Buying the next bitcoin cycle through MSTX calls

Bitcoin has run in four-year cycles, and each climb has been smaller than the one before it. If that carries on, the run beginning at this cycle's bottom is worth about 3.94x, which puts bitcoin near $332,588 in the second half of 2029.

This report sets out a position built on that view: what we would buy, what it costs, what it pays if the view is right, and the several ways it does not work. The instrument is call options on MSTX, a fund that targets twice MSTR's daily move. That choice buys enormous convexity to the cycle and charges an unusually high rent for it, and most of the report is about the rent.

The position

Four call options, one on each of four expiries, for $47,920 against a $50,000 budget. Whole contracts, and only strikes with a real book behind them.

If bitcoin follows the projected path, that position is worth $527,389 by 2029-01-19, or 11x what it cost.

Spread over four dates rather than concentrated in the single best-scoring contract, for two reasons. The cycle's timing is the weakest part of the view, so a position that needs one particular quarter to deliver is a bet on the calendar as much as on bitcoin. And the contract that scores best on expected value is reliably the one with the most tail and the least open interest, which is a number you can compute but not buy.

Why we think bitcoin goes there

Three complete cycles sit in the price record. Each one bottomed, climbed for roughly 35 months, and topped:

The last row is the claim. Here it is drawn, with the three cycles that argue for it:

The solid line is what bitcoin did; the dashed line is that last row. The last top was 2025-10-06 and down legs have run about a year, which puts this cycle's bottom at 2026-10-06. Until that date passes the live price stands in for the bottom, so the run starts from $84,348 rather than a figure typed in weeks ago.

Now the arithmetic that produces the dashed line, because it is the only part of this worth arguing with.

The multiples fall fast and unevenly: 109.5x, then 20.9x, then 7.9x. Each is roughly a fifth of the one before, then roughly a third. There is no obvious next term in that sequence, which is why the raw multiples are no use for forecasting.

Take logarithms and the same three numbers become 4.7, 3.04 and 2.07. Now the pattern is plain: each one is about two thirds of the last, 0.65 and then 0.68. That is the regularity the whole report rests on, and it is worth being clear that it is a regularity in the logs and nowhere else.

Carry it one more step. 2.07 times 0.66 is 1.37, and e to the 1.37 is 3.94. So the coming run is projected to multiply the bottom by 3.94, which from $84,348 is $332,588.

Said plainly: each cycle has delivered about two thirds of the previous one's log return, three times running. A fourth cycle obeying the same rule is a triple, not another ten-bagger. The position below is built for a triple.

The shape of the projected climb is last cycle's own path, smoothed. Replaying it day for day would assert that a correction lands on the same date as last time, which is one draw from a distribution rather than a property of cycles.

Run the cursor along it and the chart reports the whole chain at that date: bitcoin, what MSTR is worth at the multiple we assume, and what MSTX is worth after the leverage and the drag have taken their cut. Those three numbers are the entire argument, and watching the third one lag the second is the fastest way to see what the fund costs.

How wide the claim is

The dashed line is a single path, and a single path is a bad way to state a forecast. The shading around it is 40,000 simulated paths, drifting along that projection but carrying bitcoin's measured 45% volatility, the mNAV band, fat tails and a daily-reset fund. The inner band is the middle half of outcomes, the outer one the middle eighty per cent.

Read the projected column against the median beside it: they track, which is the drift doing its job. Everything else in the row is volatility, and the width comes straight out of bitcoin's measured 45%. Over the 2.32 years to the last expiry that is a log spread of about 0.70, which puts the best tenth roughly two and a half times the median. There is no extra assumption in those columns, only that number compounding.

That arithmetic has an uncomfortable consequence, and it is better stated than buried. The best tenth of paths finishes around 9 times the bottom, which is a bigger cycle than the last one delivered. Our entire thesis is that cycles are getting smaller. So the top of that band describes a world our own argument says is over.

That is not a bug in the simulation; it is the price of bolting a symmetric volatility model onto a directional view. The drift carries the thesis, the noise does not know about it, and the two disagree in the tails. We read the middle of the distribution and treat the upper decile as what the volatility permits rather than as anything we are forecasting.

A note on that volatility, since it is doing so much work here. It has not been constant across cycles, and it has not simply been falling either:

It rose into the second cycle and then dropped hard in the third, as bitcoin got larger and better arbitraged. We use the trailing measurement, 45%, rather than an average across those cycles, precisely because the old regime is not the one we are in. If volatility keeps compressing, the bands above are too wide and the options are too expensive. If it reverts to the 76% of the second cycle, they are too narrow and the calls are cheap.

By the last expiry the middle eighty per cent of paths still spans a range several times wide. That is the honest width of a cycle forecast, and it is why the position is built from four dated contracts rather than one.

