SetupAlpha

Professional-grade trading strategies from research to your portfolio.

[SetupAlpha @SetupAlpha]

Join the community here: https://t.co/AQ92OPg9BH

[SetupAlpha @SetupAlpha]

We’ve officially hit the #3 New Bestseller on Substack! 🏆

Knowing that these deep dives into backtesting, RealTest, AI and quant workflows are helpful to you is what keeps me going.

[SetupAlpha @SetupAlpha]

A Monte Carlo test is useful for one reason:

> Your future trade sequence will not match your backtest trade sequence.

Risk comes from the paths capital can take, not just the endpoint.

[SetupAlpha @SetupAlpha]

The full checklist, results, and RealTest code are in the article: https://t.co/WB0CFI78nv

[SetupAlpha @SetupAlpha]

From $10k to $3.96M

As someone who tests RealTest strategies for traders who care about costs, this Dow Award paper is worth studying for one reason: “when leverage earns its place”.

The full checklist, results, and RealTest code are in the article:

[SetupAlpha @SetupAlpha]

Watch the full Gemini and RealTest portfolio test here: https://t.co/dUtcBHBVfz

[SetupAlpha @SetupAlpha]

I tested Gemini on a RealTest portfolio build.

In this test, it combined 5 strategies, added one Risk Percent control, avoided duplicate symbol exposure, then upgraded weights with rolling 50-day return correlation.

[SetupAlpha @SetupAlpha]

Don’t optimize your strategy for the ideal scenario.

Build a system that still works when you’re emotional and the world does something new.

[SetupAlpha @SetupAlpha]

Full breakdown, including the exact prompts and charts: https://t.co/pHCk7CAIPb

[SetupAlpha @SetupAlpha]

I gave Codex, Gemini, and Claude Code the same RealTest job:
build SPY strategy
run base test
optimize MAs
run walk-forward
make charts
All 3 finished.
But there was one catch.

[SetupAlpha @SetupAlpha]

Copy this strategy here: https://t.co/rLh9OXV6fg

[SetupAlpha @SetupAlpha]

This trading strategy a very simple.

> Gaps below yesterday's low.
> Buy.
> Wait for a green day.
> Sell.

The backtest had 953 trades and a 64.01% win rate.

[SetupAlpha @SetupAlpha]

Enjoy the full video here: https://t.co/C8QLKv1xdA

[SetupAlpha @SetupAlpha]

In theory AI can write a trading strategy in seconds.
I tested OpenAI Codex, Gemini, and Claude Code on the same SPY moving-average strategy.
My goal was to see which one could move through the actual RealTest research workflow with the fewest fixes.

[SetupAlpha @SetupAlpha]

Bad backtest does not mean backtesting is bad.

It means the test was weak.

[SetupAlpha @SetupAlpha]

The full test is here, including the RealTest scritp + rules: https://t.co/nQc0RXvfqo

[SetupAlpha @SetupAlpha]

The useful test is not trend following vs mean reversion.

It is whether they fail differently.

In this RealTest run, the 50/50 book had lower MaxDD than either standalone system and the best MAR.

The best standalone strategy was not the best portfolio ingredient.

[SetupAlpha @SetupAlpha]

You can download the RealTest script and strategy rules here: https://t.co/Xlek6BkoGF

[SetupAlpha @SetupAlpha]

Last week, I published the Nasdaq momentum strategy.

Right now is one of the best periods for momentum this year.

Conditions like this usually appear only once or twice a year.

You can download the RealTest script and strategy rules here:

YTD Results.

[SetupAlpha @SetupAlpha]

Full breakdown, including the RealTest code and the rules I removed: https://t.co/DhRQPefGqt

[SetupAlpha @SetupAlpha]

I added 8 filters to a mean-reversion strategy.

In-sample: 4 rules: 7.18% CAGR 12 rules: 8.62% CAGR

Out-of-sample: 4 rules: 11.78% CAGR 12 rules: 4.51% CAGR

The cleaner backtest was the worse strategy.