# PRIORAXIS

## Cycle-Adjusted DCA for Bitcoin — Method, Evidence, and Boundaries

**Version 1.0 — August 2026**

---

### §1 — The Problem

Most Bitcoin accumulation plans fail at the same moment: when the market moves. A flat DCA ignores where the cycle stands — the same amount at the top of a mania as at the bottom of a capitulation. The common alternative is worse: a plan adjusted by feel, which means a plan invented halfway through a crash. A plan invented halfway through a crash isn't a plan. It's a panic wearing a plan's clothes.

This is not speculation about other people's behavior; it is measured. The Bank for International Settlements, simulating a steady $100-per-month purchase plan begun whenever each user first opened an exchange account, estimates that 73–81% of retail investors lost money on Bitcoin between 2015 and 2022. Read that carefully: the simulated plan was disciplined monthly DCA, and most simulated investors still lost — because of when they arrived. Adoption follows price by roughly two months, so each cycle's newest cohort buys nearest the top. US ownership surveys tell the second half: participation fell sharply after the 2022 crash and, by the 2023 survey published in mid-2024, had not recovered even as price made new highs; later readings show only partial recovery — some who left did not come back. The same record holds the opposite cohort: long-term holders — in Glassnode's on-chain measure, coins unmoved for five-plus months — whose supply has repeatedly grown through bear markets to new highs, including the current one. Each cycle splits its participants into a washed-out cohort and a committed core, and the split is decided in the declines.

The arithmetic beneath this is worth one paragraph, because it is the paper's foundation. Entry timing matters in exact proportion to how short the horizon is: for a lump sum it is nearly everything; spreading purchases dilutes it; years of holding dilute it further. Time repairs almost any arrival — but only for the holder who stays, and staying through an 80% decline is the one thing arithmetic cannot do for anyone. Volatility is not a footnote to Bitcoin; for years yet it is the asset's defining property, and what it attacks is not portfolios but resolve.

You may know the sequence from the inside: the crash that made continuing feel reckless, the skipped buy that became three, the vow at the bottom — and the re-entry later, higher, when buying felt safe because everyone was doing it. No backtest simulates that sequence, because every backtest assumes the one thing the sequence destroys: that the plan keeps being executed.

The problem, then, is not information — cycle indicators are public and abundant. The problem is reading them the same way, every week, for years: without drift, without mood, without the quiet override that follows a red month. That is a discipline problem twice over — knowing where you are arriving, and staying once you have — and discipline is exactly what a rule carries better than a person.

### §2 — What PRIORAXIS Is

PRIORAXIS is a published, rules-based reading of where Bitcoin stands in its cycle, translated into a weekly DCA weighting: more when the cycle reads cheap, less when it reads expensive. It arrives as one email every Monday — the reading, the weighting, and the reasoning — the same rule for everyone, published before the week it governs. Every past reading is kept, unedited. Between them, the reading and the weighting address the two failures §1 measures: the arrival is no longer blind — every week's buy is sized to where the cycle actually stands — and the staying is no longer unaccompanied: the same rule, the same reasoning, every Monday, including the ones that follow a crash.

PRIORAXIS is a measurement, not advice. It does not know its readers, does not size anyone's position, and does not predict. It states where the market sits against its own history and leaves the decision where it belongs — with the reader. It is written for the person deploying real money on a schedule — the one who has, at least once, watched a red week and not pressed the button. The founder runs his own accumulation under the same rule, on the same Monday.

### §3 — The Method

PRIORAXIS reads the cycle through four instrument families, combined into one composite: the PRIORAXIS Risk Factor. Macro context. On-chain valuation — where price stands against what holders actually paid, drawn from realized-value metrics computed directly from public blockchain data. Market structure. Long-horizon asset positioning — how the asset is held and by whom, over horizons of years rather than weeks. An execution layer that translates the composite into the week's weighting.

Each instrument is normalized by rolling 208-week percentile rank — every reading is scored against the last four years of its own history, not against a fixed threshold. Fixed thresholds age; a percentile does not. The composite maps to five zones — Deep Value, Value, Neutral, Elevated, Distribution — and the zone sets the week's DCA weighting: more when the cycle reads cheap, less when it reads expensive. In practice, the weighting is a multiplier on the reader's own base amount — the reader owns the plan; the rule scales the week.

