Conversion Score calculates how often a player has won the point when attacking.

Steal Score calculates how often a player has won the point when defending.

_{^{INSIGHTS: Novak Djokovic & Stan Wawrinka In Conversion Scores}}

_{^{INSIGHTS: Novak Djokovic & Stan Wawrinka In Steal Scores}}
In the example above, Djokovic had superior Conversion and Steal Scores compared to Wawrinka:
Djokovic won 22/29 points when in attack = 76% Conversion Score
Djokovic won 12/28 points when in defence = 43% Steal Score

Attack and defence definitions:
Attack: a player has an advantage in the point and a higher probability of winning the point.
Defence: a player has a disadvantage in the point and has a lower probability of winning the point.

Artificial intelligence considers multiple data points when identifying whether the player is in attack or defence. These include:
- The quality of the incoming ball. E.g., A player receiving a lower quality shot will increase the chance of that player being in attack.
- The player’s location on the court. E.g., A player contacting the ball further up the court will increase the chance of that player being in attack.
- The player’s contact point. E.g., A player contacting a very low ball, or playing on the stretch, will increase the chance of that player being in defence.
- The opponent’s recovery position. E.g., A player creating space by moving their opponent out of position will increase the chance of that player being in attack.

Calculating Conversion and Steal Scores
Conversion Score=(Points won in attack)/(Total points in attack) | Steal Score=(Points won in defence)/(Total points in defence)

Figure 1 shows the Tour average Conversion and Steal Scores for ATP players.

_{^{Figure 1 Tour average Conversion and Steal Scores for ATP players}}