Faith can move mountains, but T can move markets.
St Tropez, France
- Tweets 10,329
- Following 1,045
- Followers 676
- Likes 10,753
Pinned Tweet
You ever wonder why someone would be shitposting on here if they could really do this?
Like genuinely ask yourself… If you had the talent to make a 160x in a few months, let alone 24 hours, would you still be here chasing engagement?
The best traders in the world don’t post PnLs for engagement.
The largest desks don’t flaunt their volume for clout.
Why can’t we identify the largest whales holding $btc or $eth?
Simple - they don’t want you to know who they are. They don’t care about clout or engagement. They care about revenue and profit.
And at the end of the day, do they really want a target on their backs? Obviously not.
It’s the same every cycle.
Clout chasing, engagement farmers acting like the world’s best traders when in reality the best traders on here don’t have to fill their ego or increase their follower count to make themselves feel relevant in this space.
Look at @cobie, @GCRClassic, @GameOfDeception, @davthewave, - fuck even the boy @trading_axe.
Yeah they post, yeah they are calling higher, yeah they engineer engagement but tell me the last time you saw them flaunt a PnL?
If you’ve been here long enough, you see that when someone truly makes it in this space, they tend to fade off into the abyss, because at the end of the day, if you had life changing money, would you really still be spending your time on here?
But hey, if you’re taking advise from someone named “Shardi B” - maybe you’re not intended to make it.
Ok guys, I showed you 6k to 600k yesterday
TODAY, IN ONE DAY
We are now 6k to 1.1m
WHO IS YOUR QUEEN?
16
5
154
53,182
XBT-Skills (4ed7c36) now supports Opus 5.5 on medium, high, and max effort levels with ultracode support.
Changelog:
- Opus 5.5 measured end to end and given its own layer
- The default reviewer seat moved to Opus 5.5
- The skills control run ran and published
- Small tweaks on ELI5 and redman skills
GitHub repo: github.com/XBTDMT/xbt-skills…
96
Will drop Opus 5.5 update tomorrow... But for now it looks very good.
Making my skills pack for Claude public. It focuses primarly on Opus and Fable.
Composed of a harness and suite that aims to do 2 things:
- Optimize cost vs quality
- Reduce hallucinations in long lasting sessions
Included is an internal benchmark showing success on both fronts for 4 out of 5 Anthropic models.
The eval tool is included in the repo so new models can be tested and tuned in the same fashion.
GitHub repo: github.com/XBTDMT/xbt-skills…
147
Making my skills pack for Claude public. It focuses primarly on Opus and Fable.
Composed of a harness and suite that aims to do 2 things:
- Optimize cost vs quality
- Reduce hallucinations in long lasting sessions
Included is an internal benchmark showing success on both fronts for 4 out of 5 Anthropic models.
The eval tool is included in the repo so new models can be tested and tuned in the same fashion.
GitHub repo: github.com/XBTDMT/xbt-skills…
146
I think for daily use Opus high still wins but can not include workflow Fable 5.1 low in researcher/cache intensive needs.
1
74
Simple layer makes massive improvement for everyday use and costs if it is tailored to you:
- Ask Claude to audit your skills, plugins and clean.
- Ask Clause to read your repos and analyze how you work.
- Explain how you work - the more detail the better. Use @WisprFlow to make your life 10x easier.
- Ask it to build you doctrines for each model and run a loop of improvement on each.
Then deploy into some form of blind testing enviorment internally that allows you to keep score and update score.
Fwiw it's not a cheap up front cost to benchmark multiple models you use. I spent multiple max subscriptions on the initial tests. Now I just have to test on latest models as they come out.
42
