Inside Quant Trading: The Market- Maker Algorithm.

We start our series of articles on quant trading and the related concepts. In them we will describe and analyze the issues present in the area of algorithmic trading and possible solutions that we utilize in HyperQuant to minimize the arising risks.

This article is about one of the many market-maker algorithms. You can learn more about market-making in Chapter 3.2 of our Whitepaper. The algorithm we will look into is based on limit orders on both sides of the order-book — both for purchase and sale. The aim of such algorithm is in receiving the profit from spread — the difference between given limit orders. However, the strategy of setting the orders simultaneously for best Bid and best Ask will be lossmaking because of the effects of the following factors:

  1. The possibility of taking order on the side, contrary to the price movement, is in the majority of cases higher that taking it on the side of movement’s direction. Thus, if the asset’s price is going up, the orders put up for sale will be completed more often. Accordingly, the orders for purchase will be carried out less, which results in an un-profitable position. This effect was named “adverse selection”, and it appears as a result of so-called “informed traders” being present on the market.
  2. With the existence of an open position — there is a risk of price movement in the direction, contrary to this position. For example, with the price growth — an order for sale was activated (let’s assume that the effect described in point 1 did not take place), and an open long position for one contract appeared. Afterwards, the growth changed to the fall, and if the order for sale was not carried out in time, the specified open position begins to consume the profit — if it was acquired in growth. If the movement down continues — a loss is created. This is the “inventory risk”.

Consequently, the orders management should work based on some algorithm, that will help to avoid adverse selection risk and inventory risk.

We need to get a forecast of price movement’s direction for a short period of time. Such period will be measured in seconds or even milliseconds. We will be considering the dependence of price on the disbalance of volume in the order-book. The latter equals to the difference between logarithms of the best Bid volume and best Ask volume:

That’s all for now. We will tell you more about other areas of algorithmic trading in our next articles. Make sure to check out our Whitepaper for further details. And stay tuned for our “behind the scenes” analysis!

HyperQuant Social Media