The 2024 Intern’s Guide to Trading


We recently updated our Intern’s Guide to the Market Structure Galaxy and Intern’s Guide to ETFs. Today, we graduate to discussing how trading works.

In our guide to market structure, we talked about who does trading and how quotes across the 16 different exchanges are aggregated. Let’s pick up from there and think more like a trader – starting with some basics.

What is an order?

Orders are the instructions that a trader sends. It will include the: 

  • Stock (ticker) 
  • Side (buy or sell) 
  • Size (number of shares) and 
  • Price 

What is a quote?

When we all try to buy (or sell) anything, sellers want to get the highest price possible – but buyers are always trying to get a discount or lower price. 

In the markets:

  • A seller offers to sell stock.
  • A buyer bids to buy.

That’s why when you look at a trading screen, offers are always higher prices than bids.

Although some venues have hidden quotes, the quotes you see on a screen all come from exchanges. They’re the only venues that share their bid or offer prices with the public. The benefit of sharing (or “advertising”) prices publicly is that it should also help other investors know who to go to if they want to trade. 

Those public quotes are also published in the SIPs, which computes the NBBO and is widely used to measure trade performance and protect investors from bad trades.

What is the spread?

The difference between the public bid and offer prices is called the spread. Sometimes people will talk about the spread in cents, sometimes they will convert it to “basis points” (bps), which are 1/100th of a percent. For instance, the spread in a stock like AMZN is often 1 cent, or around 1 bp. However, spreads in less liquid stocks are often much larger.

Some market makers try to “capture” spread by quoting both a bid and an offer at the same time. Buying at the (lower) bid and selling at the (higher) offer.

But don’t confuse spreads and ticks. Ticks are the legal increment that stocks can be quoted in. In the U.S., for almost all stocks, all ticks are 1 cent (at least for now), but many stocks quote multiple cents (or ticks) wide.

Chart 1: Spreads tend to be wider for less liquid stocks (and only the blue stocks trade with 1-tick spreads)

What is the difference between a quote and a trade?

Quotes are prices where trades could happen. They persist until the order is cancelled or the order quantity is completely traded.  

In contrast, trades occur when a bid and offer occur at the same price. They are completed in an instant and often mean prices change, so others have missed the opportunity to trade at those prices. 

That’s also why trades and quotes are reported and rewarded separately by the SIP.  

It’s all about supply and demand

At the end of the day, trading is all about supply and demand. 

  • New buying adds to demand for a stock. 
  • New selling adds to supply.

In fact, when you look at the amount of stock available at higher and lower prices in the screen (the “depth of book”), you see an actual supply and demand curve for the stock (Chart 2):  

  • As prices rise, there are more willing sellers, and the quantity for sale increases. 
  • As prices fall, there are more willing buyers, and cumulative bid volume increases.
  • This creates the V-shape in the chart below. 
  • Right in the middle is the “equilibrium” price, where buyers offset sellers and trades occur.

Chart 2: Trading adds to supply or demand for a stock 

Adding supply or demand moves markets

Just like in your economics courses, adding demand moves prices up while adding supply moves prices down. 

As a result, trading costs are a combination of:

  • Spread crossing costs.
  • Shortfall, or market impact, caused by adding demand (or supply) to the market.

We can see that in the reported mutual fund trading costs (Chart 3) stocks with wider spreads also tend to cost more to trade (we will talk more about how mutual funds trade later).  

Chart 3: Trading costs are a combination of spread costs and liquidity or impact costs

Importantly, in a 2020 study, we estimated that mutual fund trading costs added to around $70 billion each year, even though they average just 0.31% per trade. So, minimizing the costs of trading is important! 

Different types of order types

One thing that helps traders minimize costs are different order types. Although, as Table 1 shows, each has their own costs and benefits. 

The most basic are market and limit orders, but there are also hidden orders where quotes are not advertised. Often, they help traders try to capture some spread without showing up as new supply and demand (reducing market impact). 

Table 1: Traders’ choices and costs

More complicated order types let buyers automatically reduce their bid prices (fade the market) as sellers arrive at the market (or vice-versa).

Market makers and hedge funds also sometimes need to use short sell orders, which lets them sell shares of a stock that they don’t own.

Different order types introduce different costs

If we look at the diagram below, we see how the basic order types work. We also see the different trade-offs and costs that can arise from each. For example, for a buyer with a:

  • Market order (green line) trades instantly at the offer. That means they pay the full spread costs. However, if the price ultimately drifts higher (red line), they already begin to profit.
  • Limit order (blue line) instead waits for a seller to cross the spread. This order is “advertised” (lit), so the seller knows the buyer exists, and when a seller crosses the spread, the buyer saves the spread costs.  However, the limit order has other hidden costs.

             – If the price then gets even cheaper (black line), the buyer might wish they could have “faded” the market and bid again at an even cheaper price. That avoids what is called “adverse selection.”

             – Or, if the buyer waits too long, the price may instead rise (red line). That forces the buyer to pay an even higher price for a trade (red line), known as opportunity costs.

