Quantitative trading is divided into two stages. The first stage was before 2008, that is, before Dodd-Frank Act. There are many hedge funds or entities similar to hedge funds in investment banks, such as Morgan Stanley's PDT (process-driven trading) and Goldman Sachs' Global Alpha, and many investment banks' proprietary trading business is also very similar to hedge funds. At this stage, the entities of these hedge funds are no different from the outside hedge funds, and their businesses are similar-betting direction, partial hedging, cross-market arbitrage, and they are also brave in taking risks.

At that time, there were great advantages in quantitative trading of hedge funds in investment banks, because two points-first, banks had very good financing channels, and the financing cost was significantly lower than that of hedge funds at that time. If you have tried to set up a fund, you will know how much the cost of capital affects a hedge fund-Buffett's success for so many years can't be separated from his long-term financial leverage of 1.6 times and his capital channel below the central bank's deposit reserve ratio (details of stock trading need to be summarized and accumulated frequently). After a long time, everything is easy. In order to improve their experience in stock trading, novices can learn stock knowledge and operation skills on Fengyun live broadcast platform. Com, which is helpful for making profits in the stock market in the future.

The second is that banks have a gray information flow-customer transaction records. This trading information, even today, is very useful inside information. When Bill Gross left PIMCO a few weeks ago, all the salespeople in investment banks went crazy, constantly studying PIMCO's previous positions in their own banks, and then analyzing which bonds were most likely to be liquidated first, so as to provide trading advice to other customers. During the revival period, cooperative investment banks were replaced many times because the transaction records of their medal funds were not properly kept confidential, and many cooperative banks followed suit.

These two advantages caused the huge profits of proprietary trading at that time, and the management fully supported the star traders to enlarge their leverage in order to make a big impact-in fact, during the financial crisis, many CEOs were promoted from the trading hall to the management by virtue of the huge profits of proprietary trading-such as pandit, former CEO of Citigroup, and John Mack, the predecessor of Morgan Stanley.

This also explains why many proprietary traders who left Goldman Sachs after the financial crisis came out to open their own hedge funds when banks could not do proprietary trading, but they could not replicate the performance of that year at all-because the strength of the whole institution gained excess returns, and when they lost their capital and information advantages, everything became a cloud.

After 2008 and 2009, everything changed.

First of all, the government clearly stipulates that proprietary trading is not allowed, so all investment banks' funds are placed in the asset management department (for example, Goldman Sachs Global Alpha enters GSAM) to operate independently (for example, PDT is separated from Morgan Stanley) or go bankrupt directly (for example, UBS and Deutsche Bank).

There are also some survivors, generally in the stock trading department, who maintain a very small self-operated scale through accounting means in the name of hedging. Many investment banks have such similar teams. But it won't become a climate, and it won't cause any systemic risks-of course, all kinds of road legends have disappeared.

Is there a quantitative transaction within the bank? In fact, there is-that is, with the progress of computer technology, automated market trading. The concept of market maker has just appeared in China-because of the introduction of option market maker system. But in the United States, this trading system has existed since Wall Street. To put it simply, suppose you run a small shop and buy coke. You have two main transactions-one is to get the goods from the general distributor at a bid, and the other is to distribute them to children playing chess in the street and running in the sunset. This is a question. Bid is your purchase price, Ask is your delivery price, and Bid is generally less than Ask (unless you are engaged in charity). You keep quoting these two prices, and at the same time adjust the quotation or the corresponding quotation quantity according to your inventory-for example, if you have too much inventory and grandpa doesn't come out to play chess, you will lower the quotation, so it is difficult to get the goods, but you keep asking and waiting for someone to consume your inventory.

This process is the basic market-making transaction process. In the financial field, because there is no actual general distributor to supply, your buying and selling price is determined according to your estimate of the fair price of the corresponding assets. Usually, the equilibrium price you calculate adds and subtracts a value to form a portfolio. For a long time, this kind of quotation has been done by people, which is boring and easy to make mistakes-and it is difficult to quote quickly for option products (nonlinear prices). I have worked with options traders for a long time before, and their work is not necessarily an intellectual difficulty, but it is definitely a challenge to people's endurance-because after the market opens, they should pay close attention to quotations, one quotation and two quotations, one quotation and two quotations, just like pawns and demons in this trading hall, quotation, quotation, quotation. ......

