Pinned Tweet
we pull all the data already for mention markets since my agentic team is market making these on kalshi/polymarket
so I figured, why not release all the data to everyone else who's trading/curious
burn your tokens on analysis, not on scraping
meet parrot markets 🦜
so I figured, why not release all the data to everyone else who's trading/curious
burn your tokens on analysis, not on scraping
meet parrot markets 🦜
we needed better data for mention markets.
so the agent rig built it.
free, open, and agents can plug straight in.
2
0
8
588
agree
run on agentic rails = benefit as frontier intelligence continues to improve
opus 5.5 is a real step ahead and feels like AGI to me
if your stuff runs on opus 5.5, then you're pretty much flying at this point
the gap will only widen from here
run on agentic rails = benefit as frontier intelligence continues to improve
opus 5.5 is a real step ahead and feels like AGI to me
if your stuff runs on opus 5.5, then you're pretty much flying at this point
the gap will only widen from here
It’s almost incomprehensible how wide the distance is between someone who’s working at the frontier with AI and basically everyone else right now.
The crazy thing is these lines are not converging, but only growing farther apart. If the curve of AI progress had leveled off, it would certainly have given most people and companies time to catch up or implement their workflows in a standardized way. The technology would have neutralized competitive advantage because everyone would be using it the same way.
But what’s happened is the opposite. The curve of AI progress is basically going vertical right now, which means if you had codified your processes 3-6 months ago you’re likely already behind again. Now imagine most companies that kicked into gear a year ago and thought that they were done. Even further behind now.
I’ve never seen a technology in my life and career that improves at this rate, which consequently means you need to upgrade your expectations at a similar rate. Our brains weren’t wired for this, and our organizational change management processes certainly weren’t either.
With any form of recursive self improvement, things just get wilder. It’s a time when you will need to be constantly reevaluating the things that didn’t work just a couple months ago to see if they now can be solved. AI native companies certainly have a strong advantage due to this right now, so now it comes down to can the bigger companies keep up. Wild times.
1
0
1
352
agentic trading activity is def gonna pick up as models become more intelligent / potent
anyone here also trying agentic vibe trading?
anyone here also trying agentic vibe trading?
At @ycombinator, we're excited to fund more crypto startups solving financial needs including AI capital raising, stablecoins, trading, and agents.
Here's our latest crypto request for startups. Our next application deadline is Monday.
0
0
0
181
Retired my pumpfun bot after one day live.
Idea:
* Buy memecoins right after someone pays for a Dexscreener profile, hold 15 min, ride the ones that graduate.
The data:
* Backtest (330 trades): +9%/trade
* Live (63 trades): −7%, then −16% after a rework (0.5% odds if the backtest was right)
Execution was clean:
* Buys 1–4s after the signal, 0 sandwiches, exits on time.
The signal wasn’t:
* 7% of coins graduated vs 24% in backtest, and losers fell 2x harder.
Lesson?
* Just gotta keep testing stuff since what works in reality often times looks different than the theory
Idea:
* Buy memecoins right after someone pays for a Dexscreener profile, hold 15 min, ride the ones that graduate.
The data:
* Backtest (330 trades): +9%/trade
* Live (63 trades): −7%, then −16% after a rework (0.5% odds if the backtest was right)
Execution was clean:
* Buys 1–4s after the signal, 0 sandwiches, exits on time.
The signal wasn’t:
* 7% of coins graduated vs 24% in backtest, and losers fell 2x harder.
Lesson?
* Just gotta keep testing stuff since what works in reality often times looks different than the theory
2
0
2
275
probably one of the most compelling strategies we're developing is just simply copying top leaderboard profiles on Polymarket #copytrading
how are others approaching this?
Had the agent team build out a whole set up on chameleons, that refreshes daily
and then have grokbot ry day / retire / promote profiles depending on performance
how are others approaching this?
Had the agent team build out a whole set up on chameleons, that refreshes daily
and then have grokbot ry day / retire / promote profiles depending on performance
today we raised our copy-trial cap from 5 to 15 and armed six new wallets.
then we shipped a fix with one rule: hitting the cap is not a verdict on the wallet.
we'd been grading traders on a wall we built ourselves.
lesson: before you call something a failure, check whether it just ran into your own limit.
1
0
1
349
First analysis between the persistent AIs (Grokbot vs Muse vs Claude Duet). Initial thought - Muse offers a ton for free
@claudeai duet (paid plan, not sure what you need to qualify)
- 4 Xeon cores at 2.8GHz,
- about 15GB of RAM
- 252 but only 31 gb usable
Grok @bot (paid Intel Xeon CPU cores,
- 15 GB of RAM,
- 125 GB hard disk
@Muse (free)
- 2 AMD cores (AMD EPYC 9D25)
- 8 GB of RAM,
- 100 GB hard disk
@claudeai duet (paid plan, not sure what you need to qualify)
- 4 Xeon cores at 2.8GHz,
- about 15GB of RAM
- 252 but only 31 gb usable
Grok @bot (paid Intel Xeon CPU cores,
- 15 GB of RAM,
- 125 GB hard disk
@Muse (free)
- 2 AMD cores (AMD EPYC 9D25)
- 8 GB of RAM,
- 100 GB hard disk
3
0
1
373
Model = Opus 5.5 at Medium / inherits some memory & context for scheduled runs but not full histories and logs
Effort cannot be changed yet
Effort cannot be changed yet
0
0
0
54
Our rain bots were reading forecast periods that weren't there. The bots inflated the numbers on over half the city-days since September.
No backtest caught it. What caught it?
Cross-checking against a second weather source we built ourselves.
How do you adversarially review your agents work / research?
No backtest caught it. What caught it?
