Created Saturday 15 November 2025
At some point I worked in high frequency finance - a controversial industry at the time (still controversial now, I guess, we just have other things to worry about and so I don't hear about it as much recently) The mental image of the HF Finance is that we magically (through computers and statistics) guess the global interest in buying or selling a financial instrument, and then offer to trade with each side at a minimal spread. The ideal of an HF firm is that it participates in every trade on the market, yet ends up with zero position at the end of the day - collecting half of the bid-ask spread on each trade.
Proponents of the HF trading argue that the industry provides liquidity to the market, and therefore is a net positive to the society. Liquidity is this weird abstract thing, that when there is little of it, then two people who each want to buy and sell respectively, still cannot trade. Let's say Alice has a quantity of stock of a Company to sell. Let's say Bob is willing to buy some stock of Company - but let's say they don't know about each other. If Alice announces publicly that she wants to sell, then Bob will offer a low price to buy the stock, knowing that Alice has a need to sell while keeping his own desire to buy secret. Conversely, if Bob announces publicly his desire to buy some stock, then Alice will offer a high price of the stock, knowing that Bob has a need, while also keeping her own desire to sell secret. As a result, "normal" participants (a.k.a. long-term investors) of a financial market are forced to keep their intents secret. Finding a counterparty to your trade becomes hard. "Liquidity is low".
This is where brokers in general and high-frequency traders in particular come in. They bring their own capital to offer to buy or sell to anybody willing to trade. Alice does not need to look for a buyer for her stocks herself anymore: she just sells it to a broker at a publicly advertised price. Similarly, Bob buys his stock from a broker, not worrying whether the stock will be available for sale. The broker's job is to offer the price such that the supply matches demand, and the broker's capital serves as the insurance that that will happen -- taking that risk away from Alice and Bob.
On the other hand, high-frequency finance is said to take the bread away from financial analysts. Analysts analyze the publicly traded companies and other financial instruments to determine how much they are actually worth -- to the best of human ability. Normally, financial analysts pay for their work by trading the financial instruments that are over- or under-priced (relative the analysts' analysis) They simultaneously drive the price towards a "fair value" - and that is beneficial to other investors. However, with a successful high-frequency finance industry (in our mental model), financial analysts don't get to benefit from the results of their analysis: once they determine that a particular instrument is over- or under-priced, the price jumps back to the "fair value" driven by the high-frequency traders, who "sense" the analysts' intent to buy or sell the instrument. The contradiction here is that the high-frequency traders don't know and don't care what the "fair value" is, yet the people who do the work to determine that "fair value" don't get to benefit from their work. (Is there an analogy with the AI-generated "art" here somewhere?)
Now, I'm not ready to tackle the influence of the high-frequency industry on the financial analyst profession. Instead, I want to challenge that first claim, that the high-frequency industry contributes to the health of the financial system by providing liquidity. My claim is that the high-frequency industry has two regimes: when there is enough external investors trading in the market, then the high-frequency industry does indeed provide liquidity and price stabilization by matching buyers to sellers. However, when there are not enough external investors to "feed" the high-frequency firms, the market enters a "self-excitation regime". Remember, in our mental model for the high-frequency industry, the companies are looking for the indicators of the interest in buying or selling an instrument. When there is little trading by "external" investors, the high-frequency companies themselves become the main contributors to these buy/sell indicators. Now the whole high-frequency industry acts like an amplifier, with the microphone lying right next to the speaker. Hence the "self-excitation" regime of the financial industry.
Now, my claim is just a claim at this point. What I'd like to do is:
- Formalize it into a mathematical model.
- Find literature that discusses this topic. With how much discussion the role of high-frequency finance got at the time, there is no way that economists did not study it formally.
- Formulate a testable hypothesis based on my mathematical model and the available literature.
- Test the hypothesis.