OlsenData

 

• Forex Market
• Interval Data
• Interpolated Data
• Tick Data
• Pruned Data

Forex Market

The forex market is a worldwide OTC (over-the-counter) market.
It is NOT exchange-traded. This means that trades are bilateral agreements between parties who are not required to register their transactions with any central agency. For this reason, there can be no exhaustive transaction or volume information about the forex market (and so, we do not offer transaction data).

The forex market is, however, stronlgy characterized by the publication of quotes by market makers who contribute this information to organizations which disseminate it globally. These Bid/Ask quotes represent the nexus of this market's evolution. It is common to regard the middle price between Bid and Ask as a reasonable proxy for the transaction price.

Because quotes are non-binding, the occurrence of frivolous or bad quotes must be dealt with through the technology of filtering. This issue is particularly acute when using high-frequency data in conjunction with threshold-based indicators for making decisions.

Olsen Data is renowned as a supplier of filtered high-frequency data and as a supplier of technology for real-time handling of high-frequency data across all instrument types.

Interval Data

As the name suggests, the time stamps and values associated with ticks of interval data describe an interval of time. (This is conceptually different from interpolated data, where the time stamp and values describe a point in time.) Supported interval sizes are: 1 minute, 2, 3, 4, 5, 6, 10, 12, 15, 20, 30, 60 minutes and daily. The time stamp supplied represents the end of the interval. The fields associated with the time stamp can be any statistical construct associated with the distribution/sequence of qualified values in the interval. Unless specifically requested, intervals containing zero ticks (i.e., null distributions of qualified values) are omitted. Commonly requested fields are:

• Open: The first qualified value of the data sequence in the interval
• High: The maximum qualified value in the interval
• Low: The minimum qualified value in the interval
• Close: The last qualified value in the interval

When restricted to these four fields, interval data is often referred to as OHLC data. In the context of Standard OHLC Transaction data, the qualified values are the Transaction Prices (Tx), and each datum has the form: & ltTime-stamp> & ltOpenTx> & ltHighTx> &ltCloseTx> Within the context of Standard OHLC Quote data there is a slight complexity because we have two sets of qualified values: the Bid Prices and the Ask Prices.  In this case each datum has the following form (where Mid = [Bid + Ask]/2): &ltTime-stamp> &ltOpenMid> &ltHighAsk> &ltCloseMid> Some other fields which can be supported include:

• CloseLI: Using a Linearly Interpolated qualified value at the time stamp of the interval (instead of the last qualified value)
• OpenLI: Using a Linearly Interpolated qualified value at the time stamp of the previous interval (instead of the first qualified value)
• Count: Number of ticks in the interval
• Open Time-stamp: Reporting the time stamp of when the Open value occurred in the interval
• High Time-stamp: Reporting the time stamp of when the High value occurred in the interval
• Low Time-stamp: Reporting the time stamp of when the Low value occurred in the interval
• Close Time-Stamp: Reporting the time stamp of when the Close value occurred in the interval

Open for a given interval is the same as Close for the previous interval.  For this reason OHLC data using this definition of Open and Close contains only three fields and is referred to as Standard HLC data.

Interpolated Data

Interpolated data is regularly spaced tick data. Supported interpolation intervals are: 1 minute, 2, 3, 4, 5, 6, 10, 12, 15, 20, 30, 60 minutes and daily. Unlike pruned data—which preserves the original time stamp—interpolated data transforms tick data to supply a stylized series at convenient time stamps of fixed interval. Two interpolation algorithms are supported:

Linear Interpolation:

• If there is a filtered tick that falls exactly on the regular time stamp, it is supplied. Otherwise the values of the filtered ticks bracketing the regular time stamp are used to generate linearly (time) interpolation of the values.

• If none of the ticks bracketing a regular time stamp are within a time period equal to the interpolation interval, that regular time stamp is regarded as stale.

Previous Tick Extrapolation:

• If there is a filtered tick that falls exactly on the regular time stamp, it is supplied. Otherwise the values of the nearest previous tick to the regular time stamp are supplied.

• If the nearest previous tick to a regular time stamp is not within a time period equal to the interpolation interval, that regular time stamp is regarded as stale.

• Pricing is based on counting the number of non-stale regular ticks. This guarantees that interpolated data can never be more expensive than the tick data series from which it is derived.

Unless specifically requested, stale regular ticks are not supplied. Unless specifically requested, weekends are not actively omitted, but they would be implicitly omitted for exchange-traded instruments by virtue of the stale-tick omission. A request for active omission of weekends will have impact only if:

• an exchange-traded instrument is requested for delivery without omission of stale ticks

• a 24 x 7 market (e.g., OTC Forex) instrument is requested

Linear Interpolation is useful for studying market microstructure and indicator research; however, it should be avoided for optimizing thresholds or evaluating return expectations of trading algorithms because it uses a future data point to generate an artificial price on the regular time stamp.  For the latter, previous tick extrapolation or pruned data would be more appropriate.

Tick Data

Olsen Data collects tick-by-tick financial market data using its proprietary server technology. Our forex database goes back to 1986; other instruments were added gradually in the 1990s, leading to a tick-by-tick database covering thousands of instruments. Different consolidation services - such as Reuters, Knight Ridder, and Telerate - have been used in the past. Currently our major suppliers are GTIS and Tenfore.

• Data is collected with microsecond time stamps but is usually supplied with time stamps of one-second resolution.

Kinds of tick data:

• Synchronous Bid and Ask OTC (over-the-counter) quotes (typical for 24-hour forex market)
• Asynchronous Bid and Ask quotes (typical for exchange-traded instruments
• Transaction Prices with or without Volume (typical for exchange-traded instruments)
• Index Levels

Additional fields:

• Four-letter Institution Code: Available with OTC tick data (such as forex spot) from 1994 onwards. We do not support mappings of the codes to bank names; this field will help you identify distinct quotes but does not necessarily identify the quoter.

Olsen Filter Credibility:

• A number ranging from 0 to 1 that describes the quality of each tick.  This field allows you to adjust the filter stringency applied to the subject data.
• A number of major currency pairs - mostly against USD - until the year 1998 are available in the very popular HFDF93 data format, which gave detailed mapping of quotes to the quoting institution.

Pruned Data

The purpose of pruned data is to reduce the frequency of tick data to fit your budget. Interval Pruning is a popular choice.  This pruning method supplies the first tick in every interval instead of every tick.  1-, 2-, 5- and 10-second pruning is available.  Each tick is supplied with the actual time stamp.
 
We also support Stochastic Pruning.  In this case the sequence of ticks is uniformly pruned so as to obtain a target number of ticks in a month.  Stochastic pruning preserves relative tick density fluctuation through the day and month.