Trading tokenized stocks on-chain -> implementation with OpenClaw & Grok Bot

[sevensisters]

Hi all,

as many of you know since 2025 there are many stocks now available in tokenized form on various blockchains. These are trading 24/7 and make it an interesting experiment to exploit intraday price movements. I herewith briefly describe my approach and how I plan to generate alpha from it.

Initially the idea came from trading the main L1 crypto tokens BTC, ETH & SOL. To that extent I used the help of OpenClaw connected to Gemini 3.1 pro to code over 200 trading strategies based on 70 text books. Then I let OpenClaw fetch 1 minute candles from Binance and backtest the strategies over a 5 month period. I then used Grok bot to improve the strategies and finally backtested them against an out-of-sample dataset spanning 2 years (almost 1,000,000 1 minute candles per crypto token). Strategies were designed so they don't generate more than 10-20 trades per months even though they look at 1 minute data.

Grok was very helpful in identifying the winning strategies and to lump different strategies together into portfolios of strategies (20 strategies per portfolio); we created 4 different types of portfolios base on the out-of-sample results: the top 20 performers, the 20 with the highest sortino ratio, the 20 with the least correlation to each other and the 20 with the highest number of trades.

Grok Bot was then able to calculate the combined effect of lumping two or more portfolios of strategies together into a book of portfolios, thus minimizing draw-down. It then generated a complete live trading engine that I deployed on a Linux virtual private server to execute trades directly onchain on Solana, allowing to keep the cost for trading and slippage to less than 0.25% for a round-trip buy and sell operation.

While I designed this primarily for crypto, it turns out to work well also for tokenized stocks but limited to those with high volume and high beta. Attached are examples of different portfolio of strategies performances for TSLAx versus a simple buy-and-hold strategy as well as an example of combining two portfolios into a better Sharpe & Sortino book of two portfolios.

I would be interested to know if some of you guys have designed similar systems and what your learnings are.

Best regards,

Florian

[ustonapc234]

Florian,

My wife's crypto tutor Calvin Tsai, a farily well known trader in the crypto world has setup a team and programmed trade crypto similar to what you have done.

He was making over 200%+ annualized return for a few years until exactly one year ago today the 10/10 crash (US 19.1 bln liquidation with over 100,000 accounts liquidated just at Hyperliquid alone) that hit him hard and he lost big. He was trading at at Binance and OKX and there was almost no liquidity for even BTC and ETH for almost 10 mins during the crash.

Calvin told us recently that he has resume his program trading. Tokenized stock 24/7 for perpetual futures are already happening on Hyperliquid and Binance with high leverage in Asia and Europe.

I agree that this is something that provides high return (with risk) for those who are good at this.

Regards

James

[sevensisters]

Some guys take on too much risk and then at some point is goes against you. My strategies only sell at a profit so total loss of capital is not possible, only slow accumulation. Below I include the backtest over thie 10/10 crash last year for BTC, and you can see how individual BTC portfolios performed over that period. The yellow shaded area is the only in-sample interval which post-dated this event.

And 200% profit per year also raises red flags: probably lots of trading and lots of capital eaten up in fees.