Therefore, starting with simple assets, API transactions have been supported at the exchange level. What is a simple asset is vanilla, such as individual stocks, indexes, foreign exchange, national debt and so on. Therefore, because investment banks are market makers of a large number of assets, they began to complete the original process through computers. Later, it was found that the computer was fully qualified for this job, because the computer could calculate the inventory at high speed to adjust the quotation, and it could also quote many complicated types. Therefore, since 2000, individual stocks and indexes have been gradually automated. After 2005, the automatic market-making of individual stock options was very hot, and after 2008, the automatic market-making of foreign exchange was quite mature. Since 20 10, the automatic market-making of national debt has also risen in the United States, which is what I am doing now.

What about hedge funds? Besides the traditional quantitative alpha, can't they also do this business? In fact, many hedge funds have better automated market-making business than investment banks-such as Citadel, such as KCG. But what's the difference? There are two differences. First, many hedge funds are not designated as market makers. The privilege of DMM is that it has its own seats-DMM is also very rare in a highly commercialized country like the United States. The reason is that DMM is responsible, that is, in various large-scale financial crises, when liquidity is extremely poor, DMM still has to keep quoting, one quotation, two quotations, one quotation, like a chess piece, like the pace of the devil ... This is very dangerous when liquidity is very poor, because everyone throws you bad assets, for example, when it plummets, it is selling your bid. What about DMM's privileges? DMM can get a high percentage of rebates, which means that the commission rebate is very high. This is in return for its obligations.

The second is that most of them are hedge funds, not brokers. If you want to buy stocks, you generally don't ask them for quotations. In such markets as foreign exchange and bonds, there are two markets, one is B2C market, that is, retail market, which is basically brokerage-customer, and the other is B2B market, that is, brokerage-brokerage. Generally speaking, the bid-ask spread in B2B market is low. An example of an image is that when I was a child, I went to a wholesale bookstore to buy books. A shop didn't have a problem set, so the boss went to the next door and bought one and sold it to me. Finally, I was sure that the boss would return part of the price to the next door. The price I offered and the price my boss paid next door are the price difference between B2C and B2B markets.

The investment banks here are playing hooligans again. They have the advantage of entering the B2C market, so as long as there are enough customers, they can basically make the automation market profitable-because according to the law of large numbers, the trading volume of buyers and sellers should reach a balance within a certain period of time.

So what do hedge funds rely on-a better strategy. Hedge funds to do high-frequency market making, basically participate in the B2B market. They are not DMM, but they also make their own quotations, and then adjust the quotations by accurately judging the price trend, so as to achieve the most favorable orders for themselves, or hold the orders in line with the predicted direction for a longer period of time to achieve profitability. This behavior of becoming a market maker spontaneously instead of DMM is called open market making.

Citadel is the king of automatic option market makers. At its peak, its annual profit can reach 1 billion (in 2009), and the profit of the whole market in that year was about 7 billion. Therefore, if the strategy goes against the sky and there is no customer flow, you can also make money in the market.

In addition, there are many opportunities for hedge funds besides market-making business. Because many commercial banks are not cost-effective-such as bulk commodities. Considering a financial company, we can't just discuss the trading strategy. On the macro level, you must think about the cost of capital and other issues. This is the way to invest. For commodities, banks have done a lot of bad things before (participated in the copper trading of Goldman Sachs and the wind power trading of JP in detail), and were fined extremely high by regulators. This is the stipulation in Basel III, that is, the punishment for holding shares of 1 yuan is completely different from that for holding regulatory assets of 1 yuan overnight. For the specific algorithm, please refer to the calculation rules of RWA (Risk Weighted Assets) in Basel Accord. This series of regulations has led to a lot of new business for hedge funds-because investment banks have withdrawn. And a large number of bank talents have also flowed to hedge funds.

Now that the door is so clear, the difference between investment banks and hedge funds in quantitative trading is obvious-investment banks mainly focus on high-frequency signals centered on automatic market making, customer traffic analysis and quotation game theory. Hedge funds are mainly based on traditional quantitative alpha and quantitative asset allocation-of course, there is also open market automation.

I hope I can help you, and I wish you a happy investment!