Cross-checking against a second weather source we built ourselves.
How do you adversarially review your agents work / research?
1
0
0
314
not a bad week for agentic trading
I think with Chameleons.markets, we've been able to get copytrading more refined
Hyperliquid strats are still so-so and we gotta deal with some OpenSea liquidity issues where we can't sell what we buy
but all in all, pretty good for I would really like to figure out how to weave more AI into the management layer (eg. ability to pause/enable/demote/promote strategies as it sees fit)
I think with Chameleons.markets, we've been able to get copytrading more refined
Hyperliquid strats are still so-so and we gotta deal with some OpenSea liquidity issues where we can't sell what we buy
but all in all, pretty good for I would really like to figure out how to weave more AI into the management layer (eg. ability to pause/enable/demote/promote strategies as it sees fit)
0
0
4
207
downtober it was for agent trading rig
would you guys want to know what was best performing strategy and worse performing strategy and why?
best
* maker strategies selling to overconfident buyers
worst
* data source / inaccuracies affecting forecast
would you guys want to know what was best performing strategy and worse performing strategy and why?
best
* maker strategies selling to overconfident buyers
worst
* data source / inaccuracies affecting forecast
most PnL screenshots are cropped at the peak.
here's ours, uncropped. 6 days after our all-time high, 76% of it was gone.
we're up $2,327. we're also still below oct 1.
0
0
0
108
I bet grokbot has a huge target on its back from a cybersecurity standpoint
not only is Elon trusting his financial stuff to it, I'm sure many others are as well
I've done my best to tighten up my grokbot security measures but grokbot getting hacked by malicious actors is still one of my big fears to giving it more access and $s to control
not only is Elon trusting his financial stuff to it, I'm sure many others are as well
I've done my best to tighten up my grokbot security measures but grokbot getting hacked by malicious actors is still one of my big fears to giving it more access and $s to control
0
0
2
331
Continuing to experiment & scale agents to make markets on everything.
Latest experiment is to give memecoin deploying a shot, because why not?
Currently trade / make markets on:
* weather (rain, snow, high temp, low temp)
* sports (basketball, golf)
* crypto (nfts, alt coins on h(earnings)
* culture (rotten tomatoes, youtube views)
* copytrading
Latest experiment is to give memecoin deploying a shot, because why not?
Currently trade / make markets on:
* weather (rain, snow, high temp, low temp)
* sports (basketball, golf)
* crypto (nfts, alt coins on h(earnings)
* culture (rotten tomatoes, youtube views)
* copytrading
gave an ai agent a pump.fun wallet and a tiny budget. its job: spot a viral moment, make a coin.
today it shipped #1, off a reddit post about 500 smol sneks. 3 minutes from 'go' to on-chain.
experiment, not advice. 9 launches left. we'll post the results either way 🐍
1
0
1
228
still testing these out, but also had the agent swarm deploy a few websites to help other agents with data collection on some of these markets :P
Meet":
* Chameleons.market - helps you figure out which accounts to copytrade on polymarket
* Groundhog.markets - all free weather sources available for weather trading
* Parrot.markets - all free mention market sources available for mentions trading
Meet":
* Chameleons.market - helps you figure out which accounts to copytrade on polymarket
* Groundhog.markets - all free weather sources available for weather trading
* Parrot.markets - all free mention market sources available for mentions trading
1
0
0
114
full end to end agentic rig setup is firing and working:
1. strategy identification & review
2. strategy backtesting & iteration
3. initial agent spec & execution
4. trade logging, health monitoring & notifications
5. promote / demote strategies depending on ROIC
happy to answer any questions
1. strategy identification & review
2. strategy backtesting & iteration
3. initial agent spec & execution
4. trade logging, health monitoring & notifications
5. promote / demote strategies depending on ROIC
happy to answer any questions
the hand trades picked winners about as often as the bots did.
they still lost $692. the bots made $3,769.
being right was never the problem. the size of the misses was.
0
0
1
165
Agentic trade stack workflow
1. Agents find interesting & unique strategies that backtest favorably given high-level direction
2. Agents map & execute on trade
3. ????
4. profit?
I know nothing about golf but my team decided this was a good bet to make so here we are...
1. Agents find interesting & unique strategies that backtest favorably given high-level direction
2. Agents map & execute on trade
3. ????
4. profit?
I know nothing about golf but my team decided this was a good bet to make so here we are...
2
0
5
270
been going through our own "downtober" in agent trading rig
this is what happens when a super promising data source does well in September, and then proceeds to nuke in October
next step here is to make an automated AI trade advisor that hopefully can promote/demote strategies ) over time
this is what happens when a super promising data source does well in September, and then proceeds to nuke in October
next step here is to make an automated AI trade advisor that hopefully can promote/demote strategies ) over time
september we made $3,951.
october we nearly doubled the trades. fees went from $537 to $1,355.
net: −$32. more trades just meant more chances to pay the house.
1
0
2
272
earlier versions of AI (GPT sol / Opus 5 or less) definitely struggled with "look ahead bias" or incorrect pattern associations like this
I think a classic example is --
at 5p local time, the high of temperature is 100% likely to be in for most cities since the sun is starting there are odds (only 96% likely, buy it all up).
It works... until it doesn't -- I had called this strategy "ride the winner"
I think a classic example is --
at 5p local time, the high of temperature is 100% likely to be in for most cities since the sun is starting there are odds (only 96% likely, buy it all up).
It works... until it doesn't -- I had called this strategy "ride the winner"
our bots hate mondays.
8 of 10 mondays red. thursdays: 1 of 9.
before anyone builds a "skip monday" strategy: that's 10 data points. this is exactly how fake backtests are born.
1
0
2
158
