Abstract

Market data analysis is the starting point in a good understanding of market behaviour. Knowing market data structure and content therefore becomes essential in elaborating concrete market analysis. Throughout this article, we will present how financial tick trade data is structured and how various trade types could impact market interpretation.

Tick data represents the standard upon which the trade or quote price of a financial instrument may fluctuate. The tick provides a specific price increment, reflected in the local currency associated with the market of the financial instrument (stock, future, option, commodity, ETF, forex, fixed income, crypto, ... ) by which the overall price of the asset can change. It is the base of the Intraday and Ticks market analysis used today to understand financial markets.

Because of its high volume and complexity, market tick data can be seen in fields such as Backtesting, Algorithm Trading, Execution Algos, Risk Management, Trade Cost Analysis (TCA) and Market Abuse programs. These interpretations will mostly focus on the size or the price of trades on the timeline studied. But what about the conditions of these trades? Is there a difference between studying one or all trading conditions?

This article will demonstrate the impact of these market trading conditions on the processing of market tick data.

First, it will outline the definitions of trading conditions: What are they, how market exchanges use them, and what the main trading conditions are for an indice such as Eurostoxx 600, Nasdaq 100 or S&P 500 components?

Secondly, we will expand our focus and study three different stocks of the Nasdaq 100 in order to analyse the repartitions of the trading conditions through daily tick data on the same time period, as well as intraday data on the same market day.

Finally, we will endeavour to observe the impact of the trading conditions in the understanding of financial indicators and trade analytics like the vwap or the candlestick indicators.

This article is based on our research in our Ganymede Cloud solution using raw data from “ICE DataVault”. We processed the data through a python Ganymede notebook in order to normalise and visualize it.

Python Notebook on GitHub: Impact of Trading Conditions : Impact of Trading Conditions
Keywords : trading condition, trade, tick, Nasdaq 100, vwap, candlestick, daily, intraday, exchange, index

Table of content

• Abstract
• Trading condition
• Definition
• Un-normalized trading conditions between the markets
• Nasdaq 100 index Study
• Inside the market
• Daily Study
• Intraday Study
• Market analysis
• VWAP
• Candlestick/bars graphs (10 min sampling)
• Key takeaways

Trading condition

Definition

A trade occurs when a buyer (bid side) and a seller (ask side) match their interest on the market. The unique identity of the trade is given by at least six properties:

• The counterparts: buyer and seller
• The unique identifier of the execution
• The timestamp of trade
• The execution price
• The size: total executed shares
• The trading condition

More importantly, the trading condition is determined by the market and represents the type of trade as well as how it happened. The most common trading conditions are as follows: “Regular Trade”, “Auction Trade”, and “Over The Counter (OTC)”.

Un-normalized trading conditions between the markets

The trading conditions, or how the trades are defined, depend on where trades take place, and therefore the market or exchange. With around 1800 exchanges, each exchange has its own trading rules, and thus its own trading conditions.

In the table below, we compare the trading conditions of two stock exchanges:

• Nasdaq Global Select Market: XNGS
• Euronext Paris: XPAR

Table 1: XNGS and XPAR Trading conditions

In the trading condition example above, the table reveals many differences between XNGS and XPAR exchanges:

• The two exchanges do not have the same number of trading conditions: XNGS has 34, however XPAR has 20.
• The range of trading conditions is not the same and does not seem to follow any rule: For instance, the 34 values of XNGS range from 0 to 57, whereas the XPAR values range from 0 to 23.
• The same condition value does not have the same meaning for the two different exchanges, for example the value “8” is “Price Variation Trade” for XNGS and “Off-Market trade” for XPAR.

All these differences emphasize the importance of retrieving the very exact trading condition data for the market we want to study. A misinterpreted trading condition could have considerable consequences depending on the use case.

Nasdaq 100 index study

In the previous section, we observed the differences of trading conditions between exchanges.
In this section, we will focus on the Nasdaq 100 index and observe the proportion of each trading condition in the exchange.

