---
title: "Discussion on \"Every 'Pro' Risk Technique I Tested That Did Nothing\" | Hashnode"
description: "Discussion on \"Every 'Pro' Risk Technique I Tested That Did Nothing\". Trailing stops, risk parity, fancy overlays — I tested them honestly, and most did nothing."
url: "https://hashnode.com/posts/every-pro-risk-technique-i-tested-that-did-nothing/6a55bde038baf797529b7edb"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Every 'Pro' Risk Technique I Tested That Did Nothing](https://glassboxtrader.hashnode.dev/every-pro-risk-technique-i-tested-that-did-nothing?utm_source=hashnode&utm_medium=feed)

Trailing stops, risk parity, fancy overlays — I tested them honestly, and most did nothing.

Jul 14 · 3 min read · glassboxtrader.hashnode.dev  
Read the article

-   [#build-in-public](https://hashnode.com/tag/build-in-public)
-   [#trading](https://hashnode.com/tag/trading)
-   [#risk-management](https://hashnode.com/tag/risk-management)
-   [#python](https://hashnode.com/tag/python)
-   [#bot](https://hashnode.com/tag/bot)
-   [#backtesting](https://hashnode.com/tag/backtesting)
-   [#data-science](https://hashnode.com/tag/data-science)

## Written by

[Glassboxtrader](https://hashnode.com/@glassboxtrader)

@glassboxtrader · Zurich, Switzerland

  
[View profile](https://hashnode.com/@glassboxtrader)

No black boxes, no hype — an honest trading bot, built in public.

Building an honest, explainable trading bot in public — one PC, a 6 GB GPU, a NAS. No black boxes, no hype.

## Responses 1

Markdown supported · ⌘↵ to post

[Arnold Holm](https://hashnode.com/@stratcorealpha)

[Oct 2](https://hashnode.com/posts/every-pro-risk-technique-i-tested-that-did-nothing/6a55bde038baf797529b7edb/comment/6abf8fed630bbc28eddd6d2d)

For the drawdown brakes, I would separate the objective of reducing historical losses from enforcing a rule boundary. A challenge guard can be useful because it blocks an entry after a specified loss threshold, even if it does not improve the strategy result. A regression fixture should therefore check the threshold, reset time and restart behavior against the frozen rule, rather than judge the guard only by the backtest return.
