As I alluded to earlier today, I've spent the last two days working on an exciting feature that captures an important nuance in Bitcoin treasury management.
The idea is simple. Every Bitcoin treasury company runs some amount of amplification (leverage) — primarily preferred stock or convertible debt — against its BTC. More amplification means more BTC per share on the way up. It also means the common equity gets wiped out faster on the way down.
Until now, the Monster Model let you set a constant amplification target and watch what happened as the dynamic treasury management engine attempted to steer either company towards its target under the conditions set by your other parameters (BTC projection path,
$STRC /
$SATA growth path, USD management, etc.).
But that didn't prevent the modeled company from drifting towards a balance sheet whose equity is wiped out in a crash.
Now you can pick a share price you want
$MSTR or
$ASST to survive down to, and the model solves for the most amplification the company could carry if Bitcoin fell and mNAV collapsed.
I call it the defensive amplification target, and it's now fully integrated into the dynamic treasury engine. If the system can run nominally, it does; but if the balance sheet gets stressed according to the parameters you specify, it sounds the alarm and begins taking decisive action.
The feature calls for you to give it three things: (1) a share price to defend; (2) an mNAV to assume in the crash; and (3) how far Bitcoin might fall (as a fraction of its power law trend). The model works backward to the maximum liability load that still clears your defensive price, converts that to an implied amplification target, and pulls the company's target amplification down to that implied target whenever your own constant target would be higher.
The new feature is off by default, so if you want to experiment with it, you'll need to toggle its checkbox, which you can find in the "Defensive target" fieldset in the Amplification section of the parameter pane.
Two things I learned while building it that I didn't expect.
→ The first: capping the target isn't enough. The model doesn't hold amplification at the target — it accepts anything inside a tolerance band around the target, and only acts decisively to delever when amplification exceeds that band. So the target can read as being exactly where you set it while actual amplification sits at the band's upper edge, above your limit. The defense mechanism has to bind the upper edge, not the center; otherwise the width of the band is amplification you never authorized. At the sensitivities involved, a tolerance of 0.01 (one percentage point) in amplification ratio is worth about a dollar of share price — so your $80 floor quietly becomes a $79 floor.
→ The second: which dollars count. The model tracks "effective USD" — an internal metric inspired by the extension of the model (which was originally
@Strategy-only) to
@Strive. The Monster Model defines effective USD as all USD assets plus a user-specified haircut on any third-party perpetual preferreds the company holds. That's fine for normal reporting. But as we saw this summer, perpetual preferred equities are least saleable exactly when the common equity is under the most stress, which is the scenario you're trying to defend against. So the defensive calculation uses pure USD while everything else keeps the effective figure. Two different questions, two different answers. (If you don't like the idea of valuing PPE above $0 even during normal operation, feel free to change that parameter; it's called "Fraction PPE counted in Effective USD" and can be found in Strive's "USD and PPE Assets" section in the parameter pane.)
If I'm being totally honest, this new feature is just absolutely sick. Watching the dynamic treasury management logic simultaneously dial back on preferred issuance while ramping up use of the common equity ATM to delever so that it can defend your chosen share price at your chosen power law level is an almost religious experience. This is reflexivity in motion, and highlights the power of modeling treasury companies (and Bitcoin itself) with a non-linear dynamical system.
Also shipped:
→ A break-even mNAV basis toggle. Break-even mNAV marks where issuing common stops being accretive — but "accretive" has two meanings. Gross BTC per share is the one BTC Yield measures, and it ignores liabilities entirely. Net BTC per share counts only what's left after senior claims. Those give different answers, and the model was silently picking one based on which mNAV variant you were looking at. Now the user gets to choose, and the choice is always explicit. This is what I called a semantic bugfix in my earlier post from this morning.
→ As for the technical bugfix I alluded to — that turned out to be a false alarm stemming from Opus's misunderstanding of some of the nuances of the model's mNAV calculations. Once it understood, the issue became clearer, and the LLM agreed that there was only ever a semantic ambiguity.
→ The power law reference curves moved from +/- 40% to +/- 50% of trend, which is a more honest benchmark after the drawdown we just lived through.
→ And a batch of quieter fixes: terminology standardized across the parameter and series documentation, so "selected liabilities" and "USD assets" mean one thing each instead of three.
The new changes are live at
monstermodels.live. Please let me know if you encounter any issues, and as always, thank you for your support. 🧡
$MSTR $ASST $STRC $SATA $BTC #Bitcoin