Strategies you can actually verify
Retail systematic trading has a trust problem. Strategies sell for $500–$1,500 on the strength of backtests you can’t inspect, validated by checkmarks the seller awarded themselves. I know because I’ve bought them — and when I ran a purchased strategy set through walk-forward analysis, one strategy survived.
This publication does the verification in public.
What you’ll get here:
Replications. I take vendor claims, papers, and community strategies and rebuild them on survivorship-bias-free data with honest costs. The verdict is always free — every test, pass or fail, with full results. When a strategy FAILS, the test script ships with the post so you can re-run the failure yourself; when something actually works, the code goes to paid subscribers — working code in beginner hands without the judgment to use it is how this industry burns people, so the code that’s safe to share is free and the code that isn’t costs something. (The deeper validation battery is my own tooling and stays in-house at every tier; what you’ll always get is its full output and the published methodology it implements, cited to the source papers, so you can audit the standard even though the machine is mine.)
A validation standard. Every strategy I publish or sell goes through the same gauntlet: Pardo-style walk-forward efficiency, multi-walk cluster analysis across 20 in-sample/out-of-sample configurations, a Monte Carlo, and the Deflated Sharpe Ratio with the trial count disclosed — the Bailey & López de Prado corrections for backtest overfitting, applied for real. Strategies that fail get published as failures. The passes are believable because you’ve seen the flunks.
The three-gate validation pipeline applies to strategies that trade often enough to be statistically testable — several hundred trades minimum; that covers most of what’s sold to retail traders. Slower strategies (rotation, trend, tactical allocation) get a different, honestly-labeled standard — return-level tests, cross-market breadth, regime coverage, parameter plateaus, and mechanism — because running trade-starved statistics and printing a checkmark anyway is exactly the theater this publication avoids.
The build. A multi-part series on constructing a multi-strategy systematic portfolio from scratch — sleeve selection, stress correlation, the evidence pipeline, the kill decisions — from someone who did it and kept the receipts, including the graveyard of everything that didn’t work.
The rules I hold myself to: conclusions are free, implementation is paid. Trial counts are always disclosed. Failed tests get published. Vendors get tested on claims, not dunked on as people — with methodology attached so you can check my work exactly as hard as I check theirs. I trade my own systems with real money. I will never show you my account — what I’ll show you instead is every test, in a format you can rerun.
Nothing here is investment advice; it’s research you can verify.
Posts are drafted with AI assistance; all research and code are the author's and independently reproducible
Subscribing is free — every post lands in your inbox, and free means the full verdict, not a teaser.