Three properties of the data pipeline matter more than any single indicator. First, the price index is computed in-house from three exchanges — no vendor index enters it. Second, the on-chain metrics are computed from the public blockchain record itself, and the pipeline independently re-verifies its matching of coins spent to coins received, daily. Third, every published reading is kept as published. When a reading is wrong, it stays in the record. The method's constants — the weights, the exact zone boundaries — are PRIORAXIS's own, and may be refined as the record grows; published readings never change — and a refinement is itself an event on the record: announced in the weekly reading, dated, and timestamped the same way, with every published reading carrying the version of the rule that produced it. The constants are withheld; the claims are not — every published reading and the full backtest program are on the record. What a reader can check is what was published and when, down to the byte; what the constants would add is re-derivation of the composite, not a stronger check of the record — and an unannounced change of rule is exactly what the versioning commitment just stated makes visible. The record's integrity is checkable by anyone: the timestamps prove what existed and when, and no one's word is required for it — ours included.

### §4 — What the Backtests Showed

The claim this section tests first is the foundation of every other: not that the rule beats the market, but that it holds its schedule — every week, including the weeks that break people. So the first finding is behavioral, from the trade log. Through the 2021–22 decline, the rule deployed in every one of its 53 weeks. Through the decline in progress since September 2025, in all 38 weeks to its current low in June 2026. And in the 2017–18 collapse it deployed in 47 of 52 — the five weeks it sat out were the five immediately after the December 2017 peak, while the reading stood at Distribution, the most expensive of the five zones §3 defines.

Three price tops sit in the record, and the rule met each one the same way — by following the reading — while the reading met each top differently. In 2017 it stood at Distribution at the peak and through the five weeks that followed: multiplier zero, nothing bought. In 2021 it had left Distribution seven months before the November top, which it read as Elevated: the rule bought through that top at half weight, then scaled up as the reading fell — half, full, one-and-a-half, double by late January. In 2025 the top again read Elevated, the last Distribution reading eleven weeks prior. The universal is about the reading, not about tops: the rule does not buy while the reading stands at Distribution — whether that zone coincides with a price top is a question the record answers three times, three ways. (The 2017 readings predate the composite's full four-year history window and are computed on its early, shorter history — stated because the record states everything.)

The numbers below were produced after their own rules. Twelve candidate sentences were frozen with their pass conditions before the first run — that is, what this section would be allowed to claim was written down before the test existed, so the results could not be cherry-picked. Four were earned, five failed their conditions and are absent from this section, three were struck in advance as unstatable by any backtest. Every verdict is in the findings document. The phase boundaries and benchmarks were frozen the same way, and the misses are printed with everything else.

The market's phases were drawn by a rule, not a hand: a cycle peak is a weekly all-time high followed by a decline of at least 50% before any new high; the trough is the lowest close before the next one. Any threshold from 48% to 75% returns the same two closed-cycle peaks — December 2017 and November 2021 — so the boundaries are the market's own, not ours. Six phases result, spanning May 2016 to the present cycle.

Three arms ran in every phase, fees applied identically to all: flat weekly DCA; the PRIORAXIS ladder — the reader's own weekly base amount, scaled by the zone reading: double in Deep Value, one-and-a-half in Value, unchanged in Neutral, half in Elevated, zero in Distribution; and a perfect-foresight construct — the same weekly contributions, each spent at the lowest price in its own future. Holding that third arm requires knowing the future. It exists as an upper bound no strategy can beat, and none did: it is there to show what perfect timing would be worth — and that nobody selling it has it.

