BUF ran the ball, forcing MIA to use its timeouts. With one timeout remaining and 2:37 to play, BUF faced a 3rd and 4 at the MIA 28. BUF ran the ball for a 10-yard gain, earning a fresh set of downs and forcing MIA to use its last timeout.
At this point, MIA would have preferred to allow BUF to score a TD. The odds favored scoring a TD of their own in response. Accordingly, BUF should have preferred to take a knee rather than score then.
Here is the chart, built using this research. The red dot is where MIA found itself. The dotted black line is the Win Probability curve for allowing the TD, and for reference, the teal line is the 20 yd line.
At best, MIA would have an 0.08 WP for trying for force the stop and FG, and about a 0.24 WP for allowing the TD. According to the Time Calculator, MIA would have only about 30 sec to respond to a FG, the bare minimum time required. But having two and a half minutes to score a TD is much more likely, even accounting for the possibility MIA converts for 2-points after the TD to make it a 7-point lead.
MIA essentially cut their chance of winning by two thirds by forcing the FG.
In fact, the best thing for MIA to have done is to conditionally allow the TD on the previous 3rd and 4 play. If possible, they should allow the TD should MIA get past a 4-yard gain. This would have maximized the time remaining to counter the score, would have saved a timeout, and would have most likely caught BUF off guard before they would be wise to the strategy.