I test trading edges until they break.
Most don't survive. I show you the data.
NullEdge is a falsification-first research operation: every strategy is assumed wrong until it survives walk-forward out-of-sample testing, net of fees, across regimes. No hype, no signals group — just honest verdicts and the numbers behind them.
What I do
Edge Audit
Send me a strategy or backtest. I run it through the full falsification gauntlet — walk-forward, random baseline, cost stress — and give you an honest verdict before you risk a cent. See packages →
Honest research & reports
Data-driven write-ups on what actually works (and what doesn't) in crypto trading and DeFi. The methodology is open; the verdicts are unsponsored.
Custom research & builds
Bespoke edge research, or a paper-first trading/finance bot built and validated with the same rigor — multi-chain production systems in Rust/Solidity when you need them.
Proof, not promises
- ✅ Continuous falsification engine — default verdict REJECT; 8 edge families × 3 live feeds × 3 timeframes on real Binance data, all rejected. the write-up →
- ✅ Walk-forward, not wishful — every edge is optimized on the train window and tested on the unseen one, net of fees + slippage, across regimes.
- ✅ Runs 24/7, paper-first — an autonomous agent refreshes real data and re-tests daily; no real-money risk, no hardcoded keys, open methodology.
- ✅ Ships production systems too — multi-chain liquidation/MEV infrastructure in Rust/Solidity, fork-proven, when a build is what you need.
Research
Eight edges, three live feeds, zero alpha →
I built a bot to hunt trading edges and kill the fakes — 8 edge families across 3 live data feeds and 3 timeframes on real Binance data, every one rejected net of fees, walk-forward. The honest graveyard, with the numbers.
Your backtest is lying →
A single backtest is the most convincing lie in trading. Walk-forward is the one test that catches it — plus the four gates that finish the job. A practical guide, with real numbers.
The edge that failed, then passed →
My falsification engine killed my own +2.75% funding-carry edge as overfitting — then 3 years of data brought it back, regime-dependent. The anatomy of honest walk-forward validation, with real numbers.
The real economics of DeFi liquidations →
Why most "liquidation bot" income is a myth for solos — measured across 5 protocols + 3 chains, with the data.
Why a "faster" bot doesn't win liquidations →
The contest is detection cadence + an inclusion auction, not latency. Measured live on Aave V4 — leader share ~0.67, and why a quick VPS changes nothing.
DeFi Oracle Landscape (live data) →
Push vs pull oracles on fresh lending markets, measured live across chains — auto-generated from my scanner's dataset.
Contact
Want an honest verdict on whether your edge is real, or a system built and validated properly? Start with an Edge Audit, or reach out directly.