Table 2 : Nasdaq 100 Index Components

We chose to study the full Nasdaq 100 index where we can find the biggest and most liquid stocks. This has enabled us to form a general idea of which trading conditions are most used, if there are any exceptional conditions, or if the proportion is completely irregular. The studies will centre around the number of trades that occurred - but also on the volume traded - for each condition. In both cases, we chose to study the percentage, rather than the raw count of entries.

The two following pie charts show the count and volume proportions of trading conditions for the last 100 days in the Nasdaq 100 components.

Table 3: Top 3 Nasdaq 100 Trading Conditions

Based on the trading conditions count, we can see there are three conditions that are used for more than 99.5% of all the trades: Odd Lot Trade (73.3%) Regular Sale (18.3%) and Intermarket Sweep (7.8%).

With the study of Volume Proportion, can we observe any noticeable differences?

We can see that the most recurrent trading condition (73.3%) now only represents 18% of the volume traded. This is mostly due to the maximum size of the trades since Odd Lot Trades cover the regular trades with a volume lower than one hundred. On the contrary, the Regular Sale Condition and Intermarket Sweep include two-thirds of the volumes traded. Regular Sale has the highest volume proportion with 51.6%. Once again, this can easily be explained if we consider that this condition refers to trades with a volume higher than one hundred. Finally, we can also see that some trading conditions like Closing Prints or Form T, despite a close to null number of trades, count for more than 10% of the volume traded. This can be explained by the OTC trades and Auction trades, which are uncommon but with a huge volume traded at one time.

Inside the market

In the previous section, we observed the proportion of trading conditions for the whole Nasdaq 100 index.

But are these observations the same if we study one specific stock?

To answer this, we are going to focus on three different stock of the Nasdaq 100 index. These stocks will serve as an example for the index as a whole.
We are going to focus on the largest stock, Apple (AAPL #rank 1) a medium one, Air BNB (ABNB #rank 53) and the smallest one, DocuSign (DOCU #rank 102). In a first part we will focus our analysis on daily data for the last 100 days. In a second part, we will run the same analysis on intraday data samples of a given market day.

Daily Study

Table 4: Daily Count Proportion Comparison (2022-02-08 to 2022-04-19)

We can observe that the trading conditions proportions of Air BNB and DocuSign follow the same pattern. The main three trading conditions are the same and cover more than 99% of the trades. There is a slight difference for Apple, where the main trading conditions are the same as in the Nasdaq 100 study, however their proportions are different. There is less Odd Lot Trade (only 55%) and a lot more Intermarket sweep (18%). Why are the medium and small stocks following the Nasdaq 100 trend while the largest one diverges significantly? Before drawing any conclusions, let’s observe the volume distribution in the next section.

Table 5: Daily Volume Proportion Comparison (2022-02-08 to 2022-04-19)

If we look at the volume distribution for these 3 stocks compared to the Nasdaq 100 index, we can see that there are generally significantly fewer differences than above, when we compared trade count proportions. The difference we noticed for Apple no longer appears. One of the explanations could be the Index Sales, where one buys the Nasdaq 100 components based on their weights: Apple being the largest company would be involved in trades with higher volume, leading those trades to be tagged as Regular Trades rather than Odd Lot Trades. For these reasons there will be more Regular Trades than usual, but the volume proportions wouldn't be affected as strongly.

Let’s now compare the daily graphs of our stocks:

Daily Count Proportion for Apple - Air BNB - DocuSign (2022-02-08 to 2022-04-19)

Based on the count proportion, we can deduce that it remains constant for the time of the study. The main three trading conditions are the three visible lines. The rest of the trading conditions are overlaid with the x-axis, due to their small share in the count distribution.