One thing to expect before the table: the ladder withholds when the cycle reads expensive, and a rule that buys less near a top cannot hold as many coins as a rule that always buys, across a window that ends at that top — and in a window that begins mid-expansion and ends at a top, severely so. Trailing in expansions is the strategy's shape, not its failure mode — the question the table answers is by how much, and what happens in the declines.

| Phase | Window | Move | vs flat DCA | vs perfect foresight |
|---|---|---|---|---|
| P1 — expansion (partial: starts at data edge, no trough) | May 2016 – Dec 2017 | +4,032% | 0.326 | 0.307 |
| P2 — decline | Dec 2017 – Dec 2018 | −83.2% | 1.064 | 0.478 † |
| P3 — expansion | Dec 2018 – Nov 2021 | +1,952% | 0.965 | 0.773 |
| P4 — decline | Nov 2021 – Nov 2022 | −75.2% | 1.005 ‡ (band 0.55%) | 0.585 † |
| P5 — expansion | Nov 2022 – Sep 2025 | +659% | 0.959 | 0.898 |
| P6 — decline, open | Sep 2025 – | −51.8% | 1.001 (no verdict: no confirmed bottom) | 0.755 (no verdict) |

Reading the ratios: above 1.0 = more coins than the benchmark for the same contributions; below = fewer. ‡ Within band: the gap to parity is smaller than the movement produced by shifting the phase edge a fortnight (the band); † boundary-sensitive: that movement exceeds 3%, so no verdict is issued. Full per-phase bands in the campaign artifacts. P1 and P2 readings are computed on the composite's early, shorter history (the full four-year window completes in 2018); the same caveat stated with the 2017 narrative governs both rows.

The table shows the expected shape plainly: PRIORAXIS trails flat DCA in every expansion. The declines read differently — 1.064 through an 83% collapse, 1.005 through the next — and PRIORAXIS builds no claim on those figures: its standing rule is that it may not claim more coins than flat DCA, and it does not. The figures are stated because the record states everything.

Against perfect foresight, nothing came close, anywhere — and that is the point of the benchmark. The gap in that column is the measured price of knowing the future: unavailable to PRIORAXIS, to flat DCA, and to anyone selling it. PRIORAXIS claims no timing alpha; its hindsight arm was unbeatable by construction, and the distance to it is printed, not excused.

A phase is a deliberately hard lens on a cycle-length rule, and here is why: cash withheld near a top exists to buy the next phase's low prices. Cutting the record at the phase boundary counts the withholding and discards its payoff. The rule's own claim lives at cycle length — 430 rolling two-year windows since 2016. The full distribution: median 0.9904 coins versus flat DCA — one percent behind, against an asset that routinely moves that much before lunch; a tenth of windows ended at 0.6124 or worse; the worst, 0.3123. The tenth-percentile and worst windows are both starts inside the 2016–17 expansion — the same shape P1 shows above, carried into the rolling record. The middle of the distribution is a rule that stays within one percent of the benchmark while carrying the discipline the benchmark assumes for free; the tails are what starting mid-expansion cost, printed rather than averaged away. Which kind of start today resembles is not left to guessing — the current reading is published every Monday, before any money moves.

Read the parity as a receipt, not a tie. At cycle length the rule lands within one percent of always-buying — the median of 430 rolling windows — after refusing the 2017 top outright, buying the 2021 top at half weight, and doubling into the collapse that followed. The coins arrive at the same place; the capital travels a different road. Not one unit was deployed across the 75 weeks the reading stood at Distribution — seventy of them inside the expansions' run-ups, five at the start of the 2017–18 collapse. A third of always-buying's capital entered at the two most expensive readings; under a tenth of PRIORAXIS's did. And every week carried its published reason, set down before the week began. That road is what PRIORAXIS sells — not a coin edge, which the record does not show and the rule may not claim — but the same coins, acquired at prices a person can defend to themselves at the extremes.

One limit is stated rather than left for the reader to find: a backtest assumes its own discipline — every arm executes flawlessly by definition, so 53 of 53 is the rule's behavior, not yet a person's. Whether a person holds the line is answerable only by a record kept in public, week by week. That record is the product, and it is being kept.

One more limit belongs beside it: the rule was designed by someone who had already watched the history it is tested on. A backtest on that history is therefore a consistency check — evidence the rule does what it says on the data it grew from — not out-of-sample proof. A fitted rule would also have something to show for the fitting; this one's headline is a printed one-percent shortfall. The out-of-sample test is the forward record: every reading published before the week it governs, kept unedited, begun August 2026. This paper is the weaker instrument of the two, and it says so; the record is the stronger one, and it grows every Monday.

