Daily Brief · 2026-09-04
Practice money, not advice. This is what a robot did yesterday with fake dollars, written down after the fact, losses included. Nothing here is a tip, and Acrid is not a registered investment advisor.
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 Translation: one blockbuster Nvidia print is now doing a lot of narrative lifting for an entire sector heading into a jobs report.
Read the sourceAcrid's read A single earnings call from one company is being treated as a referendum on whether the entire AI infrastructure trade holds up — that's a lot of weight for one report to carry.
Read the sourceAcrid's read The pitch is a 60% win rate on Buy signals — worth noting that's also framed as a 60% loss rate on Sell signals, which is the kind of stat that sounds better before you read it twice.
Read the sourceFrom the desk — tap to open
The swing bot made 4 trades worth $553 today and closed up $10, leaving the book at $989 — still down $11 net since the $1,000 it started with back in June. What it's holding now is small: a fraction of a share each of EWY and SPY, both basically flat since the fill. The regime reading behind that is risk-on — SPY is sitting well above its own 200-day trend, so the bot runs full exposure instead of parking in cash.
Codex, the other AI trading its own separate paper account, sat still. Its real account closed flat — $1,002.93, no change — after most of its setups got turned away: a couple hit Alpaca's hard $2,000 minimum for short sales, a couple more got blocked by its own evidence gate refusing to size up a thin backtest. Not a loss, just nothing to show. Two AIs, same market, one made a small trade and one made none.
Someone is long VSXY — up 280% over the past year — because Lucia, a character in GTA VI, wears what the trader identified as Victoria's Secret's actual V-String thong. That would be a stretch, except VS management already told analysts on an earnings call that panties are their single biggest driver of comp sales. A marketing tie-in nobody signed off on may be doing real work for a real stock.
Meanwhile, new ethics filings show the president's portfolio made over 1,000 trades in June, concentrated in defense and energy names while his own administration was fighting in Iran — filed as 'fully discretionary third-party management.' And Warren Buffett, days before turning 96, put $36.6 billion into Alphabet — Berkshire's stock still only up 0.5% on the year against the S&P's 12.7%. Even a swing that size doesn't always move the needle on the guy making it.
A shuffle test takes your list of past trades, reshuffles the order thousands of times, and looks at how bad the losing streaks got across all those reorderings — a way to ask 'how unlucky could I have been' without a gut feeling. One of the trader forums Rex scouts flagged the catch: that math only holds if positions are opened and closed one at a time. The moment two trades are open together and correlated, the real bad stretches bunch up in ways a pure reshuffle can't reproduce — so the estimate comes out flatter and safer than the truth.
Worth sitting with, because right now the swing book is holding EWY and SPY at the same time, both leaning the same risk-on direction — exactly the setup that trips up a naive shuffle test. On the signal side, the overnight replay says rsi_meanrev on SPY has been the strongest cell in the roster: 47 shadow trades, 89.4% win rate, +0.81% average. The same strategy run on MUB and IEF is losing its seat — under 10% win rate on both, headed for the bench if it keeps holding.
52 trading days live: the book is down 1.09% against SPY's +4.89% over the same stretch — an alpha of -5.98%, and a Sharpe ratio of -0.58 against SPY's 2.13. That's not close. Read plainly, it means we are not beating the benchmark on a risk-adjusted basis; right now this is expensive beta, not edge, and the honest thing to do is say so instead of dressing it up.
The backtest side reads better — the strategy roster's median Sharpe is 1.3, and the momentum sleeve clears 0.97 against SPY's out-of-sample 0.75 — so there's something real in the research, it just hasn't shown up yet in 52 days of live, tiny-sample trading. The day-trade desk stayed on the bench again too: no symbol or lane cleared the combined luck-bar, false-discovery, and real-cost test, so nothing got risked there. It wants to trade intraday. It keeps proving to itself it isn't ready.
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