  • Hidden midpoint order (yellow line) looks to trade at midpoint between the bid and offer but are not advertised. As a consequence, hidden orders can capture half the spread, without advertising their demand (also called “signaling”). However, because they aren’t advertised, a seller may trade in other markets — and this buyer might miss a trade. Waiting increases the opportunity costs, as the offer price could rise, taking the midpoint price (mathematically) higher, too, leaving the buyer paying a higher price for a trade.

Chart 4: Buyers’ choices and consequences

As we see, trading is often a trade-off between speed and cost to trade.

Some of our own research shows that different order types can be used to trade-off spread capture and trade speed at a very granular level.

Chart 5: Markets price the cost of waiting, using different order types very efficiently

How retail trades

Data suggests that the average retail trade is small – less than $10,000. That fact is important because usually the NBBO size is much larger than the size of a retail investor’s whole trade.

That means retail market orders should be able to trade instantly without any residual market impact. It also means retail trades should rarely cost more than the spread to complete.

Because of that, retail investors usually chose between market and limit orders.

There are also rules in Reg NMS to protect retail from bad trades, such as:

  • NMS Rule 605 keeps track of all the trades executed worse than the NBBO, as well as all the price improvement wholesalers pay.
  • NMS Rule 606 tracks all the payments for order flow (PFOF) paid for retail flow.

Chart 6: Rules to keep track of retail execution quality

It turns out that, using 605 data, you can see that retail usually beat the NBBO spread, which is also called price improvement

The reason this works is that retail buying and selling is usually pretty random, something academics call “less informed.” That makes it easier for market makers to capture spread (or avoid adverse selection) trading with just retail. However, a recent academic study found that retail customers receive consistently different fill prices depending which retail broker they use, showing that if wholesalers know who their customer is, they profit at different spread capture for different customers.

The important takeaway for traders is to appreciate that the market has evolved to service retail traders very differently than everyone else. That also means that most retail orders execute off-exchange where institutional traders don’t get a chance to trade with them – something also called “inaccessible flow.”

Where do stocks trade?

We already talked about fragmentation of the U.S. market – how you can trade most U.S. stocks in many different venues, including:

  • Sixteen different exchanges (and counting), regardless of where a stocks “primary listing” is,
  • Over 30 ATSes (dark pools),
  • As well as bilaterally with a number of wholesalers or proprietary firms (single dealer platforms or SDPs).

However, what we saw with retail trading above is segmentation.

In reality, institutional orders have their own market segmentation. Most brokers offer dark pools, and most dark pools create customer tiers that allow traders to be categorized based on their likely spread capture. Then, brokers “internalize” as much trading as they can between their “less informed” customers.

The result is shown in Chart 7 below, which has circles sized by actual volumes. We see that retail and institutional trade flows are directed to a quite different group of brokers, shown by the green and blue arrows. In total, around 44% of all trading occurs before it even reaches exchanges, with the roughly 30 broker dark pools adding to around a quarter of that flow.

For any interns looking into this data, note that it comes from a variety of sources:

  • Exchanges all send their trades to the SIP, with attribution about which exchange did the trade.
  • All of the other trades, which are considered “off-exchange,” print to the SIP anonymously via one of two Trade Reporting Facility (TRFs).
  • In order to see the breakdown of the dark pool trades, FINRA reports aggregated flows that show trades for each trading venue by ticker, but on a delayed basis.
  • FINRA also publish “non-ATS” trading data. Although that is “mostly retail,” there are other trades reported that way, too.

Chart 7: Where stocks trade

This has important market structure implications, too, as it means that the market makers, who are trying to capture spreads, often only see orders after dark pools and wholesalers have profited from their own spread capture first – known by academics as “cream skimming.”

Academic research suggests that cream skimming likely makes it less profitable to provide NBBO orders on exchanges, making those markets “more toxic,” which that should ultimately make spreads worse.

However, other research shows that exchanges that use rebates are able to offset some of the adverse selection that market makers see, helping to improve spread capture on those exchanges and keep spreads tight. That’s especially important for many of the small companies with wider spreads and less liquidity.

Chart 8: Rebate markets have, by far, the most competitive quotes and offer the most liquidity

How do mutual funds trade?

Mutual funds and pension funds (so-called “institutional” traders) represent professionally managed pools of thousands of investors. That means their portfolios, and their trades, are usually much larger.

For example, Vanguard has one mutual fund with over $1.3 trillion in assets and another with almost $1 trillion invested in just 500 companies (the S&P 500). We have estimated that mutual and pension funds trade around $70 billion each day, which includes a lot of daily cashflows. That adds to around $17 trillion over a year.

Consequently, institutional trades are usually W-A-Y larger than retail orders, which means they can’t use simple market orders or even complete a trade at the NBBO.

