Live Data

Some of our features will be available during a match and are therefore driven by live match data. Live data can be very valuable for live insights and for live load monitoring for istance. However, there are some differences between live data and the post-match data we use for our post-match analyses.

When live match data is available some features will be available with live data. Most of these features will rely on both a tracking and event-data stream to have both the positions on the pitch and the events of the match.

The accuracy of the tracking data is depending on the tracking data provider and the system and techniques they use. Most of the time the tracking will be done by optical tracking (e.g. TRACAB) where the players and ball are tracked by multiple cameras around the pitch. Subsequently, the AI-based image detection and tracking algorithms will generate tracking data for the ball and each player on the pitch.

Optical tracking is getting more advanced but will still need some corrections for moments where players are ‘optically overlapping’, for example during a corner it can be very difficult (for cameras) to distinguish different players. The tracking provider will have to correct these missing frames manually during the post-processing of the tracking data. This will firstly be done by the tracking operator live during the match which will improve the (semi-)live tracking data, which is then delayed 15 seconds. Not all corrections can be done within these 15 seconds, therefore a more extensive correction will be done post-match. After post-processing, the optical tracking data based on the positions and speed are very accurate for performance analyses.

Find our more about the validity of Tracab.

During the live tracking, it will obviously be more challenging to perform any manual corrections on the tracking data. The tracking provider will have to rely on their software to correct for missing identifications and therefore missing frames for one or more players or the ball. These software algorithms are getting more advanced and therefore more robust. Another factor that influences the robustness of the tracking data is the quality and the number of cameras that are used to collect the data. Tracab for example will have more robust data with their Gen5 system compared to the previous Gen4 tracking solution. Less post-processing will be needed.

The live stream of Tracab will have a minimal delay (of 15 seconds) to allow some corrections on the data by the operator. This method will decrease the missing frames but will not have the guaranteed quality of the post-processed tracking data.

The main benefit of the Gen5 towards Gen4 lies in the increased robustness of the tracking when it comes to the optical overlapping of players.

Within your Live load monitoring dashboard, you will find physical variables such as total distance. These variables will differ between dashboards based on post-processes data because of two reasons. The first one being the difference in accuracy between live and post-processed data, which we found ranging somewhere between 2% and 4% (September 2021). The main difference can be explained by missing frames that occur during live tracking and are corrected in the post-match data (there are 0 frames missing in post-match data).

In addition to live tracking data our features also make use of a live event-data stream from an event data provider (e.g. StatsPerform). The event data exists of different feeds to provide the necessary information. The first ‘event’ data will arrive between 1 hour and 5 minutes before the fixture is due to kick-off, containing e.g. venue and lineup information.

During the match, the event data provider will update team and player statistics such as passes, shots, crosses, dribbles and more. This event feed will be updated approximately every 90 seconds.

The event data we receive is in most cases notational data and therefore sensitive for human errors. Thus, the post-match data will be more accurate by the post-processing corrections done by the event data provider. Find out more about the availability of the match event data here:

Find out how SciSports corrects for inaccuracies in the event data with the use of tracking data here.