(source: http://www.cba.ua.edu/~rpascala/impliedvol/BSOPMSForm.php)
All 3 commonly used Option Pricing Models (Black–Scholes, Bjerksund–Stensland & Binomial) that are available in retail–based platforms share a common weakness (not commonly understood by retail traders). Despite this weakness, the models are adequate for retail–based trading purposes. Here's the background to the weakness.
On expiration date itself, price can go up, down, gap up/down or drift. Meaning, unlike Implied Volatility which must converge at zero, price does not need to return to any specific point. Zero even though it has no value, is still a point to which IV must return. However, price can go anywhere. Price does need to "revert to its mean" between the start and the end of any given option's expiration cycle, simply because the data points of price can be expressed as a Measure of Central Tendency, as seen in the Standard Deviation graphs of option pricing models. Market dynamics are unstable, making price burst in a dispersed fashion; or fail to burst and remain in a basing pattern. Because of this "jump dispersion" behaviour, there is consensus amongst mathematicians/physicists/statisticians that price diffuses away from any given "value" at any point in time (until the underlying product becomes zero), with the "value" being distinctly random. There is no coil to unwind or recoil price back to where it jumped from; or, to restrain it from jumping to explain the basing pattern.
The weakness common to the 3 option pricing models is this. There is no repeatable method to measure the probability of a product's price going into/staying/coming out of its drifting pattern, away or towards the mean/median/mode.
There continues to be many modifications to the base-line/original options pricing models with improved multi-dimensional calculations. It's a never–ending endeavor to make pricing models become complete in capturing the dynamics of the market . Yet, for a home–based trader, deep understanding of the mathematical improvements are neither practical; or, necessary to improve how you theoretically price the extrinsic/intrinsic value of debit/credit spreads in your favour.
So, what do you pay attention to as a Product's price goes into a drifting pattern? Is it the Drift Rate and Pull Parameter (common to most IV mean reversion analysis)? NO. You can; but, it's not useful.
Do note that "Mean Reversion" computations require the Call Price and Put Price to complete the calculation. What is more critical is to assess the Theoretical price (not the Product's price) of Calls versus Puts for the same given strike when the Product's price goes into a drifting pattern. Calls and Puts will not be priced at equal value, simply because the product's price is drifting. The Skew (– / +) of the different strikes' IV (focus on OTM – where the bulk of retail option trading activity is) and the level of IV (low/medium/high within its extremes of 0 to 1.00), is what you need to determine in choosing Calls versus Puts; or, Calls+Puts in in constructing non-directional trades.