Original author: sysls
Original translation by AididiaoJP, Foresight News
introduction
I've been thinking a lot lately about executing large portfolios on decentralized exchanges like Hyperliquid.
Theoretically speaking, if:
- You can earn excess returns.
- Your positions and orders are open and transparent, just like on a decentralized exchange like Hyperliquid.
So:
- You should expect that there will be a type of trader who tries to get ahead of you and steal your excess profits.
- They will do this by filling the positions you want before you do.
The end result is that you will incur higher transaction costs (slippage) due to being preempted in the trade.
Imagine you want to buy $1 million worth of Bitcoin for $100,000. Someone happens to have placed a sell order for $1 million worth of Bitcoin at $100,000. A preemptive buyer sees your intention, intervenes before you, buys the sell order, and then sells you the $1 million worth of Bitcoin at $100 per dollar. The extra $100 is the slippage that could have been avoided if your intention had been concealed.
Two extremes of preemptive trading
In theory, if this situation is pushed to its "inevitable end," then almost any form of "serious trading" will be suppressed on decentralized exchanges.
However, we know this is not the case. Many highly skilled traders on Hyperliquid are engaged in professional trading with excess returns. Therefore, it is clear that the conclusion that "players with excess returns should not trade on decentralized exchanges" is not so absolute.
Can we derive an intuitive boundary for the limits of preemptive trading based on fundamental principles and existing evidence?
Clearly, if you have a small trading volume and are trading on a highly opaque exchange like Binance, the chances of you being preempted are virtually zero. A small trading volume means your trading footprint (volume) is negligible relative to the market, making you almost invisible; and even if your behavior is entirely predictable, no one can link your specific trading activities (orders placed and executed) to you.
On the other hand, the most typical example of a large and highly transparent wallet on Hyperliquid is the HLP vault itself—the public market maker vault that provides liquidity to other traders on Hyperliquid. I'm fairly certain that there are strategies specifically designed to preemptively trade HLP, and this ongoing pressure has effectively compressed market-making alpha to near zero.
HLP represents a rather extreme example. First, it possesses both the characteristics of "extremely large volume" and "extremely high transparency." It is described as "extremely large volume" because of its huge trading footprint in illiquid long-tail assets (for example, its trading volume accounts for a large proportion of the average daily trading volume).
Furthermore, it is "extremely transparent" because it is primarily a market maker, attempting to achieve the explicit goal of closing out existing inventory at a premium by providing liquidity. This means that whenever a "large" position appears on HLP, you know it will eventually need to be closed. Worse still, you can see every position and every pending order on HLP. This allows you to adjust your portfolio to sell to HLP more cheaply whenever you see it need to buy to close out a short position, and vice versa.
All these characteristics make HLPs a particularly attractive target for front-running, much like how exchange-traded funds (ETFs) are often front-runned due to their strict adherence to index rebalancing. In the hedge fund world, if you actually use the term "front-running," compliance departments will definitely flag you across various dimensions; the industry jargon is that index rebalancing teams are very adept at providing these ETFs with a service that "predicts liquidity needs and profits from them through premiums."
How does front-running occur?
In classic front-running, one market participant knows in advance what another market participant is going to do, and then takes a series of actions to profit from that information.
Here's an (illegal) example: If I'm an insurance agent and I know that my very wealthy client intends to buy $1 billion worth of a illiquid stock throughout the trading day, then at the opening of the trading session, I would place a market buy order for $1 million, and at the close of the session, I would place a market sell order for the same number of shares.
By knowing my client's intentions and actions, I was able to close the deal before him, allowing his buying to drive up the stock price and then profiting from the difference. This is highly illegal because I:
- Action based on insider information
- I have violated my fiduciary duty.
- To benefit oneself at the expense of one's clients.
However, this is a good example because it clearly shows that I was able to profit simply because I knew the intentions and actions of another market participant and could predict the outcome of those actions, thus putting myself in a favorable position.
Every day, front-running occurs on a smaller scale and with less illegality. Trading algorithms can approximate intentions without being told what is going on; they use publicly available information (orders, trades, positions) that everyone has access to. They then estimate the market behavior resulting from this approximate intention and decide whether to take action based on the expected value of the "front-running."
Therefore, we can infer that the transparency and degree of disclosure of your "intentions" are the primary determining factors in whether you will be easily preempted in a transaction.
Gradient distribution of preemptive trading
Okay, now we know that if you are small and not transparent on the trading platform, you don't need to worry about being preempted because no one can judge your intentions. Similarly, if you are large, transparent on the trading platform, and your intentions are very transparent (like HLP), then you are destined to be preempted without any chance of fighting back.
However, these extreme cases are of little reference value to the vast majority of traders. We are more concerned with the "middle ground." As mentioned above, what ultimately determines your tendency to be preemptively targeted is how transparent your intentions are.
Even if you have a large trading volume and trade on an opaque exchange, it's not easy for others to beat you to it. Your orders will be part of the daily trading volume, appearing as "large order footprints," but attributing all orders to a "single entity" is not simple unless your trading method is extremely transparent—for example, you don't have any randomization operations, you trade with split orders of fixed lot size or fixed notional amount, or you send split orders in a very deterministic pattern (such as once every 30 seconds).
