Joan Carreras

It's 3am. Your AI agents are still testing trade ideas. You're asleep. That's the point.

Performance marketing consultant👇

[Joan Carreras @joancarreras_]

Had Jarvis take an initial stab at this using Swing Surfer's API.

Daily TF

11 total symbols, split across 2 markets/universes:

Crypto:

• BTC
• ETH
• SOL
• XRP
• DOGE

Equities / ETFs:

• SPY
• QQQ
• AAPL
• MSFT
• NVDA
• TSLA

Daily timeframe, 2021-08-21 to 2026-08-21.

Tested multiple variations of ATR multiples, fixed %, candle-low stops on these 11 symbols. Fixed 1.5% led avg return; 2% looked best balanced.

Initial conclusion is that tight stops helped, but not comfy: low win rate/high stop hits/DD.

However, it shows enough promise to do a full full-universe + asset vol-adjust research!

Thanks for sharing, dude!

[Joan Carreras @joancarreras_]

Had Jarvis take an initial stab at this using Swing Surfer's API.

Daily TF

11 total symbols, split across 2 markets/universes:

Crypto:

• BTC
• ETH
• SOL
• XRP
• DOGE

Equities / ETFs:

• SPY
• QQQ
• AAPL
• MSFT
• NVDA
• TSLA

Daily timeframe, 2021-08-21 to 2026-08-21.

Tested multiple variations of ATR multiples, fixed %, candle-low stops on these 11 symbols. Fixed 1.5% led avg return; 2% looked best balanced.

Initial conclusion is that tight stops helped, but not comfy: low win rate/high stop hits/DD.

However, it shows enough promise to do a full full-universe + asset vol-adjust research!

Thanks for sharing, dude!

[Joan Carreras @joancarreras_]

Going to backtest this approach. Been using mostly lower ATR band variations as initial stop loss placement, but we'll see about this!

From similar backtesting I did in the past, I remember that this tight of a stop loss tends to cause lots of failed reentries and catching skewed uptrends that happen once in the whole sample size to make back loses.

Will report back!

[Joan Carreras @joancarreras_]

Nah... This is keeping GTA VI interest high with everybody keeping an eye on these leaks between announcements and showing content pre-release.

You will then get to see the Extended Look, love it and go crazy.

The game will release and all of these leaks reactions "it's so over for Rockstar", "they messed up" will all be forgotten and shadowed by the game's success.

Just compare the interest in Google Trends Youtube and Search from Aug 6th Extended Look announcement vs the increase now.

[Joan Carreras @joancarreras_]

@TheOneLanceB https://t.co/QHfIsdS7FT

[Joan Carreras @joancarreras_]

$TTWO Typical post you see before meaningful bounce happens haha

[internet hall of fame @InternetH0F]

Take-Two has lost around $2.8 billion in market value following the recent 'GTA 6' leaks

[Joan Carreras @joancarreras_]

Between yesterday's $TTWO earnings call and this tweet...

GTA VI is locked and loaded for November 19th. LFG.

I highly doubt that we'll close below the W neckline bottom around $220 during this correction nor leading up to the launch. Great place to reload.

- Conservative target $280-$310
- Moon target $380-$400

[XBOX @XBOX]

November 19th

Wishlist Now: https://t.co/O1RVROCl2Z

[Joan Carreras @joancarreras_]

Earnings call donzo. Can't really ask for more than this.

Got the two things I wanted for the $TTWO weekly close.

✅ Not closing below 228.5
✅ Closing above 241-243, confirming HTF higher low

It even printed a green hammer with a big lower wick.

Moon target of 380-400 seems pretty hard to imagine hitting pre-release atm, but who knows...

Get an unprecedented "extended look" reception with hype building up + degens using leverage on these on-chain platforms tokenizing the stock, and as we get close to the Grand Theft Auto VI release, things could get silly.

I think it has a decent chance to become a meme stock.

All fugazi and especulation for now.

280-300 conservative target seems pretty doable given current position we are in and the slope of the uptrend from the bottom to the target.

