Tesla recently published on their website the new version (v2) of their driving scores (upgrading from v1.2). You will find “all” the mathematics and explanation here:
https://www.tesla.com/support/safety-score#version-2.0
Compared to V1, it is interesting to observe that 2 of the 3 new factors introduced by Tesla have long been used in our Floow scores (time of day and speed). The other new score is linked to unbuckled driving and is a good example of novel insights available from original vehicle data. Our recent new product, Floow Fusion goes one big step further, combining access to smartphone as well as vehicle data, with user consent diligently managed within the insurance app through a secure guided wizard, ensuring privacy and security.
As Tesla published all the mathematics behind their scores, we were curious to compare our decade-long experience with their scores and here are the results:
Time of Day:
Driving at night is recognized to be riskier than driving during the day. It is furthermore interesting that Tesla is translating this into their scores as most of our competitors currently do not offer this score.
We compared their weights to our current standard weights based on our worldwide experience and were happy to see that our weights are very similar on Friday and Saturday nights, which are the two riskiest nights of the week. In addition, our clients will enjoy a more sophisticated weighting as we have dedicated weights for each hour and each day of the week, Monday to Thursday nights being less risky according to our experience than Friday and Saturday.
Finally, it is worth mentioning that each of our clients can fine tune those weights based on their own experience and local behaviors.
Vehicle Speed:
Tesla has added excessive speeding to their scoring suite. The score is counting the proportion of time spent over 85mph. While this is a good start, our actuaries and data scientists found that utilising a continuous function yields better insights. In addition, our algorithm differentiates between (and penalizes less) constant high speed compared to high speed punctuated by accelerations and decelerations. These calibrations translate into more predictive scores when regressed against outcome data.
Smooth driving:
Tesla score is counting breaking events over 0.3g. The formula used is the proportion of time where the deceleration is greater than 0.3g as a percentage of the proportion of time the deceleration is greater than 0.1g. In comparison Floow algorithm is significantly more granular and counts breaking as well as accelerations. For breaking events, we start counting events at around 0.08g, and we progressively penalize braking proportionally to intensity.
The remaining scores used by Tesla leverage many embedded safety sensors and as such serve Tesla drivers well.
Nevertheless, Tesla V2 score is missing what we find to be the most significant predictor of risk: the mobile distraction score.
Our new product Floow Fusion merges original data from the majority of recent vehicles with data from the driver’s smartphone to provide a “best of both worlds” vista into risk levels. If you would like to learn more about, or have a demo of, FloowFusion, reach out to The Floow team in your region:
James Cook, North America
Matthew Chalk, UK & EMEA
Sylvain Derrien, France and Spain
Elisabetta Pizzini, Italy
Share this article
