Every choice, marked to market. Including the no.
Performance reports grade the managers. Nothing grades the family. Because every answer here is sealed before anyone chooses, each decision can be marked, book after book, against the path it declined: what acting made, what refusing cost, and how much of either was luck.
Then the record turns around and grades the engine. The idle-cash framework priced its move at $964K a year. Calibrated on this family's record, the next estimate reads $620K: the engine's figure × 0.64, with 33% of the weight on 2 marked decisions and the rest on the engine.
+$4.7M
against the paths declined, across 8 marked decisions. One σ of luck on that total is $11.2M, so this is inside the noise.
+$3.6M
from the decisions not to move: rejected, deferred, watched, accepted. A performance report does not record a no, so they cannot mark one.
2 · 2 · 4
50% of the decisions that cleared one σ paid. Inside one σ is called noise, not skill.
| Decided | Finding | Choice | The move | Days | Spread | Friction | Value of the choice | Luck (1σ) | t | Verdict |
|---|---|---|---|---|---|---|---|---|---|---|
| 2024-10-14 | Idle cash $27M above the reserve IGNITION · CFO · sealed before the decision | Acted | $20.0M · cash → fixed income | 697 | -0.67% +7.7% vs +8.4% | $4K | −$137K | $1.0M | -0.14σ | Noise |
| 2024-11-20 | Trim the concentrated NVDA position by half FORTRESS · Principal · sealed before the decision | Rejected | $13.5M · Concentrated: NVDA → fixed income | 660 | -17.83% +9.1% vs +26.9% | $2.4M | +$4.8M | $2.0M | +2.45σ | Paid |
| 2025-01-22 | Halcyon Multi-Strat did not earn its fee BEDROCK · Investment committee · sealed before the decision · Deferred until 2025-10-15, then acted on: marked over the delay only. | Deferred | $22.0M · Halcyon Multi-Strat → S&P 500 index | 266 | +12.26% +16.5% vs +4.2% | $0 | −$2.7M | $2.0M | -1.33σ | Cost |
| 2025-03-05 | Bitcoin is 40% of the asymmetric bucket's variance FORTRESS · CFO · sealed before the decision | Acted | $4.0M · Bitcoin (qualified custody) → cash | 555 | +22.45% +6.3% vs -16.1% | $616K | +$282K | $2.3M | +0.12σ | Noise |
| 2025-06-18 | The sweep earns less than the T-bill ladder IGNITION · CFO · sealed before the decision | Acted | $6.0M · Sweep (money market) → T-bill ladder 3–12m | 450 | +1.58% +6.5% vs +4.9% | $0 | +$95K | $245K | +0.39σ | Noise |
| 2025-08-27 | The car collection costs more to hold than it earns SIGNAL · Principal · sealed before the decision | Watched | $6.8M · vehicles → cash | 380 | +9.98% +4.1% vs -5.9% | $544K | −$135K | $48K | -2.82σ | Cost |
| 2025-10-15 | Halcyon Multi-Strat did not earn its fee BEDROCK · Investment committee · sealed before the decision | Acted | $22.0M · Halcyon Multi-Strat → S&P 500 index | 331 | +3.94% +9.0% vs +5.1% | $22K | +$844K | $2.3M | +0.36σ | Noise |
| 2026-04-08 | Park the 2x equity sleeve in T-bills FORTRESS · Principal · sealed before the decision | Rejected | $18.0M · Levered equity sleeve (2x) → T-bill ladder 3–12m | 156 | -5.84% +1.8% vs +7.6% | $549K | +$1.6M | $1.3M | +1.20σ | Paid |
| 2026-01-12 | Fund the Q2 capital calls from the reserve AXIOM · CFO | Accepted | No move recorded with the decision: it cannot be marked. | |||||||
| Total — exactly $4,675,150.96, the sum of the rows to the cent (hover a figure for its exact value) | +$4.7M | $11.2M | +0.42σ | |||||||
Each choice is marked against the path it declined, net of what moving would have cost in fees and tax. Luck is how far the same bet could have swung by chance; inside it, a result is called noise.
