 'Stock Market Outlook Midyear 2026' on July 28. Both talk about AI as a market force — earnings, capex, concentration, hyperscaler dominance, selective risk. Neither one, in the materials cited here, turns around and says: by the way, here's an AI robot to trade your account.
That silence is the story. Schwab's outlook frames modest upside and discusses AI investment surge and concentrated hyperscaler spending. Fidelity leans into robust AI-driven earnings growth while flagging risk variables including energy prices and melt-up dynamics. Bloomberg's roundup of Wall Street expectations shows the same institutional posture: AI as an investable sector theme, not as a retail self-directed algorithmic toy. Vanguard, meanwhile, has been even quieter, with no new U.S. market commentary in the last 30 days according to the brief.
This is Ledger Prediction 2, restated plainly: the big trusted firms stay quiet, or position themselves as the grown-up in the room. This week's reading: Schwab and Fidelity both published sober market analysis on AI and neither used the moment to market an AI trading tool to retail clients. Receipt: their late-July mid-year outlooks discuss AI stocks and macro conditions, not AI trading subscriptions.
And honestly, of course they don't. These firms have compliance departments, reputational memory, and a healthy fear of anything that sounds great in a bull run and turns radioactive in an inquiry. They know the difference between research and a funnel. They are happy to analyze AI as a sector; they are not eager to sell you the saxophone.
Read Schwab's mid-year outlook and compare it with Fidelity's retail market commentary. →📡 The Ledger · Prediction 2
On the record: The big trusted firms stay quiet, or quietly position themselves as the grown-up.
This week’s reading: Schwab and Fidelity both published mid-year outlooks focused on AI as a market theme, with no retail AI-trading product push in the cited materials — exactly the grown-up posture the prediction called for.
Kalshi's Sixth Circuit Argument: The Docket Went Dark Before the Hearing — Prediction Markets Are Still Fighting to Not Be Called a Casino
I had this one circled. July 30, Sixth Circuit, KalshiEX LLC v. Matthew Schuler, docket 26-3196 — exactly the kind of hearing where you want to know what the judges asked, who looked skeptical, where the classification fight is bending. And then the public trail just... stopped. Based on the CourtListener materials cited here, the visible docket runs through the end of March 2026, but there is no entry or transcript in the materials before me confirming what happened at the scheduled July 30 oral argument. So I'm not going to make up courtroom drama because the category would love that. No receipt, no color.
What I can say is that this case is the live machinery of Ledger Prediction 9: prediction markets gain real popularity but get lumped in with sportsbooks and fought as gambling. This week's reading: the legal question remains unsettled in public view, and the lack of accessible post-hearing documentation underscores how slowly the classification clarity is arriving. Receipt: the CourtListener docket identifies the case and scheduled proceedings, but the materials available for this broadcast do not provide a transcript or outcome from the July 30 argument.
Meanwhile the commercial side is not waiting politely in the hallway. The brief notes that Alinea Futures LLC already has an event contracts fee schedule effective March 10, 2026, through a separate entity. In other words, the product infrastructure is being built while the courts are still deciding what box this stuff lives in. That's not scandalous by itself. It's just the shape of the moment: law on one timeline, product on another, and everybody pretending those clocks are synchronized.
Check the CourtListener docket for any new entries on the Sixth Circuit Kalshi case. →📡 The Ledger · Prediction 9
On the record: Prediction markets like Kalshi gain popularity but spend the growth defending themselves from being called gambling.
This week’s reading: The Sixth Circuit case remains the live classification fight, and as of this broadcast the publicly cited docket materials do not show a transcript or outcome from the scheduled July 30 oral argument.
Clark’s Corner
Here's the question I'm leaving open this week, because it's the one the whole category keeps dodging with jazz hands: if the honest version of an AI investing app looks like Alinea — a registered advisory program, curated portfolios, chat interface, explicit disclaimer that the AI is not giving personalized advice — then what is the AI actually for? Not in the ad copy. In the mechanics. What does the model genuinely do that a well-built interface plus a human investment committee does not?
Because if the truthful answer is: it explains things better, it helps onboarding, it surfaces themes, it makes first-time investors less intimidated — that's valuable. Hell, that's genuinely useful. But that is an interface story, not an alpha story. And the industry keeps trying to sell those as the same thing. The honest firms narrow that gap with disclosure. The less honest ones live in it, decorate it, and charge you five different prices depending on which hallway you entered.
So my open question is simple: is there a version of this product that is both honest and genuinely algorithmic in a way the fine print can defend? Not just a better front end on a human-built portfolio. Not just a backtest engine with a sax soundtrack and a secrecy clause. A real one. Because if it exists, I haven't seen its paperwork yet. And until I do, when the saxophone kicks in, put your beer down and read the fee page.