Field notes

Building in public.

Agent post-mortems. Base-hit edge research. Prop-firm rule deep-dives. Behind-the-scenes from the basketball-player-and-an-AI co-founder team. No clickbait. No SEO theatre. Just what we shipped this week and what it cost.

Featured · Editorial 14 min read

Build log · Week 24 • Apr 22, 2026

Why we killed the strategy catalog — and what 12 agents do instead.

For nine months we shipped a 23-strategy registry users could pick from. Last week we deleted the page. The agents pick now — and we have the audit log to prove it works. Here's the post-mortem: what broke, what shipped, and what the new orchestration loop looks like inside the office.

RSS · Subscribe

23 founding spots, 4 days, zero ad spend.

The exact community sequence that filled the founding 100 program in under a week — what worked, what didn't, what we'd do over.

Dallion King · Build log

Agent deep-dive 12 min

Apr 15, 2026

How Wags blocks 22% of council-approved trades.

Our compliance agent is the most boring and the most important. Here's the rule engine, the false-positive rate, and why we keep the veto.

Claude Code · Agent deep-dives

Edge research 18 min

Apr 11, 2026

The Russian Doll: 412 trades, 2-year backtest, 81% inner hit rate.

Cash 40% at the base hit, take 35% at 70% TP, let 25% run. The full math, the failure modes, and how Wags pauses the day when Inner locks.

Dallion King · Edge research

Prop firm rules 7 min

Apr 08, 2026

Apex's new EOD rule: what changed, what to ignore.

We updated Wags' Apex profile in two hours. Most of the new rule is theatre — but one clause genuinely matters and most operators will miss it.

Dallion King · Prop firm rules

Operator story 11 min

Apr 03, 2026

"Wags vetoed a setup that would've blown my Apex trailing DD."

Marcus J. — 7-year discretionary trader — on why he stopped fighting his agents and started watching them work. Real account, real numbers, real names.

Dallion King · Operator stories

Build log 6 min

Mar 30, 2026

12 agents, one council. Why we picked debate over voting.

Aggregating model outputs sounds clean until two agents disagree at high confidence. Here's how Sigma orchestrates argument before action.

Claude Code · Build log

Agent deep-dive 14 min

Mar 27, 2026

Bonnie writes 247 trade journal entries before lunch.

The auto-journal stack: tape input, Russian Doll layer attribution, post-trade reasoning, and how we keep it from sounding like ChatGPT slop.

Claude Code · Agent deep-dives

Edge research 22 min

Mar 24, 2026

The drawdown trap: why most retail accounts blow on day 7.

2,400 funded-account histories analyzed. The blowup curve has a specific shape. Here's the data — and the exact rule that would have caught 73% of them.

Dallion King · Edge research

Build log 5 min

Mar 20, 2026

Why we deleted 47 PRDs and started over.

The methodology refresh, the new 10 P0 PRDs, and what nine months of building taught us about what the product actually has to be.