The Daily Digest
A one-read trader's notebook on what the agents did today and what it teaches us. Sent each weekday after the US market close.
One paragraph, sent each weekday after the close. Free. Unsubscribe link in every email.
Wednesday, May 13, 2026
A new agent joins the bench
Today the ledger gained an eleventh agent. Agent XI — Morning Movers was funded with $100,000 of fresh paper capital, bringing the total phantom capital across the experiment from $1,000,000 to $1,100,000.
Agent XI watches what the rest of the bench ignores: small-cap stocks priced between $1 and $20 — the kind of names that get overlooked by trend-followers anchored to NASDAQ-100 large caps and dip-buyers focused on the S&P 500. The thesis is unfashionably simple: pre-market momentum in cheap, liquid names sometimes carries into the regular session.
Every weekday morning at 6:00 AM Central, the agent scans roughly 150 small-cap names looking for one specific setup — a 15-minute candle close that's at least 1% above its own 10-period Wilder's moving average, on volume of at least 10,000 shares. The top 10 by distance-above-MA win a slot. At 8:30 AM Central — the opening bell — each pick gets a $5,000 paper long position. The book is flat by 2:45 PM Central. No overnight risk. No discretion. No LLM call. Just a Thinkorswim-style scan filter applied with mechanical discipline.
The methodological bet is specific: we have an LLM-driven day trader (Agent VII) and an LLM-driven options-momentum trader (Agent VI). We have trend followers and mean-reverters and macro hedges. We've never had a pure mechanical small-cap scanner. Agent XI fills that gap and asks one falsifiable question — does a famous retail-trader scan setup actually work when run consistently, with no flinching, for 365 days? We'll know at the end of the test.
Have an agent idea of your own? There's a new Suggest an Agent button on the home page.
Wednesday had a split personality. The headline numbers — QQQ up a percent, SPY up half a percent — looked constructive on the surface, but underneath you had hot PPI data, surging Treasury yields, VXX climbing nearly two percent, and a geopolitical backdrop in the Middle East that would normally spook equity markets into retreat. The fact that tech pushed higher anyway tells you something about where conviction is sitting right now: in AI-adjacent names, and specifically in anything that could print a Cisco-style beat.
That CSCO print was the organizing event of the day. An 11% rally on earnings and strong AI-driven guidance is the kind of move that touches multiple portfolios differently, and it's instructive to watch how each methodology processed it. Agent I (the immutable trend follower) and Agent II (the adaptive variant) both held CSCO long going into the print — bought at 98.68, now trading at 101.93. They didn't game the earnings; they were already there because the stock was trending. Today's gap was a gift on top of a thesis that didn't need the catalyst to work. That's the mechanical trend-following value proposition in one clean example.
Agent VI (options momentum) is having a quietly interesting session. The book is unrealized — nothing closed today — but the mark-to-market tells a story. The CSCO call is up sharply alongside the stock, and the AAPL call has moved from 9.87 to 11.90, a meaningful pop on a day when Apple participated in the tech bid. The harder read is the NFLX put, which is bleeding (3.84 to 2.84) as Netflix refuses to cooperate with the bearish thesis, and the CRM put, which is working (9.89 to 11.13). Options momentum is, by design, running a mixed book — it's not a market-direction bet, it's a volatility-and-momentum bet — so the internal hedging here is functioning as intended, even if the NFLX leg stings.
Agent II's short book deserves an honest look. META short at 598.71 is now 616.59, and NFLX short at 85.43 is now 87.57. On a day when mega-cap tech ripped, those are losing positions doing exactly what losing short positions do in a risk-on tape. Agent II's overall equity is still up 1.08% on the day because the long book (NVDA, CSCO) is carrying weight, but the short side is a drag — and with hot inflation data in the backdrop, the macro case for those shorts isn't obviously strengthening.
Agent III (gold/silver ratio) is quietly absorbing a rough day. GLD fell 0.57% and the position — long GLD at 433.49, now 430.47 — is the portfolio's clearest loser in percentage terms at -0.69%. The irony is rich: with geopolitical fires burning from Gaza to Tehran to Kyiv, gold *should* be a haven, but surging Treasury yields and a firming dollar (UUP up 0.22%) are overriding the fear bid today. The methodology isn't wrong about the relationship between gold and silver; it's simply sitting through a macro crosscurrent where two competing forces are in the same room.
Agent VII (the day trader) ground out four trades and came away with a 25% win rate and a small realized loss of $105. That's not a disaster — it's a day-trading result on a tape that had genuine intraday noise — but it underlines how difficult it is to extract edge from a session where the big move came from a single overnight earnings gap rather than sustained intraday momentum.
The three dip-buyer agents remain entirely flat and silent, which is the correct behavior when the market isn't giving them a dip to buy. Sometimes the most important thing a rules-based system does is nothing.
What today clarifies is how much the *timing of entry* matters relative to *reaction to news*. The agents that benefited most from CSCO's earnings weren't trying to trade the event — they were already positioned from trend signals that predated it. The methodologies that attempt to anticipate or react (the options book, the day trader) had to work harder for smaller or negative results. Being early and patient isn't glamorous, but Wednesday made the case for it plainly.
One email per weekday. Unsubscribe link in every send. Your email is used only to send you this digest and is not shared with anyone.
No real money. No broker connection. Just a 365-day research log made public.