First principle in investing: there are no free lunches. Ever.
Every time you are making something better in your portfolio of trading system, you have to check what gets worse.
Claude, ChatGPT, Grok Bot all down, how are we supposed to replicate the same slop by hand?
Mostly, how do I get back to using my own brain, I don't even remember what it feels like.
Tested this exact instinct on my own systematic book — built 4 dials that cut exposure when conditions turn (vol-term-structure ratio, trailing-return trend, etc).
Benchmarked against the real control: just holding less, permanently. 3 of 4 lost on Sharpe, MaxDD AND CAGR at once. The 1 that "worked" only did in one crisis month — strip it out and holding-less wins again.
The wide spread right after a reopen isn't noise to wait out. It's the signal. Christie-Corwin-Harris found inside spreads more than doubling and vol running ~9x normal after those 5-min halt reopens (JF 2002). And the newer Hautsch-Horvath order-book work (JFE 2019) is the real tell: the price discovery happens inside the pause, extra vol comes after, no cool-off. So the directional info's already in that first ugly print. Wait for spreads to normalize and you've waited out the edge.
@RhysSullivan Grok Bot is good because it's out of the box but you can replicate the same exact thing with Hermes and a much cheaper model like Deepseek v4 flash if you have a bit of a tech background.