# Beware the ides of algorithmic trading — retained record

> Retained reddit record from reddit:algotrading.

Canonical: https://highsignal.app/data/records/1eb15fec0829f8d9

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:algotrading)  published 2026-10-06 17:42 ingested 2026-10-07 02:43

[original source ↗](https://www.reddit.com/r/algotrading/comments/1wz90oe/beware_the_ides_of_algorithmic_trading/)

## retained text



```
I guess kind of a vague title, but it comes with a story. I have been posting/commenting less here lately because of generally less time right now, but wanted to talk a bit about our journey.

We have been live trading since July 2025. The first 6-7 months were really up and down, as we were still working out kinks and such. Still never lost money, but also weren't beating markets. 

Our algorithm itself has been pretty unchanged since the very beginning (at least the raw buy signal part). We have improved on the exit strategy(s), how to commit capital, various keepouts that filters out some of the raw buy signals, etc. From Jan through March we did pretty well. We were beating markets by just a little bit (and if not for a couple of catastrophe trades would have been doing very well...don't get me started on our pattern of 500 good/acceptable trades and 1 catastrophe trade that knocks us back down to earth). We added a couple of keepouts to try and address those catastrophes and did a restart in mid-April.

It was humming along again with more like 1000 good/acceptable trades to every catastrophe trade and we were again beating markets by just a hair more. 

**Then I started meddling**, and really the only meddling was with our parameter optimizer algorithm. This caused us to have an atrocious month of August with -8% overall and knocked us back down to just barely 3% since the April restart. So at the start of September I made the painstaking process of reverting back to the July 23 codebase in version control and start marching forward with **only** the features that were honest-to-goodness bugfixes and not attempts at performance or optimization improvements. September was back to regular performance. We had a +2.73%, which I consider a huge win considering our benchmark of VTI was -1.03% in that same timeframe.

The moral of the story is that just as hard as it is to find something that works even reasonable well, it is even easier to break it.

tl;dr - This has probably been my biggest message across all my posts here, but I need to be far more disciplined about proving proposed improvements before I send them live.
```



document fetched 2026-10-07 00:44 · canonical https://reddit.com/r/algotrading/comments/1wz90oe/beware_the_ides_of_algorithmic_trading
