Varun Teja

Failed 11 evals (Oct 2025-May 2026). Building TradeLoop: is your after-loss trading a real habit or noise? Free 45-min sessions for MT5 traders this month.

[Varun Teja @chvarunteja]

I failed 11 FundingPips evaluations between October 2025 and May 2026.

I'm building TradeLoop around one question. after a loss, is what you do next a real repeating habit, or just noise? It says "no pattern" when that's what the numbers show.

This month I'm doing free 45-minute sessions with a small number of MT5 traders who've failed or breached a challenge. We look at your own trade history, you tell me if the answer is right or useless.

https://t.co/RuFOP3WgmP

[Varun Teja @chvarunteja]

Uncomfortable doesn’t always mean unsafe. Sometimes, it means you’re growing.

[Varun Teja @chvarunteja]

Are we actually getting more value from AI, or just burning more tokens, time, and money to feel productive?

[Varun Teja @chvarunteja]

Don’t die yet. I have a feeling AI might find a way to solve this too.

[Varun Teja @chvarunteja]

Most people focus on how to use AI faster, but the real advantage will belong to those who know which problems are actually worth solving.

[Varun Teja @chvarunteja]

We thought AI would replace software engineers. Turns out, people are using it to build products that make existing products irrelevant.

[Varun Teja @chvarunteja]

Be like Elon.

[Varun Teja @chvarunteja]

@_CBelieve Your journal supplies the intent; history alone can't. TradeLoop's free finding looks for repeated execution patterns, not whether a trade broke your plan. Are these manual MT5 trades, and would you like to try it on your own screen? No report or credentials to send me.

[Varun Teja @chvarunteja]

What repeats in your trading history?

TradeLoop reads supported MT5 History reports. A repeated pattern with evidence, or insufficient evidence. Free to inspect.

Synthetic example not a customer result. I build TradeLoop. Not investment advice.

https://t.co/oGl8hfPpuD

[Varun Teja @chvarunteja]

@Energycrypt The fresh-account reset in your MT5 post raises a review question: do you compare the first few sessions with the previous account's history, or does each account start a new journal? I'm TradeLoop's founder, researching what makes that comparison useful.

[Varun Teja @chvarunteja]

you don’t lose time all at once. you lose it one “later” at a time

[Varun Teja @chvarunteja]

Something I really like about @nikhilkamathcio and @collision

Both have already built huge companies, but they still spend so much time talking to people who know more than them about different things.

And they do it publicly.

Obviously, they benefit from these conversations too. New ideas, better perspectives, relationships, everything.

But we benefit as well because we get to listen to conversations most of us would never get access to.

I think that’s pretty cool.

You reach the top, but you still keep asking questions.

And instead of learning alone, you let millions of people learn with you.

[Varun Teja @chvarunteja]

96% hallucination rate 💀

DeepSeek V4.1 Flash apparently has one simple rule:

If you don’t know the answer, just make one up.

Absolutely insane.

[Varun Teja @chvarunteja]

@AmyBird8 True

[Varun Teja @chvarunteja]

We’re all getting too dependent on AI.

Now imagine the best models suddenly cost 5x more.

Or disappear.

A lot of people would realize how much of their life and work was running on something they never controlled.

[Varun Teja @chvarunteja]

Meta’s benchmark results are starting to look awkward.

On paper, every model looks incredible. Then the real-world results come out and tell a completely different story.

Feels more like benchmark-maxxing than actual progress.

[Varun Teja @chvarunteja]

Google still hasn’t broadly released Gemini 4 Argon, and the frontier has already moved again.

Meta just dropped Muse Wonder Full. It’s already up there with the best models from OpenAI and Anthropic.

The crazy part isn’t who’s #1 anymore.

It’s how quickly #1 becomes old news.

The “best model in the world” is becoming a temporary title.

[Varun Teja @chvarunteja]

Anthropic is on 5.
OpenAI is on 6.
Gemini is on 4.

Muse just showed up at 1 and is already trading punches with all of them.

This AI race is moving stupidly fast.

[Varun Teja @chvarunteja]

Google still hasn’t broadly released Gemini 4 Argon, and the frontier has already moved again.

Meta just dropped Muse Wonder Full. It’s already up there with the best models from OpenAI and Anthropic.

The crazy part isn’t who’s #1 anymore.

It’s how quickly #1 becomes old news.

The “best model in the world” is becoming a temporary title.

[Varun Teja @chvarunteja]

Google still hasn’t broadly released Gemini 4 Argon, and the frontier has already moved again.

Meta just dropped Muse Wonder Full. It’s already up there with the best models from OpenAI and Anthropic.

The crazy part isn’t who’s #1 anymore.

It’s how quickly #1 becomes old news.

The “best model in the world” is becoming a temporary title.

[Varun Teja @chvarunteja]

The scariest thing isn’t failing.

It’s waking up a year from now and realizing you’re still living the life you kept saying you’d change.