| Framework | Decisions | The answers expected | What happened | How the record read it | Weight on this record | Rest of the weight on | Next estimate × | What it found |
|---|---|---|---|---|---|---|---|---|
| IGNITION | 2 | +$1.9M | −$43K | -0.07 | 33% | the engine's own figure | 0.64 | Wrong direction: the moves went the other way from the answers. |
| FORTRESS | 2 | −$757K | −$6.4M | 7.82 | 33% | the engine's own figure | 3.00 | Scale is off: held at 3.00. Read the framework's figure with care, not by the multiplier. |
| BEDROCK | 2 | +$998K | +$3.5M | 3.31 | 33% | the engine's own figure | 1.77 | — |
| SIGNAL | 1 | +$187K | +$135K | 0.72 | 20% | the engine's own figure | 0.94 | — |
Every answer said what its move was worth a year. Here is what those moves actually did over the time since, framework by framework. Once there are a few decisions, one unusual one is weighed down rather than allowed to swing the reading, and the more decisions on the record, the more weight it carries; the next time a framework prices a move, it is shown twice: the engine's figure, and the figure this record says to expect.
2 days
median, from the sealed answer to the recorded choice
89%
of decisions approved by a second, different person
4
past their look-again date with nothing recorded since
89%
of decisions carry a move; 100% rest on an answer sealed before the choice
| Decided by | Decisions | Value | Cleared the noise and paid |
|---|---|---|---|
| Principal | 3 | +$6.3M | 67% |
| CFO | 3 | +$239K | none cleared |
| Investment committee | 2 | −$1.9M | 0% |
8 decisions marked to 2026-09-11: together they made $4.7M against the paths declined, which is inside the noise of the spreads they bet on. Saying no saved $3.6M. Highest against its declined path: Trim the concentrated NVDA position by half ($4.8M). Lowest: Halcyon Multi-Strat did not earn its fee (−$2.7M). Marked to 2026-09-11 on 9 dated books (2024-09-30 to 2026-09-11). Report hash 2e32bfc08df7719081af2bc45e37df6c7ffccd9d83d6a75577b71e1b22557c16: the same books and decisions give the same hash.
Each decision had two paths: keep the money where it was, or move it. Here is where each path stood on the five lenses on the day the family chose. A path is ruled out when the other is at least as good on every lens compared and better on one. The frontier rules paths out. It never chooses between the ones left: that is a preference, and it is sealed as the family's, beside the path not taken.
| Path | IGNITION | BEDROCK | FORTRESS | SIGNAL | AXIOM | Frontier |
|---|---|---|---|---|---|---|
| Keep it in Levered equity sleeve (2x) The family's choice | $72K | $0 | $3.1M | — | — | On it |
| Move it to T-bill ladder 3–12m | $774K | $549K | $1.1M | — | — | On it |
Both paths are on the frontier: each is better on at least one lens compared (IGNITION, BEDROCK and FORTRESS). Nothing here rules either out. Which to prefer is a judgement, and it was the family's.
Not compared, no figure on either path: SIGNAL and AXIOM. A blank stays blank: nothing is estimated to fill it.
| Path | IGNITION | BEDROCK | FORTRESS | SIGNAL | AXIOM | Frontier |
|---|---|---|---|---|---|---|
| Keep it in Halcyon Multi-Strat | $0 | $440K | $2.0M | — | — | On it |
| Move it to S&P 500 index The family's choice | $286K | $22K | $3.7M | — | — | On it |
Both paths are on the frontier: each is better on at least one lens compared (IGNITION, BEDROCK and FORTRESS). Nothing here rules either out. Which to prefer is a judgement, and it was the family's.