If you can conceal your intentions—for example, by randomly selecting the size of your trades, randomly timing and spacing your split orders, and avoiding placing buy orders that are excessively large relative to your average daily trading volume or the number of orders on your order book—it becomes very difficult for others to attribute your orders to a single individual. The market may perceive a general level of buying interest, but it may be unable to attribute this interest to a specific party with access to alpha rewards, and therefore will not price liquidity accordingly.
Fortunately, we can actually extend this concept to transparent exchanges. Although Hyperliquid and Lighter have large vaults and their operations are relatively transparent, actually getting ahead of these vaults in trading is not easy.
The conclusion is: unless you are a very large entity (such as an institutional vault managing hundreds of millions of dollars in assets), you almost don't need to worry about being preempted.
Limitations of front-running
Attempting to generate alpha returns from front-running without breaking the law is itself a practice of alpha strategy. Since you are modeling your intentions based on publicly available information (orders, trades, positions), you inherently bear the risks associated with that model.
Orders, executions, and positions may be visible, but the intent is not. A limit order placed there could represent alpha returns, inventory management, or hedging. Models that assume every order has alpha returns will be gradually eroded by countless misjudgments.
Furthermore, even assuming you can extract the intent relatively accurately, alpha returns are not "omnipotent." All alpha returns contain a certain amount of statistical noise, and your portfolio is not only exposed to the statistical noise of alpha returns, but also bears the additional model risk of misunderstanding certain behaviors as alpha returns.
You might say that if I blindly copy the target's actions 1:1, I can definitely capture all the alpha gains—but the problem is that this exposes you to the risk of being exploited. If you send the same buy order every time the target makes a move, then when the target wants to sell, it can place a limit buy order, see that you've placed the same order, immediately cancel its own order, and then sell to you instead. So you see, blindly trading ahead of others can itself create vulnerabilities.
It should also be remembered and recognized that alpha returns have a time span. Some alpha returns are fleeting and may not even be usable by the attacker themselves (such as the alpha from high-frequency trading); while other alpha returns last for a very long time, and the attacker may give up because they are unwilling to share such long-term risk with you (such as rebalancing trades that last for days or weeks).
Finally, even if you have an extremely experienced frontrunner watching you, the actual impact is only a few basis points. If you truly have consistent alpha returns, many strategies can easily absorb these few basis points of additional cost.
How to avoid becoming an easy target
Even knowing that things aren't that simple, as a smart market participant capable of generating alpha returns, your task remains to conceal your intentions and make it as difficult as possible for attackers to get ahead of you.
You can do many things, varying in complexity and effectiveness. Your first priority should be to persistently collect telemetry data and logs so you can quantify the specific "degree" (if any) of being preempted. You can do this by analyzing a large sample of orders and the markup prices, slippage, and impact costs of trades.
Then, once you have the data, you can take a series of defensive measures. A common thread in these measures is that you should make it less obvious whether you want to buy or sell, how much you actually want to buy or sell, how urgent you want to buy or sell, and whether you are trading an alpha position or a hedging position.
Some simple ways to obscure your intentions are to post two-sided offers simultaneously, using random sizes, and operating at time intervals that are not always deterministic.
One (high-level, sophisticated) way to effectively obfuscate your positions is to split your portfolio into multiple wallets, each maintaining a roughly neutral long/short position and good margin efficiency. Within each wallet, you hold both alpha-generating and hedging positions. Some wallets might have 80% alpha and 20% hedging; others might have 80% hedging and 20% alpha. Over time, you rotate the "type" of each wallet, randomly introducing new wallets and discarding older ones.
This means that if attackers only track one wallet, they might end up tracking a wallet primarily used for hedging, thus getting trapped in losing positions designed for hedging purposes. If they track all wallets, you can further obscure your true intentions through a series of contradictory actions. What that might look like is left to the reader's imagination!
Finally, there are already some (external) solutions available on the market to address this issue. I haven't personally used them, but in essence, they address privacy concerns through one of two methods:
The process involves aggregating orders, performing internal matching, placing the remaining orders on a decentralized exchange for execution, and finally returning the positions to you—this is not much different from the practice in hedge funds where a central liquidity ledger aggregates orders from various strategy groups and then allocates the positions back.
Your order will be split into multiple wallets along with other users' orders in the scheme, executed on a decentralized exchange, and then the position will be returned to you.
in conclusion
If you're a small-scale retail investor, you probably don't have much to worry about, even trading on a transparent decentralized exchange. Front-running has its limitations, making it difficult for others to truly profit at your expense.
That said, as your trading volume increases and the quality of your alpha returns improves, it will naturally incentivize early adopters to keep an eye on you. At that point, you should invest more resources in obscuring your intentions and making things as difficult for them as possible.
This problem is by no means "solved," and for any institution or trader conducting large-scale transactions in an open, decentralized, and transparent liquidity venue, it will be an ongoing "cat and mouse game."