Instead, institutional brokers need to use additional techniques to keep trade costs as low as possible, including:

  • Working orders: Brokers will usually “work” orders for mutual funds over a number of minutes or hours. That means they split larger “parent” orders up into smaller (child order) pieces. That way, each child has a smaller impact on supply and demand and, therefore, price.
  • Hiding: Others in the market are always looking for signs that a stock will rally or fall (to save themselves money trading). Posting orders in dark pools or using hidden order types on exchange allows investors to be in the market without advertising they are there.
  • Smartly routing: Different stocks have wider spreads, longer queues and more depth, and some venues have different trading costs, too. An algo and smart router can choose different paths and prices for each child order throughout the day to improve the price and speed of trading, including using dark pools to sometimes trade with less informed flow.

In fact, there is evidence to show that brokers even tune algorithms to account for small differences in exchange fees that allow some orders move up NBBO queues faster than others. That, in turn, changes the trade-off between explicit trading costs and opportunity costs.

Chart 9: Even at the same NBBO price, a faster moving “queue” can change order execution prices

Sometimes, especially for an institutional buyer with “more behind” (more of the order still to trade), even adverse selection can be a good thing, as it increases the trade done, and the next order should be at an even better price.

Some claim that this creates a conflict of interest as brokers pay the exchange fees while investors get the better fills – and vice versa. However, an important study with buy-side trade data found that net realized spreads are statistically identical regardless of whether trades happen on maker/taker or inverted exchanges. In short, as long as provided commissions are bundled to include exchange fees, investors should be indifferent to where brokers route their orders.

Chart 10: Exchange fees are a fraction of most spreads

Importantly, institutional investors have the data to measure and manage this conflict.

Firstly, their execution experts (or smart interns) can ingest FIX tag information on each trade that shows which venues each “child” order traded in. From that, they can determine how their brokers are routing flows, as well as estimate the net fees being paid (or rebates earned) by each broker. They can use that to determine if one broker’s net execution costs (commission plus shortfall) seem out of line with others. 

They can also use institutional 606 reports that the SEC created, showing high level disclosures on this activity too.

How fast should you trade?

In most cases, the data shows that trading is a trade-off between how fast a trader can trade and how much their trade costs.

Which brings us to an important question: How fast should you trade?

In reality, the optimal trading speed depends a lot on what you and other investors know.

There is a mathematical way to optimize this problem, which we discussed in How Fast Should You Trade?  This shows that you need to understand the trading trade-offs:

  • Market impact is created when you add more demand to the market, so prices rise to attract more sellers. The faster you do that, the faster prices rise – adding to trade costs.
  • Alpha in the trade. For a portfolio manager, alpha is good, as it represents the amount a stock outperforms the market. But trading alpha measures how fast the stock goes up when you want to buy it, even if you don’t trade – so it’s an opportunity cost.
  • Trade size reflects how much your order changes the normal supply and demand. Generally, larger trades cost more.
  • Liquidity in the stock determines the minimum time a trade size should take to finish. Smaller-cap stocks typically have less liquidity, which limits how fast you can build a large holding in those stocks.
  • Spread costs add up. Generally, the wider the spread, the more expensive a trade will be (Chart 3). That’s because investors typically need to cross more spreads than they can capture.
  • Risk is a factor too. All other things being equal – why wait if the costs are roughly the same, if for no other reason than loss aversion. Behavioral science shows that individuals feel the pain of losing is around twice as bad as the pleasure of gaining!

Once you know all this, you can (theoretically) estimate how trading costs, opportunity costs and risk change over time. After doing that, you can see what trading speed will minimize all the different trading costs – weighing the alpha (opportunity costs) of trading slower against the market impact (cost) of trading faster. In the diagram below, for example, “X” marks the spot!

Chart 11: Optimal speed to trade-off impact and opportunity cost can be mathematically determined

People trade at different speeds throughout the day

Complicating the problem above is the fact that trading dynamics change throughout the day. For example:

  • Spreads are usually wider in the morning.
  • Volatility is typically higher in the morning, and around events and news.
  • Trading activity is higher in the mornings and afternoons – and slower around lunchtime – forming what’s known as a VWAP curve, or smile.

Chart 12: Trading speeds change over the day

The close is usually the most liquid part of the day. But open and close work differently than trading during the day. Rather than a bid and an offer creating spread costs, the market open and close are auctions. In these auctions, buyers and sellers add orders, and the “clearing” price is found – where supply equals demand – literally a single price where buy shares equal sell shares.

On specific days in the year, like when index funds all need to trade or futures or options expire, closes are even larger.

Don’t stress — computers do most of the work for investors

Although this all sounds complicated, the reality is that computers (trading algorithms and market maker models) do most of the trading these days, and they can be optimized with data and programmed to fix much of the complexity that human traders face. Some likely even incorporate machine learning and artificial intelligence.

It’s also important to remember that most of the market is also interconnected and automated. The SIP and NMS rules require it.

So, the biggest input required from most investors is to decide what stocks they want to buy, tell the algorithm how fast they need to trade, and sit back and watch as fills come in. 



Source link