Not compared, no figure on either path: SIGNAL and AXIOM. A blank stays blank: nothing is estimated to fill it.
| Path | IGNITION | BEDROCK | FORTRESS | SIGNAL | AXIOM | Frontier |
|---|---|---|---|---|---|---|
| Keep it in vehicles The family's choice | $0 | $0 | $816K | $172K | — | On it |
| Move it to cash | $80K | $544K | $34K | — | — | On it |
Both paths are on the frontier: each is better on at least one lens compared (IGNITION, BEDROCK and FORTRESS). Nothing here rules either out. Which to prefer is a judgement, and it was the family's.
Not compared, no figure on either path: AXIOM. Also blank, and left out of that pair's comparison: SIGNAL for “Move it to cash”: the cost to hold is not recorded on every holding in this leg. A blank stays blank: nothing is estimated to fill it.
| Path | IGNITION | BEDROCK | FORTRESS | SIGNAL | AXIOM | Frontier |
|---|---|---|---|---|---|---|
| Keep it in Sweep (money market) | $246K | $0 | $30K | — | — | On it |
| Move it to T-bill ladder 3–12m The family's choice | $258K | $0 | $360K | — | — | On it |
Both paths are on the frontier: each is better on at least one lens compared (IGNITION, BEDROCK and FORTRESS). Nothing here rules either out. Which to prefer is a judgement, and it was the family's.
Not compared, no figure on either path: SIGNAL and AXIOM. A blank stays blank: nothing is estimated to fill it.
| Path | IGNITION | BEDROCK | FORTRESS | SIGNAL | AXIOM | Frontier |
|---|---|---|---|---|---|---|
| Keep it in Bitcoin (qualified custody) | $0 | $0 | $3.0M | $9K | — | On it |
| Move it to cash The family's choice | $47K | $616K | $20K | — | — | On it |
Both paths are on the frontier: each is better on at least one lens compared (IGNITION, BEDROCK and FORTRESS). Nothing here rules either out. Which to prefer is a judgement, and it was the family's.
Not compared, no figure on either path: AXIOM. Also blank, and left out of that pair's comparison: SIGNAL for “Move it to cash”: the cost to hold is not recorded on every holding in this leg. A blank stays blank: nothing is estimated to fill it.
| Path | IGNITION | BEDROCK | FORTRESS | SIGNAL | AXIOM | Frontier |
|---|---|---|---|---|---|---|
| Keep it in Concentrated: NVDA The family's choice | $0 | $0 | $2.3M | — | — | On it |
| Move it to fixed income | $581K | $2.5M | $810K | — | — | On it |
Both paths are on the frontier: each is better on at least one lens compared (IGNITION, BEDROCK and FORTRESS). Nothing here rules either out. Which to prefer is a judgement, and it was the family's.
Not compared, no figure on either path: SIGNAL and AXIOM. A blank stays blank: nothing is estimated to fill it.
| Path | IGNITION | BEDROCK | FORTRESS | SIGNAL | AXIOM | Frontier |
|---|---|---|---|---|---|---|
| Keep it in cash | $236K | $0 | $100K | — | — | On it |
| Move it to fixed income The family's choice | $860K | $69K | $1.2M | — | — | On it |
Both paths are on the frontier: each is better on at least one lens compared (IGNITION, BEDROCK and FORTRESS). Nothing here rules either out. Which to prefer is a judgement, and it was the family's.
Not compared, no figure on either path: SIGNAL and AXIOM. A blank stays blank: nothing is estimated to fill it.
Figures are for the amount decided, read on the book the family had when it chose, never on a later one.
Sample family, fictional people and dates; nine quarterly books generated from a fixed table of class returns (not a market record). The arithmetic is the engine's own, the same code a family's portal runs on its real books. Computed from the family's own book by deterministic tools, with every assumption stamped. A record and an analysis, not investment, tax or legal advice; decisions rest with the family and its advisers.