Compilation｜Odaily Planet Daily Azuma
Preliminary summary
- Odaily Note: For details of this part of the story, please refer to "HyperEVM’s “No. 1” Arbitrage Team Reveals Strategy: How to Move Out $5 Million in Half a Year》。
Turn the clock back to October 2025.
For eight months before that, we had been running the #1 arbitrage bot on HyperEVM.
But the drudgery is almost over. We have been competing with Wintermute for the past few months, and now that new players have joined the game, the revenue has been significantly compressed.
It doesn't matter. My buddy and I are used to this. We never try to compete head-to-head with the institutions and their legions of workers over the long term. We can't do it, there are only two of us.
Our advantage has always been to deploy a strategy as quickly as possible, and then wipe out the profits as much as possible before the big players enter the market.
They can't deploy a strategy in 48 hours. They have regulatory constraints, internal processes, approval procedures, etc. We don’t have those, we just need to be as fast as possible.
So, it’s time to find your next chore.
We started thinking, where is the next opportunity?
That just happened on October 10th. The crypto market looks completely hopeless, everyone has been screwed over, and there are no exciting new opportunities left to explore.
So, we started looking around for new battlefields. On October 13, HIP-3 went live on Hyperliquid. Three days later, TradeXYZ officially launched its first stock perpetual contract market - XYZ100.
Considering that Hyperliquid still has over 40% of its token supply yet to be distributed to the community, we thought it might be a good idea to generate some trading volume on HIP-3.
In fact, this is exactly the same thinking as when we discovered the HyperEVM arbitrage opportunity eight months ago: at the time, we just wanted to brush some spot trading volume on Hyperliquid.
We didn’t know if this would work, but we decided to give it a try – develop and operate a stock perpetual contract robot that arbitrages between HIP-3 and traditional financial markets.
Step one: Enter the traditional financial world
One thing to note upfront is that before this, we had never traded stocks in our lives.
We don’t even quite know what futures are. Basically, we know nothing about traditional finance.
The only thing we know is,IBKR was a very competitive platform for what we wanted to do, so we decided to look into it first.
In the first few days, I was pretty much just trying to figure out how to use the IBKR platform. I basically took screenshots of everything and sent them to Claude: “What is this?” “What does this mean?” “What do I do here?” “How do we hedge XYZ100?”
Basically, this is how we started learning TradFi. Meanwhile, my buddy started looking into IBKR’s API, trying to figure out what could and couldn’t be done.
Coming from the cryptocurrency world, he was used to connecting to an exchange API and getting the program running very quickly.
But IBKR is a whole other world. Market data subscription, contract specifications, order types, permissions, API restrictions, TWS, IB Gateway...We had a lot to figure out, and we weren't even sure if it could be done at first.
However, after a week of playing around with IBKR, we started to find some solutions.
Build an arbitrage bot
This strategy is actually quite simple.
We treat IBKR's price data as "real prices" and continually check for arbitrage opportunities on HIP-3.
If a certain market on HIP-3 is trading at a discount relative to IBKR, we will open a long position on HIP-3 —— Only when our order on Hyperliquid is filled will the corresponding short position be established on IBKR.
If the trading price on HIP-3 is at a premium relative to IBKR, we will do the opposite - short on HIP-3, and then go long on IBKR after the order is filled.
In theory, it's very simple, but in practice, we need to set a lot of parameters for each HIP-3 market.
Let’s take the NVDA arbitrage strategy on the IBKR side as an example:
["NVDA", 55, 400, { maxDelta: 800, slippage: 0.1 }]
- “55 shares” is our minimum hedge size. Since IBKR's fees are as low as $1, we don't want to execute a lot of micro trades. Therefore, we will let the Delta exposure accumulate until we reach 55 shares of NVDA and then hedge it on IBKR.
- “400 shares” is our maximum hedge size for a single IBKR order to avoid excessive slippage.
- "maxDelta: 800" is our safety valve. If for some reason our trades on IBKR continue to fail, causing the position difference between the two markets to reach 800 shares of NVDA, the robot will simply stop trading on that market.
- “Slippage: 0.1” is the maximum slippage we allow when hedging with IBKR.
Next is the HIP-3 side:
NVDA: pair("NVDA", "xyz:NVDA", {makerSize: 400, makerOffsetBuy: 0.12, makerOffsetSell: 0.12, cancelDelta: 0.02, takerRatioBuy: 0.05, takerRatioSell: 0.1, takerMin: 1, takerMax: 2000, limit: 110000, makerEnabled: true, preMarketOffset: 0.04 })
It looks complicated, but the logic is actually quite simple.
