tradebotsecrets · research record · 2026-08-30
Two operator questions answered from the local full daily history (2010-08-18 onward, close basis,
extremes measured from the data itself): (1) is the social-media “1,064 up / 364 down” cycle model a better
projection than our halving-anchored 526/889 calendar? (2) are cycle bear-market drawdowns shrinking on a
deceleration curve, and where does that put this cycle’s bottom? Scripts:
run_cycle_anchor_comparison_v1.py, run_cycle_drawdown_deceleration_v1.py.
Our model: high = halving + 526d, low = halving + 889d. The social-media model: high = prior cycle low + 1,064d, low = high + 364d. Since 889 − 526 = 363 ≈ 364, the down-leg is effectively identical — the models differ only in what anchors the top.
| Cycle | Actual high / low | 1,064/364 error | 526/889 error |
|---|---|---|---|
| 2017 top → 2018 low | 2017-12-17 → 2018-12-16 | +3d / +3d | 0d / +1d |
| 2021 top → 2022 low | 2021-11-09 → 2022-11-22 | −5d / +9d | +21d / +36d |
| 2025 top | 2025-10-07 ($124,777) | −14d | +9d |
Mean absolute timing error: ~7 days for the low-anchored model vs ~13 days for the halving model (whose exact 2017 hit is its own fit artifact). Clean-era trading replay 2015→2026 (1x calendar long/short, fees cancel): low-anchored causal chain +102%/yr log-growth vs halving +93%/yr vs buy-and-hold +51%/yr.
| Anchor | Bottom call |
|---|---|
| Halving 2024-04-20 + 889d (ours) | 2026-09-26 |
| Actual 2025-10-07 high + 364d | 2026-10-06 |
| Pure 1,064/364 chain (2022 low propagated) | 2026-10-19 |
History slightly favors anchoring the down-leg at the realized high: 2017 high + 364d hit the 2018 low exact to the day; 2021 high + 364d was 14 days early on the 2022 low. Working bracket: 2026-09-26 → 2026-10-20, center of mass 2026-10-06. No long flip before Sept 26; full flip confidence only after Oct 20.
Bear drawdowns per cycle (all-time-high tops only, which correctly excludes the 2019 relief rally; max close-to-close drawdown before the next top):
| Cycle top | Bottom | Drawdown | Shrink ratio vs prior |
|---|---|---|---|
| 2011-06 ($34) | 2011-11 ($2) | −93.2% | — |
| 2013-12 ($1,137) | 2015-01 ($172) | −84.9% | 0.911 |
| 2017-12 ($19,280) | 2018-12 ($3,232) | −83.2% | 0.981 |
| 2021-11 ($67,562) | 2022-11 ($15,760) | −76.7% | 0.921 |
| 2025-10 ($124,777) | provisional 2026-07-01 ($58,534) | −53.1% | 0.69 so far |
Each completed bear retained ~0.91–0.98 of the previous one’s magnitude (geometric mean 0.94) — the deceleration is real and surprisingly regular. Projecting it onto the current cycle from the $124,777 top:
| Deceleration model | Projected drawdown | Implied bottom |
|---|---|---|
| Linear magnitude trend | −71.7% | $35,318 |
| Constant magnitude ratio (0.94) | −71.8% | $35,130 |
| Exponential retention growth | −63.4% | $45,682 |
The provisional low (−53.1%) is shallower than every deceleration projection. Two mutually exclusive readings: (1) the curve broke upward — the institutional/ETF era structurally dampens bears and the July low is the bottom, making all three calendars late; (2) the curve holds — the on-schedule bottom is −63% to −72%, i.e. $35–46k, arriving inside the Sept 26 – Oct 20 window.
Close basis throughout — intraday wicks run a few percent deeper. Cycle extremes are data-measured (±365d window), not chart folklore. n = 4 completed cycles: every projection here is curve-reading, ranking-grade only, and none of it is FDR-confirmable. The 2026-07-01 low is provisional until the cycle closes. This page records evidence, not financial advice.