End-of-Day Tick Statistics for Execution Planning

Summary

The document outlines end-of-day statistics that can be computed for each ticker after a market closes. These include price and size summaries for bids and asks, spreads, executed volume, volume share, VWAP, execution counts, and intervals between trades. It also proposes calculating historical summaries over recent ordinary business days, using uniform intraday buckets whose lengths vary with the ticker’s trading activity.

The example connects these features to a buy-side VWAP strategy seeking to reduce market impact. Historical volume and bid-price behavior within a time bucket can help estimate available liquidity, set displayed order size and price, and decide whether a lagging order should cross the spread or wait for later volume. The document is a proposed feature list and use case rather than a validated forecasting method. It does not specify data-cleaning rules, treatment of auctions or outliers, or evidence that these statistics improve execution outcomes.

Key ideas

• End-of-day tick summaries can include price, quote size, spread, volume, VWAP, and trade timing statistics.
• Intraday statistics should be computed in uniform time buckets suited to each ticker.
• Historical bucket-level volume and quote data can inform expected liquidity during execution.
• A VWAP strategy can use these estimates to adjust order size, price, and urgency.
• The proposed statistics are a starting point and do not establish predictive performance.

Tags

# Simple EOD computations for tick data


# Simple EOD computations for tick data












As part of End-Of-Day calculations once a particular market/exchange has closed for all the tickers traded on that market one may typically compute the following properties:

- OHLC

- Bid/Ask Price (mean, median, stddev)

- Bid/Ask Size (mean, median, stddev)

- Spread (mean, median, stddev)

- Executed volume (total, mean, stddev, min, max)

- Percentage of total daily volume

- VWAP

- Time between executions (mean, stddev)

- Number of executions

- Perform points [2-9] over the last N non-extraordinary business days (where N is 10 and 20) obtaining mean, stddev, min and max

The statistics are summaries computed over uniform buckets of time for each ticker. The buckets can be of size 30sec, 1min, 2min, 5min, 10min - which bucket is used for a particular ticker is dependent on the nature of the ticker.

For example, a 10BD mean of executed volume and bid price (with their stddev) in a particular bucket for a well behaved ticker can help a vwap buyside mean oriented market impact minimizing strategy predict how much volume could be expected in that interval, allowing it to make better decisions about how much volume it should have out, and at what price levels, and whether or not it's falling behind and if so should it cross the spread now or will there be ample volume for it to catch up in the next intervals before its end-time is reached.

My question is: In addition to the above what else does one typically compute? - and if possible why?

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.