is a rating system that penalizes based on the OT either being beaten or allowing himself or the rusher to enter three zones around the QB: buffer (1.5yds), danger (1.0yds) or sack (<.7yds). It is based on a scale of 0-3 and comprises the primary metric for overall evaluation akin to PFF ranking of OL. These are strictly speaking position based or coordinate based ratings relative to the QB. Qualitative metrics are noted below I use two scales the QBI and the QBI+Style Metrics for evaluation. QBI data is held in the qbi_metrics table in the db.
Is an evaluation of such things as Momentum, the euclidean distance relative to start and end of play during interface between blocker and defender, the QB-DE rating and euclidean scores. So that 3 metrics are created through the block metrics algo: qb-de rating (0,1), qb-de euclidean (float), momentum (0-100) and efficiency (0-100). QB-DE Euclidean is the measurement of whether the distance has increased from the start of itnerface to the end of the interface b/w the Blocker and Defender, increase in distance of defender from QB. Momentum is a measure of whether the blocker slows the rate of closure on the QB, measured on a frame by frame basis, score reflects total percentage of frames where momentum decreased during interface. Efficiency is a measure of whether a block is using arm block or close or control block style (i.e. body close to other body)
Is an evaluation based on the concept of divergence of the defender from taking a straight line toward the quarterback, the ability of the blocker to deter the rusher from the shortest path toward his target, the QB, it is also based in measuring the euclidean. This euclidean is a measurement from the defenders position compared to his predicted position on a straight line to the QB from his previous position. The higher the euclidean distance the better the result, the higher one is off the straight line to the QB.
Is a measure that looks at the d between the DE and the QB. If the d of the DE increases, between the QB and DE, from the start of the interaction to the end of the interface length a positive score is given and the QB-DE Rating is given as 1, otherwise it is scored as 0 (0,1). If the euclidean average value is above 4yds then a positive score is given. It is important to note that this value usually increases on edge rushes when the arc is affected by the OT to push the rusher away or outward from the QB, arc increases. This can be seen in the plots provided at the link below. A source of noise for this measurement is a scrambling QB that either scrambles toward the DE or away, future versions will need to detect this anomaly.
born and raised near Canton, OH, home of the HOF, I developed an intense interest in all things football, so it is natural to apply Data Science to football as an adult. I use Python for backend development of algorithms to put the ratings together and HTML5 and Javascript for web and mobile presentation layers. Read Research Paper about these Metrics
