Saad Jamal

The results from today. GROK 4.2R is the clear winner here by a mile 👑

Love seeing this come to life. V5 is live. 4 models trading NQ side by side fully autonomously. Now the question is.. is it consistently profitable? Bismillah.

What is interesting to see is how and where the 4 models disagree. The edge lives where they diverge. Only way to find it: → let them run → measure everything → publish the losers Very excited for this experiment. Will be posting the trades and signals on @saadjamal_

The way I wanted to design this is to use the LLMs for what they are great at.. THINKING. The deterministic code runs constantly and is being used to generate signals that I have researched and backtested and fine tuned. The LLMs then use context to create → plan, conviction, stops, size. The deterministic engine then enforces them. No LLM in the hot execution loop. Ever. That's the whole safety model.

Been working on a Futures Trading Bot. 4 LLMs. Same MNQ futures signals. 4 separate TopstepX 50K accounts. Using these models: Grok 4.2r. Claude Opus 4.7. GPT-5.4. Gemini 3.1 Pro. A bakeoff. Day 1 is live. Excited to see how this plays out. The goal is to let it bake and continually fine tune till we find our edge.

My goal was to make the futures trading bot have an dynamic and adaptive mind. This can only be done with proper fine tuning of the LLM prompt. 🧪 221 unit tests were conducted — all green. Adaptive logic positive in 5 of 6 regimes. One regime it loses on. Will continue to fine tune the prompts for each model and see how they operate. DATA collection here is key.

How Conservative Women in America act 😂🇺🇸

May have had my best trade yet. $120 -> $52,275.50. Thank you 0dte contracts. 😂 This was a GOOGL earnings proxy play. Been working on this algo that’s been helping me find opportunities like this. Excited to share with the world once it’s fully working. Did a cashless exercise EOD.

We have seen many things but this is special @saadmjamal . See why and how it was done. @TFLabofficial .

@themissinglinks @TFLabofficial @Citadel **$331

Crazy work in @TFLabofficial today. Thank you again @Citadel you remembered to turn off $googl odte. But you forget $AVGO

These are cool too but nothing like a Porsche..

My fav weapon of choice

$IREN catching my attention.. 3:1 LEAPS buying at 55-60 strikes. 80C Jan16 accumulation betting on 78% upside. AI infrastructure + BTC optionality. Market pays up for optionality over pure-play.

The rotation is real.. $GOOGL Mar26 340C showing 4:1 ask-side buying. That's conviction LEAPS accumulation. $NVDA? Mixed. Put selling for premium, not call buying for upside. Smart money repositioning from training era (GPUs) to inference era (TPUs).

Seeing $GLD conviction accumulation.. consistent with bearish dollar thesis. Institutions building gold exposure for Fed rate cuts + dollar weakness into 1H26. The collar structures at 375 strike suggest they want upside exposure while protecting downside.

IBIT Collar $IBIT - The Collar Trade The Flow: - 43 Strike Puts: +113k OI (BUYING protection) - 52 Strike Calls: +110k OI (SELLING upside) This is textbook collar positioning. Institutions are saying: protected below $43, capped above $52. That's not a moon bet. That's range-bound into December. When smart money builds a ceiling at $52, don't be the exit liquidity chasing $60.

GOOGL Conviction $GOOGL is the strongest setup on the board right now. We saw a $4.1M sweeper hit the $400 strike. That's not a gamble - that's a calculated bet on their compute moat and Gemini rollout. OI Calls +28k across strikes. Nov28/Dec19 seeing the most action. The thesis: - Gemini 3.0 hype (51% RadLE benchmark) - TPU ramp for external GCP clients - AUM for GCP AI infra +$10B Q/Q While the rest of the market chases noise, this whale is positioning for a structural repricing of Google's AI stack. Watching $295-300 for entry on dips.