BL-EC-005: Capacity and Venue Access
BIJECTIVE LABS
BL-EC-005: Capacity and Venue Access
Status: RELEASED
Classification: PUBLIC
Revision: 1
Date: 2026-09-07
1. What was reachable
| Reachable | Not reachable |
|---|---|
| Bybit USDT perpetuals | Options, any venue |
| Hyperliquid perpetuals | Colocated execution |
| Bybit spot, majors only | VIP or institutional maker tiers |
Three of the four unreachable items are consequences of capital scale and are
covered in BL-EC-004. The first is different: an options desk is an access
question rather than a size question, and it is the one that blocks the only
signal class the programme validated out of sample twice.
2. Spot availability, and a verdict that turned on it
The cash-and-carry evaluation is the clearest case in the record of a verdict decided by what could be reached rather than by what was measured.
| Stage | Result |
|---|---|
| Instruments with net-positive carry after costs | 14 |
| Of those, with a spot market on the venue | 1 |
| Yield on the one executable instrument | 2.7 pct annualised |
| Internal bar | approximately 10 pct annualised |
Thirteen of fourteen carries were measurable and not transactable. The strategy required simultaneous exposure in perpetual and spot, and the spot leg did not exist for the instruments where the carry was worth having.
The row is classified ENVELOPE-BOUND on venue access. A reader operating where
those spot markets exist is looking at a different question, and the record
carries the measurement rather than only the conclusion.
3. Options access, and the class it blocks
The magnitude class is the one signal family that survived every test in both programmes. It validated out of sample twice, once as a lift measurement falling from t=+10.25 to t=+4.72, and once as a Brier skill score of +0.01251 in sample against +0.01053 out.
Its direct monetisation is an options position. The variance risk premium that would be harvested measures +8.10 volatility points at t=22.0 on BTC across 1,277 days, which is not a marginal finding.
Both routes are closed here. Selling the premium requires margin of 6,345 USD per contract against a declared capital scale of 1,000 to 2,000. Buying it when conditioned was falsified outright, because the fragility gate raises the premium rather than lowering it, meaning the options market has already priced whatever the signal detects.
What remains is the defensive use, which is the one component from either programme running in production.
For a reader with options access, this is the largest unexplored route in the record. The signal is validated, the premium is measured, and the monetisation path has never been tested by anyone at Bijective Labs.
4. Universe
| Measurement | Date | Count |
|---|---|---|
| Bybit USDT perpetuals enumerated | 2026-05-31 | 570 |
| Passing the movement filter | 2026-05-31 | 192 |
| Instruments in the Programme II research panel | 2026-07-27 | 154 |
| Bybit perpetuals in the spread census | 2026-08-01 | 676 |
The venue grew by 106 instruments over roughly nine weeks. That is worth recording rather than smoothing, because it affects how the two figures should be compared: the 570 and the 676 are the same venue at different times, not a discrepancy.
The 154-instrument research panel is a deliberate subset. It was selected on data availability across all four required streams over the full research era, so that the panel is balanced rather than survivorship-weighted toward instruments that launched recently.
5. Capacity was not measured
This is the largest gap in the volume and it is stated plainly.
No strategy in this programme ran at a size where its own transacting moved the book it was measuring against. Market impact therefore appears nowhere in the cost identity, and no capacity ceiling was estimated for any row in the Kill Ledger.
The consequence for a reader is specific. Every cost figure in this record is a floor that omits the reader's own footprint. A strategy that clears cost here at 1,000 USD may not clear it at 1,000,000, and this record cannot say where the crossover is.
Two rows would need capacity work before any deployment at size.
The cross-sectional families rebalance a wide panel, which means many small legs rather than a few large ones. Turnover measured 0.776 per rebalance on the residual reversion construction, and cost consumed 84 percent of the gross result at that turnover. Capacity there is a turnover question rather than a depth question.
The market-making family is capacity-bound by definition, since a quote is only filled to the extent that somebody wants the other side.
6. Data substrate as a form of access
The two venues sit at opposite ends of a trade, and no amount of capital changes which one answers which question.
| Source | Depth | What it carries |
|---|---|---|
| Bybit historical, 154 instruments | approximately 4.8 years | Klines, funding, open interest; anonymous |
| Hyperliquid | 2026-07-03 to 2026-07-24 | Wallet-level positions, entry and liquidation prices |
| Total | 8.20 GB |
The identity class in BL-P2-003 exists only because of the second substrate and
could not have been produced by the first at any depth. The out-of-sample
validation of the direction and carry classes exists only because of the first
and could not have been produced by the second at any richness.
Access to data is an envelope parameter in the same sense as a fee tier, and it is the one most often treated as fixed when it is not.
7. Cross-references
BL-EC-004 for the capital scale that fixes most of section 1. BL-P2-003 for
the class-level view. BL-SD-016 cash and carry. BL-SD-038, BL-SD-047,
BL-SD-048 for the volatility rows. BL-FM-002 section 7.
Copyright (c) 2026 Bijective Labs. All rights reserved.