Daily Volume Proportion for Apple - Air BNB - DocuSign (2022-02-08 to 2022-04-19)

Based on the graphs, the proportions are not stable. Over the time interval studied, Regular Sale and Odd Lot Trade are always the biggest volumes, but their proportion can change between the days (from 30% to 70% in one day for Regular Sale). The uncommon but high-volume trading conditions can easily be seen at the bottom of the graphs. We noticed that they have an irregular trend, but easily impact the overall proportion with a peak at 20% or 30% for the highest. The irregularity of the 3 main trading condition lines is linked to the rare peaks of uncommon trading conditions, especially for AirBNB and more noticeably for DocuSign.

Intraday study (5 min sampling)

The previous sections gave us an idea on how the market deals with trading conditions, and their proportions in trade counts or volume. While those studies were carried on daily data samples, we should also question the subject of intraday data behaviours under trading conditions. What happened during a trading day, regarding the trading conditions?

Let’s visualize our studied day using 5 min bars, focusing on the number of trades for each trading condition:

Intraday Count for Apple - Air BNB - DocuSign (2022-04-18)

Our first observation is the “smile”: for every trading condition we can see a decreasing number of trades during the first half part of the day and an increasing number of trades during the second one, thus forming a “smile” shape. The end of the “smile” is higher compared to the whole day, meaning the activity on the market is at its peak. Only three trading conditions easily appeared on the three graphs, in correlation with our past observation that they cover more than 99% of the total trades.

Can we notice any similarities based on the shape of the graphs when focusing on the subject of the volume traded?
Let’s see the three graphs focusing on the volume traded:

Intraday Volume for Apple - Air BNB - DocuSign (2022-04-18)

We can hardly distinguish the “smile” shape as it was the case before, especially for the DocuSign graph. We can still notice the peak of activity at the end of the market, with high activity and therefore high volume. What is interesting to observe is that just after the end of the market day, there are Post Market Trades or End of the Day type Trades that are taking place. They did not appear in the Count graphs due to their small number, but we can see them in Volume graphs. This indicates the importance of monitoring the post-market period in intraday-based studies. Another important point is the irregularity of the volume traded for DocuSign.

After daily and intraday studies, we were finally able to accurately picture what the trading conditions in the market are, how they are used and their impact on the trade analysis. To conclude our focus on trading conditions, we will investigate to see if market analysis like vwap or candlestick indicators can be influenced as well. For example, if we only keep one single trading condition, how does this impact our market analysis?

Market analysis

The previous section outlined the distribution of trading conditions for some index components and how those distributions behave in a daily and intraday time frame.

The following section studies how trading conditions filtering impacts two widely used indicators in market data analysis:

• Volume-weighted average price, or VWAP, is an analysis indicator used on intraday charts that resets at the beginning of every trading session. It's a trading indicator that represents the average price a financial instrument has traded at throughout the day, based on the volume and the price.
• Candlestick (also known as a bar) is a type of indicator used in technical analysis that displays high, low, open, and closing prices of a financial instrument for a given time period (ex: 5 min). The wide part of the candlestick is called the "real body" and tells us whether the closing price was higher or lower than the opening price.

Taking into account our previous observations, we are going to focus on six different filters in and arbitrary trading session:

• All the trading conditions: VWAP (orange line). It's the reference
• The three main trading conditions (Regular Sale, Intermarket Sweep, Odd Lot Trade): VWAP_069 (red line).
• Everything except the main three ones: VWAP_no069 (green line)
• Individual Analysis:
• Regular Sale condition: VWAP_0 (black line)
• Intermarket Sweep condition: VWAP_6 (purple line)
• Odd Lot Trading condition: VWAP_9 (yellow line)

VWAP

Let’s study the VWAP indicator of our six combinations for a given normal trading day.

First, we can see that five combinations generally follow the same trend, and the 6th deviates noticeably (green lines). The significant difference of the VWAP_no069 is due to the lack of trades compared to the others (less than 0.5% of the trades seen in table 3). We can however clearly see the break in the line for AAPL caused by one trade with high volume (0.5% of the trades but 10% of the volume).

