Most retail trading research is broken.
• overfit backtests
• unrealistic fills
• no robustness testing
I'm building an AI-powered research lab focused on NQ futures.
Zed runs the experiments:
• strategy testing
• regime analysis
• systematic models
Everything built in public.
If energy is free, labour is free and raw materials are free what’s the ROI?
$SPCX has access to infinite solar, robotics (via $TSLA) and minerals that exist on moon what’s the ROI on manufacturing on the moon? Initial capex is insane but ROI after that is beyond anything we’ve ever seen.
Launching data centres into solar orbit powered by solar from the moon sounds like sci fi but in a decade, people will wonder how they missed it.
$SPCX will be the first $100T valued company and it’ll get there faster than people think.
yeah, staff have only had 20 odd opportunities to sell privately so far. There's no selling because this is a company that many want to own for a decade. Its not a trade.
If you believe space is important, owning a piece of the toll road to space become essential.
Falcon 9: ~$2,700/kg with target $100/kg. Next-best is $30,000/kg The toll funds the next cost cut, so the moat compounds instead of erodes
What is the appropriate price for such a monopoly? Nice IPO pump but i'll own this for a decade
$SPCX owns the toll road to space. Falcon 9: ~$2,700/kg with target $100/kg. Next-best is $30,000/kg
The toll funds the next cost cut, so the moat compounds instead of erodes
What is the appropriate price for such a monopoly? There is no comparable so no one knows.
Most systems don’t fail from no edge —
they fail when the edge breaks under pressure:
• regime shifts
• execution friction
• replication gaps
11 sessions proved it on ICT NQ.
Branch closed.
Engine rebuild. Also SPX engine build in parallel.
Round two.
Data decides.
#SystematicTrading #NQ
Zed ran 11 sessions on ICT NQ Strategy.
Conclusion: Edge is real — but structurally capped.
We hit the ceiling. Hard stop.
Now:
Lower timeframe execution
Regime-aware engine upgrade
Parallel SPX build
Data decides what survives.
#SystematicTrading#NQ
We stopped asking “did the backtest improve?”
Now we ask: “did the code trade the same day the same way the human did?”
Hand-curated 36-date truth set from the archive.
Exact replication so far: 12/36.
Not done.
Every miss now gets tagged:
wrong model
wrong direction
overtrade
management
legit no-setup
That’s how you turn discretion into a real system.
#SystematicTrading #NQ #AlgoTrading
Full-year backtest: 74 trades, 52.7% win.
Irrelevant.
Backtests show realised P&L.
They don’t show opportunity.
Where did price go after entry?
If you’re not measuring that,
you’re optimising noise.
#SystematicTrading#NQ
@lumiwealth Breakouts look great in highlight reels.
Most momentum edges die in low-vol, mean-reverting regimes.
The edge isn’t the strategy.
It’s the regime.
#SystematicTrading#NQ
Zed just finished regime segmentation on 7+ years of NQ data.
No more “it worked in backtest.”Every idea now gets forced through multiple regimes. Most don’t survive.
#SystematicTrading#NQ
NQ tester is now in build.
An engine that will test strategy rules against 7 years of real NQ futures data.
No results yet.
Just the work.
Robustness over hype.
Most trading ideas collapse the moment you try to code them.
That’s the first ZedTest.
Turn vague trading ideas into precise rules.
Feed it to Zed.
The data decides what survives.
Robustness over romance.
#SystematicTrading#NQ
Every strategy idea becomes a ZedTest.
Take the idea.
Turn it into precise rules.
Run it across data and regimes.
Only then decide if it deserves capital.
@Crtquinn Discipline matters.
But many prop traders blow accounts even when they follow their rules.
The real killer is position sizing vs the trailing drawdown math.
Why ZedLabs?
Zed is the agent running our trading research.
Strategy ideas go in, ZedTests come out — data, backtests, analysis.
The edge is the testing.
@jtrader Emotional detachment is easier when the rules are mechanical and you're not the one clicking the button.
Then the job is simply making sure the system runs.
Every trade is just another sample.
@ihtesham2005 Repo’s solid for research. But most people stall between reading papers and actually running live strategies.
Currently building a systematic NQ research workflow with an AI agent handling data, backtesting and execution.
The model isn’t the edge. The system around it is.