You can when it's absurd.
Laughton can actually be a serial killer, kick babies and punt puppies and if he puts up 40 points in the NHL and gets a contract, he'd still make more than 800K.
Even if you discount that to 30 or 35.
The analytics helps when you're trying to determine if a player who has good production, and you need other measures to help determine if that increases or decreases his value by a fixed amount. TK is at present, a near PPG player over his last ~131 games. He's getting X value (whatever you deem is fair market) around 7.5, 8, 8.5, 9 or whatever.
If he's exceptional in other areas, he may get 9.5. If he stinks in other areas, he may get 7.5. But there's a guardrail that he'll never get less than as a near PPG player in market environment. So in that context, this model has it so far off that you've blown past the guardrails and are into the abyss. TK will never, for example, sign for a $4M contract - ever. No matter what the analytics state, even if he's the worst player in the league at everything else hockey. His production implies some minimum. Same way, a player who puts up (consistently) 35-40 points will never have a value of 800K (league minimum) - regardless of any 'context' you provide. This also does not account for off-ice factors, which are clearly valued. In the Sanheim vein, Sanheim is shown to have negative contract value. $6.5M for a top pair d-man, with production and stats to back it up. You can argue whether a good contract or a fair contract, but negative?
The model is broken.