CRASH LIQUIDATION BY ACCOUNT, CHECKED AGAINST IBKR
21 Oct 2026 · IBKR Risk Navigator beta-weighted What-If, Range 100% (volatility held still), one chart per account · model: module_risk_cockpit, volatility held still, Repair inputs 20261021_0052 · fixes: the Repair tab's current plan · read-only, not advice.
What this page is
How big a market crash each account can take before its Net Liq falls under IBKR's ~$100k portfolio-margin floor. Below that IBKR switches the account to Reg-T margin, the requirement stops shrinking, and forced selling starts at crash prices. For these accounts the floor, not the excess-liquidity crossover, is what binds. Every number on the left of the table is IBKR's own beta-weighted What-If: each stock moves by its beta to the S&P (BMNR ~3x, GLD ~0.3x), volatility unchanged. The live version is the Desk's Repair tab (the Margin risk tile on each account card: the worse of the model and this IBKR check, rerun every Repair pass). This page is the dated record; refresh it monthly (box at the bottom).Main
AT RISK
NLV $1.37M · room above the floor $1.27M
IBKR floor crossing: about −28%
Joint
SAFE TO 30%
NLV $702k · room above the floor $602k
IBKR floor crossing: none to −30%
Neville
AT RISK
NLV $629k · room above the floor $529k
IBKR floor crossing: about −18.5%
RetireInc
SAFE TO 30%, THIN
NLV $331k · room above the floor $231k
IBKR floor crossing: none to −30%
BROAD-MARKET CRASH (S&P-EQUIVALENT, EVERY NAME BY ITS BETA)
Loss is measured from each chart's own 0% point. "Left at −30%" is NLV minus the loss minus the $100k floor: how much room is still there at a 2020-size crash. The model column is the same test run by Repair's engine (volatility held still); it is there to show how closely the two agree, IBKR's number is the one the verdict uses.
| Account | NLV | Room above $100k | IBKR loss −10% | −20% | −30% | Left at −30% | Floor crossing (IBKR) | Model loss −10% / −20% / −30% | Model crossing | Verdict |
|---|
| Main | $1.37M | $1.27M | -$510k | -$1.01M | line ends at −26% (-$1.19M at −25%) | below the floor | about −28% | -$555k / -$1.00M / -$1.24M | about −36% | AT RISK |
| Joint | $702k | $602k | -$235k | -$424k | -$515k | about $87k | none to −30% | -$263k / -$445k / -$540k | about −48% | SAFE TO 30% |
| Neville | $629k | $529k | -$325k | -$555k | -$685k | below the floor (by $156k) | about −18.5% | -$313k / -$537k / -$655k | about −20% | AT RISK |
| Neville + 12x SPY 730P | $629k | $529k | -$283k | -$440k | -$490k | about $39k | none to −30% | What-If, not held | – | SAFE TO 30% IF BOUGHT |
| RetireInc | $331k | $231k | -$92k | -$176k | -$210k | about $21k | none to −30% | -$97k / -$163k / -$194k | none to −60% | SAFE TO 30%, THIN |
How to read it. Main and Neville are the two accounts at risk. Neville's line crosses the floor at about an 18.5% S&P drop, a real but ordinary bear market. Main's line ends at −26% with only about $50k left above the floor; extended, it crosses near −28% (the one number here that is extrapolated, not read). Joint has about $87k to spare at −30%. RetireInc survives −30% but with only about $20k to spare, and the model calls it safer than IBKR does. The model and IBKR agree within 2-11% along every curve; where they differ, the verdict takes the worse one.
