Why price moves like a drunken sailor (AKA market microstructure basics)

Exploring the impact of buyers and sellers on market prices

In this article, we will explore:

• A simple model for how market prices move

• The two dominant perspectives on price movement

• Long and short term price impacts, and the role of investor psychology

Basic principles

If we assume that the market consists solely of rational participants, then goods would only trade at prices that are dictated by the intrinsic value of that good.

But what we see is prices that move around too much to be explained by fluctuations in intrinsic value alone. Furthermore, there is a tendency for investors to trade far too much, and price tends to be positively auto-correlated over longer timeframes. ¹²

Auto-correlation is just a fancy way of saying that the price of something today is related to its price in the past. e.g. if a bottle of rum was $20 yesterday, it’s likely that it will be $20 today and rather unlikely that it will be $2 or $200

To put it differently, the price of something today contains some information about its price tomorrow. The implication being that the intrinsic value of something is in some way impacted by the price it trades at.

Let’s now make a somewhat realistic assumption that the market consists of many different participants with varying degrees of sophistication, and wildly different ways of rationalizing prices. Perhaps they have a special crystal ball, or rigorously backtested strategy, or maybe they have regulatory requirements, or some other structural incentives…they all rationalize the price of the same thing in a different way. But when they trade the same thing in the same market, regardless of their level of sophistication or method of rationalization, their contribution to price fluctuations are the same.

Whether I’m a party host who needs 50 bottles of rum, or a pirate who needs 50 bottles for my the next treasure hunt, or a bartender who needs to get rid of 50 bottles - the price of a bottle of rum will shift similarly, regardless of which one of these three is out on the market

The two perspectives: efficient market driven vs. order flow driven..

This impact on price, regardless of how informed an investor is, was shown by Gabaix and Koijen³. Put simply their model found that:

• Buying $1 of an individual stock increases the market capitalization of that stock by $M in the long run with M ~ 1 even if it is an uninformed trade

• Buying the market as a whole (e.g., VTI, SPY, etc.) has an even larger impact with M ~ 5

This is in contrast to rational models that predict that uninformed trades would not have a substantial impact on price with M ~ 0.01

Impact on price does not scale linearly with order volume

What we observe in most financial markets, is that large orders are usually split up into multiple smaller orders. The smaller orders are executed incrementally using a combination of order types over a span of time (minutes to days).

The Square Root Law defines the average price impact of a large order with volume Q, over a period of time T.

---

This law clearly states that price impact does not scale linearly with the size/volume of an order. The Square Root law also implies a few interesting things:

1. The second half of the large order impacts the price much less than the first half

   1. This means that there is some time period (Tm) within which the influence of past trades is relevant, after which time price is potentially ‘back to normal’

2. The total market cap is seemingly irrelevant to the price impact, rather it is the volume traded over the time period (Vt) that is important

3. A relatively small total order size can have a large price impact

   1. Trading 1% of daily volume moves the price by sqrt(1%) = 10% of its daily volatility

Volatility is an important concept to understand, you can find the basics in a previous post below

A lightweight introduction to volatility

What about the long term impact on price?

The Square Root Law is limited as it only explains the impact on price over a limited period of time from when a large order begins execution. Latent Liquidity Theory (LLT)⁴ purports to explain longer term price impacts and relies on some specific assumptions about market participants, namely:

• There exist long term investors with a closely held ‘reservation price’, sort of like a private opinion and not necessarily posted on the open market for everyone to see

• Reservation prices are updated as a function of time, incoming news, price changes, etc.

• The collection of reservation prices of all market participants represents the available liquidity in the market even though it may be ‘latent’

These reservation prices remain ‘sticky’ during Tm, that is, long term investors are slow to change their reservation price over the duration that a large order has hit the market (and is impacting price). But when reservation prices are revised, they have a tendency to distribute themselves around the new market price.

These assumptions imply that the market price is considered together with other factors an investor considers (e.g., news, fundamentals, etc.) when rationalizing fair value of a bottle of rum

So LLT implies that the long term impact of large orders:

1. Decreases over time down to a value that depends on the reason why the large trade was triggered in the first place

2. Is independent of how long it takes to execute the large order

3. Is inversely proportional to the volume traded

4. Is linear with respect to the total size (volume) of the large order, and

5. Is proportional to the volatility observed over Tm (the time period within which the large order has an impact on price as per the square root law)

To simplify this…The larger the order, and the more volatile the market is, the higher the permanent impact will be on price. This impact could be offset by choosing instruments and/or time periods with higher volumes being traded.

To Conclude:

• If order flow is the dominant cause of price changes then you get paid for anticipating behaviours of others as opposed to correctly forecasting intrinsic/fundamental value

• There is likely more signal to be found (i.e. statistical patterns) in larger cap, more actively traded instruments

• Prices of long term market participants remaining sticky but then redistributing around a new price may be a reflection of anchoring and the law of large numbers

Additional reading, fantastic paper “The Inelastic Market Hypothesis: A Microstructural Interpretation”

---

Thanks for reading…if you found this post useful please forward it on to your friends and colleagues!

1. Summers (1986). Does the Stock Market Rationally Reflect Fundamental Values?

2. Odean (1999). Do Investors Trade Too Much?

3. Gabaix and Koijen (2021). IN SEARCH OF THE ORIGINS OF FINANCIAL FLUCTUATIONS: THE INELASTIC MARKETS HYPOTHESIS

4. (Mastromatteo, Toth, and Bouchaud (2013). Agent-based models for latent liquidity and concave price impact