- "makerSize" defines our pending order size, while "makerOffsetBuy / makerOffsetSell" defines how big a spread we want to get relative to a reasonable price.
- "cancelDelta" tells the robot that when the price changes significantly, it needs to cancel the original order and place a new order.
- For taker transactions, "takerRatioBuy / takerRatioSell" defines the price difference we need to take the initiative; "takerMin / takerMax" controls the size of the transaction we are willing to execute.
- "limit" is the maximum total position size we are allowed to hold in this market, and "makerEnabled" just allows us to choose to turn on or off pending orders (maker).
- Finally, “preMarketOffset” will add some extra spreads during the pre-market trading session, since liquidity is much poorer on the TradFi side during this time.
first transaction
At the end of October we were finally ready to give it a try.
The first few days were a bit chaotic. We have been struggling with the IBKR API and sometimes lose connection and my buddy has to figure out a way to keep everything connected and functioning properly at all times.
But we soon discovered that the opportunities were plentiful. Basically, it feels like picking up money.
Throughout November, we executed approximately $850 million in volume on HIP-3, with profits exceeding $500,000. This is pretty cool.
December was a little calmer. We did about $550 million in volume and were still profitable, but we actually started to wonder if we should focus on other things.
It does make money, but it doesn't seem like a gold mine.
However, we decided to continue. As always, as long as there is profit to be made, it is usually difficult to stop.
Precious metal mania
The real outbreak occurred in January of this year.
Gold and silver started to rise like crazy, and the demand on Hyperliquid was completely insane.It’s starting to become almost too easy to make money, and it’s a market like this that we’ve spent the last two months preparing for.
One of the issues we face is liquidity. Basically, everyone wants to go long commodities on Hyperliquid, which means we need to keep putting more money on the IBKR side to hedge.
We continued to add funds to IBKR, but moving such a large amount of funds also caused problems for the bank. EtherFi was the MVP on this one, they allowed us to withdraw large amounts very quickly.
Throughout January, we completed $1.7 billion in trading volume on Hyperliquid and generated over $600,000 in funding fee revenue alone.
But like I said before, there are only two of us. We have no internal processes and we move very quickly. Basically, everything is tested directly in production, and sometimes, that comes at a cost.
On January 27, I just arrived in Dubai and was about to go have coffee with my friends and chat about robots. Suddenly, I received a liquidation alert from IBKR on my phone.
I have absolutely no idea what's going on. It was still very early and there were no particularly big fluctuations in the market.
I log into IBKR. It turns out we were net short $120 million worth of gold futures. And gold is in the midst of a sharp rise.
We shut down the bot immediately.At that moment, my whole body was shaking. I'm really afraid of being raped, because all transactions on IBKR are basically executed by robots, and I am not that familiar with IBKR.
Over the next 15 to 30 minutes, I manually closed $120 million worth of my short gold position.
Later in the afternoon, when the market opened, we were finally able to calculate our losses—a loss of $1.1 million.
This is really uncomfortable. But we don't have time to cry about it, we have to figure out what's going on and fix it as soon as possible.
In the end, it was discovered that the reason was actually ridiculously stupid. There was a problem with the data refresh of the IBKR API.
The bot believed that there was a delta difference between our positions on both sides of Hyperliquid and IBKR and kept shorting gold on IBKR in an attempt to correct a delta exposure that actually didn't exist.
again and again; again and again; again and again...
It wasn’t until it cumulatively shorted $120 million worth of gold that we started getting liquidation alerts.
Clearly we need more controls.
We need to make sure that the data we get from IBKR is indeed up to date. We also need to add additional checks before allowing the robot to continue expanding its position. What’s more, bots were simply not designed for today’s huge trading volumes and such density of opportunities.
We spent the whole day fixing everything. The next day, a new version of the robot came back online, but our confidence had been shaken. Maybe we don't even know what we are doing. Maybe the risk-to-benefit ratio is simply not worth it.
We had just lost $1.1 million over a stupid problem, and now it was starting to look like the robot was going to break again at any time. This is the first time since the robot went online that we have seriously thought about it - should we stop like this?
But you should know us by now...We're just two greedy bastards.After losing seven figures, we weren't going to give up. We will only work harder.
And I think this may indeed be one thing we are better at. We have developed many robots together in the past, and almost every time we have gone through a stage of crazy losses, but for some reason, we have always managed to come back in the end.
We don’t spend hours sitting there crying. We try to figure out what went wrong, fix it, and move on.
Until the loss is earned back, this loss will basically become a taboo topic among us brothers.
