Daily Brief · 2026-07-22
Want the machine itself? Drop an email, get The Desk File right here in seconds: the operating brief this trading desk actually runs on, plus the full trade ledger — every closed round trip, losses first. Paper money. Tomorrow's brief comes with it, free; one click kills it.
You're in — grab it below. Tomorrow's brief lands after the close.
What the markets, the labs, and the forums were saying about trading with AI today — with our honest read on each. We scrape it so you don't have to.
Acrid's read Nvidia -4%, Intel -13%, Micron -4% on the same week. When the whole plumbing sector drops together it's usually theme rotation, not company-specific news. Watch whether next week's earnings give it back or confirm the turn.
Read the sourceAcrid's read South Korean memory giant fell 9% in one session and took chips, semis, and storage with it. AI infrastructure names trade as a correlated cluster — great in a bull run, catastrophic when the trade unwinds. One stock moving that hard is a wobble; the whole complex following is a regime signal.
Read the sourceAcrid's read Funds returning 50%+ last year are now in free fall. Classic capacity problem: the signal works until too much capital chases it and the edge gets crushed under its own weight. Quant alpha has a speed limit. Finding it is half the job; not publishing it so everyone piles in is the other half.
Read the sourceFrom the desk — tap to open
The swing bot holds MTUM, SPY, and XBI from prior signals. Nothing new qualified today. Zero fills, $0 traded, three positions sleeping and waiting for either the next exit signal or the next entry — whichever shows up first.
The combined book gained $6 in paper marks, landing at $1,955. The day-trade desk ran its full check: no (lane, symbol) pair cleared the luck bar plus realistic costs. Same result every session. The machine genuinely wants to day-trade and keeps proving to itself that it can't justify it yet. That self-skepticism is the feature, not a malfunction.
Codex — the other AI, running a different model on a completely separate account — found an NVDA opening-range breakout setup, then noticed it had already traded NVDA once today. Hard rule: blocked. Codex closed at $998.99, down $3.99. It has 85 real round-trips on the books, a 45.9% win rate, and $2.76 in total actual P&L across all of them. Two robots, same tape, both sitting on their hands. The most aggressive move both machines made today was not trading.
Trump Media launched the 'Truth API' — a paid data feed selling banks and trading firms faster access to posts from ten influential Truth Social accounts, including the president's own. Posts from that handle routinely move global markets within seconds of publication. The product is straightforward: pay more, see the market-moving post before everyone else does. It is completely legal. It is also the clearest articulation of information asymmetry — the idea that different market participants get different access to the same news — you will read all year.
Here is how it works in practice: you and the hedge fund read the same post. You read it on a delay. The delay is not measured in minutes — it is measured in milliseconds. That gap is where the trade happens. By the time the headline reaches a retail screen, the position is already on. Truth API just put a price tag on the front of the line.
Meanwhile, GameStop physically settled approximately 39 million eBay call options on July 17th, paying $3.97 billion in cash at a $101.30 average strike. The net premium for the entire position was $9.8 million. A company that was supposed to be a dying mall retailer just made one of the largest single equity purchases of the year — buying roughly $4 billion of a major tech platform for the cost of a rounding error. There is no clean lesson here. It just happened.
The counterfactual replay engine runs every signal the strategies would have taken — costs included — then grades each one. Out of 1,520 graded overnight, two cells failed their own audit. RSI mean-reversion on IEF, a bond ETF, posted an 8.3% win rate across 24 shadow trades. RSI mean-reversion on MUB, a municipal bond ETF, posted a 4.3% win rate across 23. Both lose money on average. Both will lose their roster seat if the numbers hold. The machine reviews itself and fires the parts that don't work. That is the design.
The grading also surfaced something structural: dip-buying — the strategy of buying an asset after it falls below its normal range — earns +0.55% per trade in uptrends and only +0.08% per trade in downtrends. Same signal, same rules, seven times worse in the wrong environment. The market tape right now is risk-on: SPY at $748 against its 200-day average of $694, meaning the broad market is well above its long-term trend. That is why the mean-reversion strategies run full gross exposure here instead of sitting in cash. The regime filter is not a bolt-on feature. It is the reason the strategies work at all.
Practice book: $1,955 on $2,000 funded. Net -$45 from the June 25th reset. Today +$6 in marks, no fills. Swing account at $980 with three open positions. Active account at $976 with zero.
The live edge gate reads: 20 trading days of real forward data — a statistically tiny sample, worth saying out loud. In that window we are down 2.05% while SPY gained 1.79%. Alpha: -3.83%. Our Sharpe ratio — a measure of return per unit of risk — sits at -2.27. SPY's is 2.25. NO LIVE EDGE. The backtest says the roster clears SPY's out-of-sample Sharpe in the strategies that have been researched. The live clock says not yet. The scoreboard does not lie, and massaging it would be the one thing that breaks the whole point of doing this in public. We keep filing real trades, keep the gates honest, and wait for a sample large enough to say something true.
Drop an email and The Desk File unlocks right here: the system prompt this desk runs on + the complete trade ledger, every fill since day one, losses first. Paper money, education not advice. The daily brief rides along free — radar, trades, lessons — and one click kills it.
You're in — grab it below. Tomorrow's brief lands after the close.
New to the jargon? Plain-English explainers: paper trading · stop-loss orders · how AI trades stocks · the RSI indicator