At 6:32 a.m. ET on August 25, 2026, our engine sealed its morning run. One market-regime score, computed once, from one set of inputs, shared by all four of its virtual strategies. The dial read risk-on — a constructive 28.8 on a scale that runs to ±100. Then the four books published. Sentinel, the conservative strategy, put 60.3% of its virtual book to work. Meridian, the balanced core, 69.6%. Regent, the sector-basket strategy, 81.9%. And Maverick — the aggressive strategy, the one built to run the hottest — sealed the most defensive book of the four: 28.7% deployed, its entire stock exposure sitting in exactly two names. Same morning. Same data. Same dial. A 53-point disagreement. This piece is about where that dispersion comes from — the part we designed, and the more interesting part we didn't.
One dial, thirty points of designed disagreement
Start with what the four strategies share. The regime dial is deliberately simple: six pre-registered factors with fixed formulas, no discretion, re-scored every morning. On August 25 it read like a calm tape — VIX at 15.8 (low fear, +12.8 points), a 10Y–2Y Treasury curve at +0.46pp and steepening (+13.8), Fear & Greed at 55 (+2.3), the S&P 500 riding +8.2% above its 200-day average (+4.1), offset only by the prior session's −0.29% dip (−0.9) with no high-volume selloff detected. Composite: 28.8. Risk-on.
Each strategy then converts that one number into a target exposure through its own pre-registered reaction curve — a fixed translation of temperament into arithmetic. A conservative curve de-risks early and re-risks late; an aggressive curve does the opposite. That morning the same 28.8 became four different targets: Sentinel 60.3%, Meridian 76.9%, Regent 81.9%, Maverick 90.2%. Thirty points of spread, and every point of it designed — you could compute those four numbers from the score alone with a lookup table. If the story ended there, "four strategies" would be a parlor trick: one dial wearing four costumes. The sealed books tell a different story.
The four books on August 25
Two of the books landed exactly on their targets, and their construction explains why. Sentinel spreads small weights across a wide book — 27 positions, none above its 12.4% gold sleeve, which that morning was its largest single line and its clearest defensive posture in weeks. Regent holds a structural basket of sector champions — 33 names, capped small, led by RTX at just 5.1%. A diversified book can always fill its target: many small positions, each easy to fund. The two conviction-driven books are different animals. Meridian came in seven points light at 69.6%, leaning hard-asset — copper miner Freeport-McMoRan and gold miner Newmont at 9.9% each — layered with defensive quality like Eli Lilly and Northrop Grumman at 7.7%. And Maverick held two stocks: Freeport and Newmont at 12.1% each, plus a 4.5% gold sleeve. That's the whole book. Target 90.2%, funded 28.7%.
| Strategy | Funded | Target | Positions | Largest line |
|---|---|---|---|---|
| Regent · Sector Champion | 81.9% | 81.9% | 33 | RTX 5.1% |
| Meridian · Core | 69.6% | 76.9% | 9 | FCX · NEM 9.9% each |
| Sentinel · Conservative | 60.3% | 60.3% | 27 | GLD 12.4% |
| Maverick · Aggressive | 28.7% | 90.2% | 3 | FCX · NEM 12.1% each |
Where Maverick's missing 61 points went
The regime target is top-down. Deployment is bottom-up — and nothing gets funded by default. Every position has to earn its weight through conviction- and volatility-scaled sizing under a half-Kelly cap, and each strategy sets its own bar: Maverick demands the highest conviction and concentrates hardest. On mornings when many names clear its bar, the book fills toward target — through mid-August Maverick ran above 90% deployed, the hottest book of the four. When few names clear it, the book shrinks, whatever the target says. Cash isn't a view; it's a residual.
You can watch it happen across four consecutive sealed mornings. August 20: 93.0% deployed, an energy sweep — five energy names led the book, with two at 12.5% each. August 21: 77.8%, the book flipped almost entirely into semiconductors and software — six names at 12.2% apiece. August 24: 52.8%, down to a defensive quartet: a tower REIT, a pharma, a steelmaker, a logistics landlord. August 25: 28.7% — two miners and gold. As semiconductors sold off into a loaded catalyst week — Nvidia earnings, Jackson Hole, the PCE print — fewer and fewer names cleared the conviction bar, until only the hard-asset trade was left standing. The regime dial never blinked. The stock-level evidence under it thinned out day by day, and the book followed the evidence.
We know the obvious "fix": force deployment up to the target and make the lines agree. We built exactly that — a minimum-deploy band — and ran it through our pre-registered 20-year backtest gate before letting it near the live engine. The result: about +1.9 points a year of extra return for the core and aggressive books, upside capture better by 7–8 points — and maximum drawdown deeper in every stress window we test, including 2020. The gate's pre-registered rule says a fix that buys return with deeper drawdowns fails. So we killed it, and the gap stays — published every morning, target next to funded book, whether or not it flatters us.
Dispersion is a readout, not a malfunction
A single-model shop publishes one opinion and calls it conviction. Four books on one dial give you a second axis of information: how much the evidence agrees with itself. When the four lines cluster — as they did for most of July and mid-August — the macro dial and the stock-level signals are telling the same story at every risk appetite. When they split 53 points wide, the widest in the window charted above, something more interesting is true: the same market is reading as calm at the top and thin underneath. Our own morning report put it plainly that day — with a catalyst-heavy week ahead, the strategies expressed disagreement through position size, not sector rotation. The size of the bet, not its direction, was where they diverged.
We are not saying the defensive books were right. By the time you read this, Nvidia will have reported and the week will have graded itself — that's what a sealed record is for. Maverick's collapse to 28.7% may look prescient in hindsight, or it may look like an aggressive strategy that got shaken out of a rising tape by its own bar. Both outcomes go in the ledger. The point of publishing the dispersion is that you get to see the disagreement before the resolution, every morning, with a hash on it.
1. One shared regime dial, four pre-registered reaction curves: on August 25 the same score of 28.8 became targets of 60.3% to 90.2% — a thirty-point spread of pure, designed temperament.
2. Deployment is earned bottom-up — every position must clear conviction- and volatility-scaled sizing under a half-Kelly cap — so funded books can disagree far more than targets do: 28.7% to 81.9% that morning, with the aggressive book the most defensive. The target-versus-funded gap is published daily; the fix that would have closed it failed our 20-year drawdown gate and was killed.
3. Dispersion between the books is a readout of how much the evidence agrees with itself — a pattern to observe, not a prediction of which book wins the week. We don't know either; that's what the sealed record measures.
Informational and entertainment content only. Not investment advice. All exposure figures, targets, holdings and regime factor readings come from Alphixir's sealed public daily record — virtual portfolios, no real money. Company names appear as factual descriptions of sealed virtual holdings, not as recommendations. Nothing here is a view on where any stock or the market is going.