Vice City, Leonida Keys await us, gents.

[Joan Carreras @joancarreras_]

@petergyang I don't even bother looking at outputs until there's a fresh agent without context that has been asked if it agrees with the plan, and then once implemented multiple reviews until all pass.

[Joan Carreras @joancarreras_]

@LindaRaschke Yup

And lots of times price first shows its hand and then catalyst gets announced afterwards.

[Joan Carreras @joancarreras_]

Duuude, I think I've been running these model tests wrong, and it took watching GLM 5.3 burn $5.53 to notice 😭

I told it to find a swing strategy that isn't already in my saved list. So it read all 196 templates, saw that mean reversion, momentum, breakouts and gap plays were already covered, and decided the only thing nobody had touched was calendar stuff. Which is exactly where it spent the budget.

Turn of month on equities, then the crypto weekend effect. Both killed.

Grok 4.6 said the same thing about the same account last week: my strategy list already has the good edges in it, so all that was left for it to try was the weak stuff.

So the rule I thought was making this harder maybe just sent every model straight to the ideas that don't work?
Next model that drops, I'm running it on a virgin account with no history, and I want to see which edges it reaches for when nothing is off the table.

Maybe I could also point one at my existing strategies and tell it to make those better instead. Then I'd have two tests, one for finding brand new edges and one for improving the ones I already trust.

I could make a benchmark out of this series, uhmm. Thoughts?

Oh well, I guess these videos have been more of a sniff test than proper model reviews, but at least I was swimming enjoying summer time 🏊

[Joan Carreras @joancarreras_]

@PKycek @joyflow4 Agreed, the more open hours for NASDAQ, the more non-Americas regions can trade with better liquidity during their normal waking hours. It's good for business, and I guess NA, Wall Street, etc will adapt with more automated systems and less discretionary trading.

[Joan Carreras @joancarreras_]

@TedHZhang Basically survive and keep playing the game long enough so when good times comes, you capitalize on it 🤙

[Joan Carreras @joancarreras_]

@CastilloTrading The final shot is a gem.

Kudos to Clemente, Leonardo DiCaprio has nothing on him 😂

[Joan Carreras @joancarreras_]

@JoachimMo1985 @PKycek Is that because you focus some of your strats on these auction times specifically?

[Joan Carreras @joancarreras_]

@PKycek Haha pre-AI era I thought I was a hacker getting some pinescripts up and running.

Then AI started to get smarter about a year ago, and realized how little I knew about statistically sound methods that quant firms use 🥲

[Joan Carreras @joancarreras_]

@lonextrades 100%

I keep seing content online of some people saying that they don't use stop losses and they just have a mental stop loss.

That is all good and dandy until that one day where you get a crazy move you weren't expecting and get wrecked.

Speaking from experience haha

[Joan Carreras @joancarreras_]

@TheFlowHorse Nice, remove the brain rotted audience that need the flashy stuff to keep watching.

They are probably the ones that give you the most headache anyways 😂

[Joan Carreras @joancarreras_]

Yep, get used to stepping on the gas, feeling like you are playing with profits, etc

That's why I started using lower ATR band (for longs) as a way to size my positions.

Usually getting the massive win tends to be because you enter a position that had low volatility at entry, thus you size accordingly because you have a tighter stop loss. After it's done it probably made a substantial move and the ATR volatility is way higher.

When using this method, it allows you to position smaller and gives more wiggle room for the stop loss.

This can also be paired with: Hey if prior win was a RR of more than X and volatility has spiked more than Y then the dynamic position size should be smaller.

[Joan Carreras @joancarreras_]

@elonmusk Haha I feel like that when the agent is going mid task and I add an additional request like "btw not interrupting the main task, but do this too and don't forget to X, Y and Z" 😂

[Joan Carreras @joancarreras_]

It feels dumb at first, but that is why I long when momentum algo tells me to even if below resistance.

Sometimes you get these pumps outta nowhere and the only signal you had was volatility compression into slighty higher lows that tick up the trend metrics just enough to give the signal.

Vamosss Duck