*(Pre-registrations frozen before their runs and timestamped on the Bitcoin blockchain; documents, sha256 digests and proofs are published at prioraxis.com/verify/. Phase and per-phase figures — PHASE-GRID-R1: phase table `f2d07c1b…`, results `dd7b676f…`, raw output `528c699d…`, findings `7f802a7b…`. Rolling-window median — SCEN-DCA: `rolling_windows.csv` `f0dc3f02…`. Deployment shares — JOURNEY-R1: findings `f7a21027…`.)*

### §5 — The Sell Side

PRIORAXIS's published readings and weightings govern accumulation. A sell-side framework — when the same zone logic argues for reducing rather than adding — exists as design, built on the same instruments and the same zone vocabulary. It has not been through the backtest program the accumulation side survived, and PRIORAXIS does not publish sell-side guidance until it has. Its parity test is scheduled behind a live parallel trial of the sell-side scoring, now running. When it runs, this section will carry the result — whichever way it falls. And if the test does not support publication, the sell-side framework will not be published: the same bar the accumulation side had to clear.

### §6 — What PRIORAXIS Never Does

PRIORAXIS never touches funds. It never asks for exchange keys, and never will — the promise is architectural, not procedural: there is no PRIORAXIS system to which a key could be given. It never personalizes: the same rule, the same reading, the same weighting for every subscriber, which is also what keeps the reading a publication rather than advice. It never predicts — a zone is a measurement of where price stands against the market's own history, not a forecast of where it goes. And it never edits its record: a published reading stays as published, including the wrong ones.

### §7 — The Record

Every Monday reading since publication began is kept at its original wording — dated before the week it governed, checkable against what the market then did. The record is the method's real audit surface. PRIORAXIS asks to be judged on that record rather than on this paper: the paper describes the method once; the record demonstrates it weekly. It is public at prioraxis.com/the-record/, kept since August 2026. Where the cycle stands this week is itself published — the current reading is on the site and in every Monday email.

### §8 — Who Runs This

PRIORAXIS is built and operated by Bojan Zimmermann, an engineer by training, based in Slovenia, active in Bitcoin since 2017. He built the rule after doing everything §1 describes, in his own account, across more than one cycle — the system exists because its builder needed it first. The system is the product of years of systematic-accumulation work, consolidated into the framework this paper describes. He runs his own accumulation under the same published rule, on the same Monday, with no private variant. PRIORAXIS holds no client funds, executes no trades, and sells one thing: the reading, on schedule, with its reasoning shown. PRIORAXIS is one person, and the design accounts for it: the method and record are published and timestamped, and a subscription is cancellable monthly — a reader's exposure to the founder is at most one month deep.

### §9 — The Path From Here

PRIORAXIS develops in one direction: a deeper record, verified harder. Each step below is gated on a condition, not a date — it ships when it is earned, in the sequence the conditions allow.

**The sell-side result.** The sell-side framework's parity test runs behind that live trial. Its outcome — either way — enters this document and the record.

**The member tier.** A paid tier carrying the full weekly reading: the composite, the weighting the rule derives, and the reasoning at depth. It opens when every feature it describes exists in the present tense — PRIORAXIS does not sell futures.

**The conformance ledger.** The record, made mechanical: what the rule said in advance, graded against what followed, published as a running ledger rather than an archive. Designed; built when the record is long enough to grade.

**The founder's record.** The founder's own accumulation under the published rule, documented when its own publication gate is met — the same sequencing as everything above it.

What PRIORAXIS will not develop is already stated in §6, and it is permanent. The path adds depth to the record and rigor to its verification. It does not add prediction, personalization, or custody — those are not missing features. They are the boundary.

---

### Version History

**v1.0 — August 2026. Signed 2026-08-15.** First publication. Accumulation-side backtest program complete (PHASE-GRID-R1: six mechanically defined phases, 2016–2026; pre-registration frozen before the run and timestamped on the Bitcoin blockchain; documents and proofs at prioraxis.com/verify/). Sell-side framework exists as design; its parity test is scheduled and this document will carry the result.

---

*PRIORAXIS publishes a measurement, not advice. Nothing in this document is a recommendation to buy, sell, or hold any asset.*