Percentiles hide something an option holder cannot afford to ignore, though. They report where bitcoin ends up, not how it travels, and a call is paid on the journey as much as the destination. Here are sixty individual paths from the same simulation, with the projection drawn solid over them:

Almost none of them looks like the smooth line. They overshoot, stall for a year, round over early, or arrive late, and the smooth path is the average of all of that rather than a description of any of it. Two of those behaviours cost real money here: a path that arrives after 2029-01-19 pays nothing however high it eventually goes, and a path that spends its first year flat still pays the fund's rent the whole time.

What sits between bitcoin and the position

Two things, and both of them matter more than the choice of strike.

MSTR is not bitcoin. It holds 0.00215 bitcoin per share, so each share is backed by about $181 of coin, and it trades at $161.61. That is 0.89 times the bitcoin behind it: MSTR is currently worth less than the coin it owns. We assume that multiple settles at 1.5 over the horizon. That assumption is doing real work. Going from 0.89 to 1.5 is most of a double before bitcoin moves at all, and it is the single least defensible number in this report.

MSTX is not MSTR held twice. It targets twice MSTR's move every day and resets at every close, and resetting costs the square of volatility. MSTR has been running at 81% annualised, which at two times leverage is a drag of 66% a year, before the 7.8% of fees and financing on the borrowed half.

How much of this is the multiple

MSTR at 0.89 times its coin is the assumption doing the most work, so it is worth pricing separately. Holding everything else fixed and changing only where the multiple settles:

Two things fall out of that table. The position does not need the multiple to recover: at 0.9, roughly where MSTR trades today, it still returns 2.7x, because bitcoin tripling carries it on its own. And the multiple is worth about 4 times the outcome across the range we would call plausible, which is a large enough swing that anyone who disagrees with our 1.5 should read their own row rather than our headline.

Note the middle column. Even assuming it settles at 1.5, the multiple only reaches 1.41 by the last expiry, because it reverts at a half-life rather than jumping. The position is not paid the full recovery.

What it pays on the way

If bitcoin tracks the dashed line, the chain compounds like this. Read the bitcoin column down and everything to its right follows from it.

The first rows are the point. The position is under water for the best part of a year while bitcoin climbs, because every strike is well out of the money and the fund is paying its rent the whole time. A holder who cannot sit through that does not get the rest of the table.

What the simulation says, and what it cannot

Running 40,000 paths around the projected line, with bitcoin's measured 45% volatility, the mNAV band, fat tails and the daily reset all switched on:

It reports a 91% chance of profit and a 4% chance of the whole position expiring worthless. Those numbers are worth very little, and it is important to say why.

The simulation randomises the path, not the thesis. Bitcoin's drift is pinned to the cycle projection, so every one of those paths is a world in which the four-year rhythm holds and the decay lands where we fitted it. What is being measured is noise around an assumed destination. A 91% win rate is therefore a statement about volatility, not about how likely we are to be right.

The honest reading is that the distribution above is conditional, and the unconditional risk is dominated by one thing: whether the cycle repeats a fourth time. That is not a number we can produce, because we have three observations of it.

What would make us wrong

The cycle does not repeat. Three completed runs is three observations, and the regularity of the decay across them could be coincidence. If the rhythm breaks, nothing downstream of it survives: the bitcoin path, the MSTX price and every strike are wrong together rather than one at a time. This is the whole risk, and it is not in the distribution above.

It repeats, but late. The longest listed MSTX contract expires 2029-01-19, and we project the top on 2029-08-30, roughly eight months later. The position cannot hold to the top even if the top arrives. A cycle that runs a few months long turns a good outcome into a total loss, which is the sharpest edge on this trade and the reason the legs are spread across dates.

mNAV does not recover. We assume it goes from 0.89 to 1.5. If MSTR keeps trading at a discount to its own coin, a large part of the projected return simply never appears, and no amount of being right about bitcoin replaces it.

The fund does not survive it. MSTX is a leveraged product in a volatile underlying. A single day past the leverage limit ends it, the sponsor can close or reverse-split it, and listed options would be adjusted rather than paid. Our simulation puts that at a low rate; we do not think a low rate on a three-observation model is reassurance.

The book is not there when it matters. Open interest on these strikes runs from 28 to 300 contracts. Prices here are the ask, so the entry is costed honestly, but an exit before expiry may not be available at anything like model value.

What we are actually doing

Sizing to a loss we are content to take, and treating the whole $47,920 as spent on the day it is committed. A position whose median outcome is 17.16x and whose modal outcome over a wide range of unmodelled futures is zero should never be a position you need to be right about. It is a call on one specific, dated, falsifiable claim: that bitcoin does a fourth cycle, smaller than the last, roughly on time.

We think that is worth 47,920. We do not think it is worth more than we can lose.