Even if the 5 others follow the same trend, there still are small differences between them. For all graphs the VWAP_6 line for the Intermarket Sweep condition (purple line) has a noticeable gap from the others. In the AAPL graph, it can be seen at the end with a 0.2-unit difference. But in the ABNB graph, the gap is created at the beginning of the day, with a 0.5-unit gap at its peak.

The three stocks seem to have a similar trend. We are going to study in detail the end of the day of AAPL. Note that AAPL tick size is $0.01.

Table 6: End of the Day AAPL VWAP Study

The difference between the reference VWAP and "Three main trading conditions" is greater than 2.57 ticks. It is significative for some scenarios. Ganymede is capable of filtering any combination of trade conditions.

Candlestick/bars graphs (10 min sampling)

Our second market analysis is based on the candlestick indicator. It will be carried out on the same day as the VWAP study. Using this python notebook plugged to Ganymede API, we built 6 candlestick graphs using the same trading condition combinations for the same three stocks. In the next table, the results for the AAPL stocks are shown.

Table 7: AAPL Bars graphs 6 Combinations Comparison

In these 6 graphs, we can see the similar trend of AAPL prices: Low volatility at the beginning of the day, and finishing with a decreasing trend.

However, the aforementioned graphs aren’t strictly similar. There are some construction differences in the bars, such as their colours or their height. These differences can create patterns which can easily be interpreted by analyst or, contrarily, it could break them.

We will now analyse DocuSign 10 minute bar graphs at the end of the studied day, from 20:00 to 21:00 (UTC time). Below is a comparison between the “All trading conditions” combination and the “Main three combination”:

In these graphs, we can see a difference at 20:10. In the first case the candlestick is green, whereas in the second one it is red. This means during the 10 minutes between 20:10 and 20:20, one interpretation would state that the price is decreasing while the other would state that it is increasing.

Let’s study this candlestick in detail.

Table 8: 20:10 Candlesticks Numerical analysis

What is happening during these 10 minutes?

We can see that there are nearly as many trades in both cases (~0.5% gap). This refers to our previous part where we studied the main three trading conditions which cover more than 99,5% of the trades of DocuSign. The “Main three trading conditions" should be as representative of the market as that of “All trading conditions”. The difference in those candlesticks is therefore due to the 0.5% trades that do not appear in the second case. In the table we can observe that both cases share the same high, low and closing price but not the same opening price (0.22$ difference) resulting in the change of color. This change is most likely caused by one trade following a trading condition excluded in the second combination. If this trade, priced at 144.7475, took place before the first Odd Lot, Regular or Intermarket Sweep trades, it could cause the graph construction to change.

Although the 6 combination candlestick studies show similar market trends, the last part demonstrated the importance of carrying out in-depth analyses of the graphs. Because one construction change could cause the creation of a different interpretation pattern, analysts need to be aware of those slight differences between each combination and know their causes.

Key takeaways

Through a few well-known market data analyses on a few ubiquitous Nasdaq 100 stocks, this article intends to sensitize the reader to the importance of trade conditions when it comes to tick data (or tick derived data, like sampled indicators)

Here are a few key takeaways:

• Trade conditions impact many use cases. They have to be part of the equation and considered with care.
• Ignoring trade conditions (or being oblivious to them) is most likely to generate errors, problems or mishaps.
• Using pre-generated sampled historical data (ex: n-minutes bars) is likely hiding the details under the rug, be inquisitive about how they were built.
• In the few Nasdaq 100 examples, three trade conditions accounted for the overwhelming majority of ticks, but as low as 0.5% of the remainder could produce noticeable effects.
• Carefully understanding and considering market phases (pre-market, continuous trading, auction and post-market) is also paramount.

Ganymede gives complete control over the trade conditions filtering (and market phases) at the tick level, so you don't get into trouble. Request a demo here.