Cross-domain synthesis
One instrument, eleven studies.
Every study on this site, run through the same unchanged core - and the honest verdict from each, side by side.
Delta Motion is one signal-detection engine, not a collection of tricks. The same core reconstructs a fair reference, measures divergence and lead-lag, filters artifacts, and validates against a null with every parameter fixed in advance. The table below is that engine applied eleven times across ten domains - six in crypto, one in wholesale power, one in retail fuel, one in cloud infrastructure, one in sovereign rates - with nothing in the core changed between them. Each cell traces back to a published study you can open and check.
The portfolio is deliberately crypto-heavy; energy and sovereign rates are the genuinely independent asset classes here, and they are where the claim that the method travels gets its real tests. The strongest evidence on this page is not the successes but the downgrades - two on replication, one on a geography placebo. Twice we pre-registered a replication of our own most confident finding, and twice it did not survive: SOL's funding-basis edge proved specific to a single week, and the ETH/BTC triangular reversion proved to be tick rounding rather than a binding force, once a second, stricter null was added. Reporting those in full is the point.
| Domain | Sharp reference | Deviation size | Lead-lag | Reversion vs surrogate | Verdict |
|---|---|---|---|---|---|
| Crypto - cross-exchange spot (BTC) | Cross-venue consensus (de-margined) | ~1 bp gap; clears retail/mid cost bands in ~0 of ~10,000 min per window | ~0 (<=0.5s); live re-confirmation only, not replicated | 20-52% vs ~2% surrogate in 6/6 windows; survives the low-information-minute mask | Efficient yet cointegrated - a real binding force pulls the venues together below cost. Replicated across six windows. |
| Crypto - perpetual basis (BTC/ETH/SOL) | Perpetual-spot no-arb / funding tether | Basis 0.7-3.0 bps (sigma) | - | Majors random walk 6/6; SOL reverts only in the overlap window (11.6% vs 2.7%); funding<->basis +0.14 to +0.77 (18/18) | Majors efficient; SOL's edge window-specific (did not replicate); funding-tethering replicates everywhere. |
| Crypto - triangular no-arb (ETH/BTC) | No-arbitrage identity (direct vs implied) | Max excursion 6-14 bps; never clears mid/retail in 60,480 min | - | Real 42-43% vs ~2% surrogate BUT below the 46-52% quantization null, 6/6 | Consistent to within one tick; untradable. The reversion is tick quantization, not a binding force - the stricter null overturned our own earlier reading. |
| Crypto - stablecoin peg (USDT/USDC) | The $1.00 peg | USDT ~-4 bps avg, drifting -9 -> -2 bps over a week; venues agree to ~0.5 bp; USDC within +/-1 bp | - | Not stationary (ADF cannot reject a unit root, p~=0.13); reverts no faster than surrogate | Independent venues agree on the price; the discount itself drifts and isn't self-correcting. An integrity signal, not a trade. |
| Energy - day-ahead vs real-time (CAISO) | Day-ahead clearing price | RT ~$3.3/MWh below DA (~9%); 95% CI -3.7 to -2.9 | - | Deseasonalized ACF1 +0.30 (z +15) vs value-permutation null; NP15 hub +0.70 | Genuine day-ahead premium and decaying predictability. One month, two hubs. |
| Energy - day-ahead vs real-time (NYISO) | Day-ahead clearing price | RT-DA median negative in 6/6 months (-2.7% to -7%); 95% CI excludes zero each month | - | ACF1 +0.35 to +0.65 (z +9.6 to +17.8), 6/6, vs value-permutation null | The CAISO premium and persistence reproduce on an independent grid over six months - a market-design property, not one grid's quirk. |
| Crypto - alt cross-exchange lead-lag (SOL) | Cross-venue consensus (timing) | - | 0.00s in all 14 windows; 5 cleared the power gate; floor ~0.5s, densest ticks ~0.08s | - | No resolvable lead - the alt is priced across venues contemporaneously, just as BTC is. A clean null at a stated resolution. |
| Crypto - DEX vs CEX lead-lag (ETH) | CEX order-book price | - | <= 2s block floor on Base (0-2s), CEX-first sign; 0 of 8 windows resolved | - | The on-chain price tracks the exchange inside a single 2s block - correctly signed, below resolution. A clean null. |
| Retail - fuel price dynamics (France) | Local (departement) consensus | |deviation| ~2 c/L (tail ~9-10 c/L); mean-reverts at z>=8 in 4/4 windows | - (H2 underpowered on daily data; no resolvable area-lead) | Fires 20-53% of departements locally, BUT a scrambled (random other departement) reference fires as often or more, 4/4 | No evidence of local competitive convergence beyond generic price stationarity - a geography placebo does not support the local reading. Methodology-grade; the surviving positive (synchronized identical-price feeds) is thin and fully caveated. |
| Cloud infra - cross-AZ spot (AWS us-east-1 g4dn) | Same-region leave-one-out consensus (vs geography placebo) | - (pre-data close; no analysis window examined) | - (H2 deferred; blind to contemporaneous coupling) | Non-identifiable - reflection problem; a foreign-region placebo false-detects, the true common factor over-corrects | Identification failure: contemporaneous cross-AZ coupling is not separable from a shared-regional-factor null by this design. Closed methodology-grade, pre-data. Rules coupling neither in nor out - not an efficiency null. |
| Sovereign rates - US/euro-area AAA 10Y | US CMT 10Y vs ECB AAA 10Y; daily yield changes | Same-day co-movement rho ~0.44-0.57 (the dominant link) | Lag-1 rho +0.20..+0.27 full / +0.16..+0.22 trimmed; 3/3 blocks; not separable from the ~4h close gap | Spread reversion a null (0.293 vs 0.344 surrogate p99) | Material lag-1 coupling, replicated 3/3 and robust to a top-5% vol trim; dominated by same-day co-movement, the lag-1 reported as coupling, not leadership. |
Two rows look alike and are not. The cross-exchange BTC gap and the ETH/BTC triangular residual both stay tiny and both mean-revert far above a random-walk surrogate - the surface reading is "efficient but bound" in each. Yet only the first is a genuine force. A second null - rounding the fine reference to the coarse quote tick - reverts even more than the real triangular residual, which marks that reversion as arithmetic, not arbitrage. The cross-exchange gap has no such explanation and survives every check, so its binding force is real. Same family, opposite verdict, separated only because the right null was chosen in advance.
Exploratory analysis of public data. Not investment advice or a recommendation.
The point
Read down the verdict column and the pattern is the product: genuine, untradable structure where the market has it - cross-exchange cointegration, the day-ahead premium on two independent grids - clean nulls where it does not, and priors honestly downgraded - two on replication, one on a geography placebo (retail fuel). A pipeline that found structure everywhere would be the thing to distrust. This one does not, and it says so on the record. The discipline of refusing signals that aren't real is the capability.