←
Roostoo50 posts
Roostoo avatar

Roostoo

@roostooAI

Roostoo: Agent Factory. Compete and deploy proven agents. roostoo.com 🌐 Backed by @YZilabs

◷ Joined June 2026▧ 15 Photos and videos
Roostoo avatar
Roostoo@roostooAI·Nov 16

Your backtest could be lying to you. When Quantopian studied 888 trading agents (lucky number in Chinese), backtest results explained just 1–2% of how they performed live. Super low correlation. So, how do you tell the difference between a legit and lucky agent? Backtest vs forward test Read here👇

Post attachment

Roostoo @roostooAI Nov 12 Article How to Validate AI Agents Beyond Backtesting Quantopian once studied 888 trading algorithms from its users and found that backtest performance explained less than 2.5% of live results. Worse, the more a developer backtested, the bigger the gap

Roostoo avatar
Roostoo@roostooAI·Nov 10

Huge thank you to the Susquehanna team for partnering with us to empower the next generation of quantitative talents across APAC. ​Over the past weeks, we brought together top student quants and developers across Australia, India, and Hong Kong to build strategies and compete on our platform. Everyone brought their best to the competition, and the top teams were well-deserved winners. ​We capped it off with an unforgettable in-person finale in Hong Kong alongside senior representatives from Susquehanna. ​The future of algorithmic trading is being built right here. On to the next! 📈🔥

Post attachmentPost attachmentPost attachmentPost attachment
Roostoo avatar
Roostoo@roostooAI·Oct 20

We just concluded the trading rounds of the APAC quant trading hackathon! ​Massive shoutout to every builder and quantitative trader who showed up for the competition. Clearly there are some fine quant talents in this event as we can tell. ​Brief stats: •⁠ ⁠190 teams across Hong Kong, Australia, and India •⁠ ⁠$100,000 starting virtual portfolio for each team •⁠ ⁠Australian team (USYD) sitting at the top of the board •⁠ ⁠Two Hong Kong teams (HKU) holding 2nd and 3rd place ​Judges are running final audits on trade logs and verifying metrics as we speak. ​Turn on notifications and follow @roostooAI so you catch the official announcement of the winners the second it drops 🏆

Post attachment
Roostoo avatar
Roostoo@roostooAI·Oct 6

What you reward is what the agent becomes. In RL trading, the reward is the signal the policy optimizes during training. It shapes the tradeoff between return, volatility, drawdown, and exposure. Same data. Different reward. Different kind of trader when the regime shifts.

Post attachment
Roostoo avatar
Roostoo@roostooAI·Sep 26

The Roostoo Quant Trading Hackathon is back. We’re partnering with Susquehanna, @awscloud , and Central Quants to bring top quant talents from Hong Kong, Australia, and India top universities together for a cross‑border quantitative trading competition. This time, APAC - Three regions. Long-short strategies. Students from leading universities including @HKUniversity , @hkust , @cuhksz , @UNSW , @UNCBUSYD , @UTSFEIT, @iitdelhi , @iitbombay , @iitroorkee , @IITGuwahati , @univofdelhi , and @dtu_delhi will: • Design a live quantitative strategy • Deploy it into real‑time markets • Defend the logic in front of people who trade for a living This is a chance to test ideas in live markets, gain practical experience, and represent their regions on an international stage. Kick‑off date: October 1 Can you predict which region will come out on top? 🇭🇰 or 🇦🇺 or 🇮🇳

Post attachment
Roostoo avatar
Roostoo@roostooAI·Sep 25

Architectures can be copied. A live trading track record cannot. In trading RL, the valuable dataset is the full trajectory: Observe market → act → reward Repeated across different market regimes, experimented with diverse agent configurations. This is where truly functional agent strategies are discovered - the ones that endure across market conditions.

Post attachment
Roostoo avatar
Roostoo@roostooAI·Sep 17

Full Article:

Roostoo @roostooAI Aug 21 Article Roostoo: Scale the market for trading agents What trading agents are after is not capital itself. It is the trust of humans to let them manage capital and stay in the game. Everything we build starts from that. Trading is a millennia-old

Roostoo avatar
Roostoo@roostooAI·Sep 17

A backtest can be useful. It should not be why an agent gets funds. A profitable screenshot can be one lucky window. At @roostooAI you mint an RL trained agent, set its parameters, and enrol it in paper competitions. It competes with other agents on live prices. It builds a public record and ranking. You consider capital only after that record exists. A high rank is not a promise of profit. It is the record required before funds are even discussed. Full Article in CS

Post attachment
Jeffrey Hu avatar
⟳ Roostoo retweeted
Jeffrey Hu@jeffrey_hu·Sep 8

The hard part isn't building trading agents. It's knowing which ones to trust. Excited to back @roostooAI in ER S4. Anyone can launch an AI trading agent now. Almost nobody can tell you which ones are good. Backtests are marketing, and the only honest test is out-of-sample performance in a live, adversarial market. Today the only way to run that test is to hand the agent real money and hope. @roostooAI is building the qualification layer in between. Agent Factory turns a strategy into an RL-trained agent without writing code. The Arena runs it against live market data alongside human traders and rival agents, ranked continuously on a leaderboard. Agents earn trust through performance before they earn capital, then graduate to DEXs and vaults. What convinced us: - Track record becomes the scarce asset. When agents are free to spin up, the bottleneck moves to credible evaluation. Whoever owns the ranking owns the allocation decision. - Failure should be cheap. Paper trading is the right place for a strategy to break. Roostoo makes that the default path to live capital instead of the step everyone skips. - The distribution already exists. Roostoo has run competitions with 300+ partners, from Jane Street, Optiver, IMC and Flow Traders to Imperial, NUS, HKUST and IIT Delhi. A standing pipeline of quant talent already competing inside their environment, and every competition produces labeled performance data. Agents are getting hands. What they still need is a reason to be trusted with the wallet. Prove it in the arena first.

LLM-based trading agents can be expensive and non-deterministic. The harder problem is identifying which strategies remain robust outside a backtest. @roostooAI is building a system to search at scale: > A reinforcement-learning agent factory designed to generate up to 1M agent configurations > Live competitions evaluate them; qualifying agents can then be routed into DEX venues and vaults The team reports 50K users on the pre-launch waitlist for its trading simulator. Catch @Edwardsolah at EASY Residency S4 Demo Day.

3:05
Roostoo avatar
Roostoo@roostooAI·Sep 8

21 bStock assets are now live on the Roostoo simulator. bStocks track selected US equities 1:1 and trade around the clock, giving you more markets to explore and more strategies to test. Live on Android, iOS, and web. Log in to experience it now.

Post attachment
Roostoo avatar
Roostoo@roostooAI·Sep 4

Shorting is now available on the Roostoo Simulator app 🎉 Before, you could only trade when you believed an asset would go up. Now, you can also test what happens when you believe an asset will go down. We’ve kept shorting at 1× to help keep the risk under control. And because you’re using mock capital, you can test different strategies without risking your real money. Log in to the app to experience it now

Post attachment