The next day, silver had a wild retracement after hitting all-time highs, and at one point, there was even a roughly 3% price differential between Hyperliquid and IBKR.
We made about $600,000 in profit from that.
We fucking are back again!
Liquidity management and speed upgrade
By this stage, we had made quite a bit of money. And, we know how the game works.
If there is so much money to be made here, then there will definitely be more people, as well as their huge legions of workers, coming to compete for this piece of cake. Therefore, we must become stronger as soon as possible.
The first issue is funding.This is completely different from pure cryptocurrency arbitrage - in crypto markets, rebalancing funds between different exchanges may take less than five minutes; but here, we have to move funds in and out of IBKR through a bank.
Therefore, we designed a dynamic system that adjusts the strategy based on the liquidity available on one side of IBKR. When liquidity on IBKR is low, we are willing to pay a cost to close existing trades to free up funds.
At the same time, we will require a larger spread before opening a new position. When liquidity is higher on IBKR, we do the opposite and we are willing to accept tighter spreads to open new positions and deploy capital more aggressively.
The second issue that needs improvement is speed.Until then, we have been using IBKR's price data as our source of true prices. This certainly works, but this data source is relatively slow.
As more and more players enter the game, we know that sooner or later it will become a race to the bottom. Continuing to rely on IBKR’s data is no longer enough. So, we started looking for alternatives and found Databento.
With Databento and a Nasdaq data license, we have access to a much faster source of direct market data.
We applied for access in January and were finally approved at the end of the month.
From precious metals to oil
By February, the metals market was still hot and we did about $1.5 billion in volume. And as if that wasn't enough, Trump's decision to bomb Iran in late February caused wild swings in the markets and pushed oil prices above $100.
At this point, we were making approximately $60,000 to $120,000 per day from arbitrage spreads and funding fees. Of course, except for Saturdays and Sundays, when traditional financial markets are closed and we are bored to death.
If we "only" made $40,000 in the past 24 hours, we might even feel like something must be wrong.
So we check the robot, adjust parameters, and try to figure out what's going on and how we can optimize it further.
Claude was a great help to us in this process. We could feed it all the trading data from Hyperliquid and IBKR and let it analyze: Where are we losing the most money? What went wrong? What else can be improved?
This is actually the first time we’ve used AI to analyze transactions, and it’s made quite a difference. Even when things are going well, we still maintain this almost obsessive state.
Basically, it's the only way we know to stay ahead of the curve.
My buddy and I have been talking about robots all day long. He was pushing code updates almost daily, and I was constantly adjusting parameters based on what was happening in the market.
semiconductor mania
At the end of April, as the conflict in Iran began to cool down, we thought that this crazy profitability was finally coming to an end.
We've been making about $500,000 a week for the past few months, and we just can't think of anything else that could continue to create arbitrage opportunities on this scale. Then suddenly semiconductors and all the "bottleneck trades" started to explode in full force.
Stocks like SNDK and MU started trading like pure meme coins. This is so crazy.
We started building this robot in October when basically nothing was happening in the market. Since then, we have experienced the precious metals market, the oil market, and now, semiconductors are starting to be hyped like the altcoin Dogecoin on BSC.
What is happening in this world?
Obviously, there's a lot of luck involved. We happened to be in the right place at the right time and had a suite of products that were already working and specifically suited to this market environment.
But I think there was some good judgment in the early belief in HIP-3, equity perpetual contracts, and more specifically, TradeXYZ.
In May, June and July, we were still doing $1.5 billion to $2.5 billion in volume per month and making about $400,000 to $500,000 a week steadily.
Conclusion
It’s now early September. Since we started, many institutions have entered the market.
Ethena has also announced plans to enter the equity basis trading space in the coming weeks. For us, this opportunity may be ending.
But it’s definitely been a wild ride. In 10 months, during what essentially felt like a “crypto winter,” we achieved:
- HIP-3 and IBKR sides combined for $32 billion in trading volume;
- 1.5% of total TradeXYZ trading volume;
- $10 million profit.
Of course, the reason why all of this can be achieved is that we have a large amount of liquidity that can be deployed.But looking at actual capital deployed, annualized returns are still around 35% to 45%, depending on the stage.
More importantly, this is a great opportunity to really enter the world of traditional finance for the first time and understand how it really works.
Ten months ago, we had never traded stocks and barely even knew what futures were. Now we have traded $32 billionSize stocks and related contracts.
The only thing left now is to pray that Hyperliquid Season 3 comes soon.
Thank you for reading to the end. We'll be back with the next story.