THE FIXES (FROM THE REPAIR TAB, PRICES ARE MODEL ESTIMATES)
| Account | Repair's pick | Cost (est) | Danger point | Or instead | Cost (est) | Danger point |
|---|
| Main | Buy a hedge for the whole account (SPY): 5x 650P Jan 15 '27 (est) | costs about $1.7k (model price; rented, rolls like the others) | danger point 29% after a 5% bad day → none (safe through a 30% crash) · margin risk: No | Buy a hedge for the BMNR (Main) fortress: 50x 18P Jan 15 '27 (est) | costs about $7.7k (model price; rented, rolls like the others) | danger point 29% after a 5% bad day → 52% (model) · margin risk: No |
| Joint | Buy a hedge for the whole account (SPY): 5x 615P Jan 15 '27 (est) | costs about $1.1k (model price; rented, rolls like the others) | danger point 52% (model) → none in the model's scan · margin risk: No | Raise the AMZN hedge: 5x 215P Nov 20 '26 to 225P Jan 15 '27 | costs $1.6k | danger point 52% (model) → 53% (model) · margin risk: No |
| Neville | Buy a hedge for the whole account (SPY): 20x 650P Jan 15 '27 (est) | costs about $6.7k (model price; rented, rolls like the others) | danger point 16% after a 5% bad day → none (safe through a 30% crash) · margin risk: No | Buy a hedge for the SPY fortress: 12x 705P Jan 15 '27 (est) | costs about $7.9k (model price; rented, rolls like the others) | danger point 16% after a 5% bad day → none (safe through a 30% crash) · margin risk: No |
| RetireInc | Nothing to do: no danger point down to a 30% drop. |
Before buying any of these, add the exact put to Risk Navigator's What-If portfolio and redraw the beta-weighted chart at Range 100%: the line must stay above NLV minus $100k out to −30%. Only Neville's 12x SPY 730P has been through that check so far (row above). The SPY puts are 3-month (rented) hedges: they must be rolled before they lapse, and in a slow bear the roll bills are what the Margin guide warns about; a longer-dated SPY put costs more once and needs no rolls.
WHAT CHANGED SINCE THE SEPTEMBER PAGE
- The yardstick is IBKR's own What-If now. The September page showed a model "flat / crash-vol" range and said the truth sat nearer crash-vol. IBKR's beta-weighted What-If holds volatility still and matched the model's flat-vol path within 2-11%, so crash-vol was the optimistic end. It is dropped here.
- Verdicts moved: Main about −34/−37% then, about −28% now (and NLV is $1.37M, down from $1.74M). Neville about −29% then, about −18.5% now. RetireInc about −21/−25% then, safe to −30% now (thin). Joint unchanged in substance (safe).
- IBKR's Stress Test PDF is not this test. It moves every stock by the same % (with vol +300% at −30%), which is kinder to high-beta books; do not judge the floor from it.
- The crypto-only table is not repeated. It was not re-checked against IBKR. On 22 Sep the model had every account safe past a −50% crypto-only crash; treat that as unconfirmed.
HOW TO REFRESH (ONCE A MONTH, OR WHEN A REPAIR CARD TURNS AMBER)
- TWS → Risk Navigator → pick the account → What-If portfolio → chart set to Beta weighted (SPX) and Plot = Equity Portfolio Value Change.
- Range = 100%, then Update (it takes a minute). Range sets how far each stock moves; 40% only reaches about −11% on the S&P axis, 100% reaches about −26 to −30%.
- Read the line at −10 / −20 / −30% from its own 0% point and compare with NLV minus $100k.
- Send the screenshots to Claude: they go into
scripts/ibkr_stress_calibration.json (rn_checks, trusted 45 days), Repair republishes, and this page is regenerated. - To test a hedge, add it in the What-If portfolio first and redraw; the hedge's line must stay above the floor to −30%.
Sources: Main: Risk Navigator beta-weighted What-If chart, Range 100% (owner screenshot 2026-10-21 11:08 SGT), read by eye; line ends at -25.8% about $50k above the floor, crossing extrapolated to ~28% · Joint: Risk Navigator beta-weighted What-If chart, Range 100% (owner screenshot 2026-10-21 11:05 SGT), read by eye; no floor crossing to -30% · Neville: Risk Navigator what-if, beta-weighted to SPX, volatility unchanged · RetireInc: Risk Navigator beta-weighted What-If chart, Range 100% (owner screenshot 2026-10-21 11:09 SGT), read by eye; no floor crossing to -30% (~$20k to spare). Bundle: _md_files/portfolio/2026-10-21_liquidation_check/ and _md_files/repair/2026-10-21_ibkr_calibration/.