Yards per passing attempt is the foundation for quarterback performance. It’s the reason why since the 1970 AFL-NFL merger teams that win YPA by any margin have won about three quarters of games.

But YPA doesn’t just predict wins and losses. It also predicts the quarterback stat that we care most about in fantasy with our signal-callers: passing touchdowns. The TD% of attempts for quarterbacks closely tracks YPA. Last year, 17 of 30 qualifying QBs (224 attempts) had a variance in these two stats, adjusted for league average where 100 is exactly average, of just seven points.

The bottom line: YPA is the dog and TD% is the tail. If a QB has a much stronger TD% (and thus number of TD passes) than his YPA, he hasn’t really earned them and last year’s total has less predictive weight. And if he’s sporting a much stronger YPA than TD%, he was probably just very unlucky in converting touchdowns last year and thus is a good bet to do much better now. Why isn’t the higher TD% the number to bet on instead of the YPA? Because you always bet the bigger sample.

Last year, we applied YPA to quarterbacks and concluded that Cam Newton, Blake Bortles, Ryan Fitzpatrick and Derek Carr would throw far less TDs on a percentage basis. We were right on all but Carr, who was down but still close enough to the prior year to count as a miss. So three of the four predicted busts were correct. (Note we clearly said to ignore the model on Aaron Rodgers and Eli Manning.)

The top buy in last year’s column was Matt Ryan, who we said was THE late-round QB to get last year given he was available often in the last round of 12-team leagues. We also said the two other values were Ben Roethlisberger (TD rate spiked dramatically as predicted, 4.5% to 5.7%) and Jameis Winston (4.1% to 4.8%). While both of them were hits, they almost seem like misses compared with that top-line call to buy Ryan at 2016 ADP. He finished with an NFL best 7.1% TD rate and 9.26 YPA among QBs with the minimum amount of attempts.

No one was really drafting Fitzpatrick with a serious pick so let’s just cross him out. And Roethlisberger was not an ADP bargain either. So relative to ADP, four hits and one miss. That’s pretty good. Let’s see if we can repeat it with this year’s outliers.

The major buy on the board is Washington QB Kirk Cousins. He earned about a 5.8% TD% last year, meaning he should have tossed 35 touchdowns. Few are treating him as a QB of that caliber this draft season (current ADP of 100). So, always gamble when it’s free. Cousins is a top-tier QB at a second-tier price even with the loss of DeSean Jackson. Note that red-zone efficiency has no correlation that I can find year to year and Cousins, who was among the worst last year, was among the best in 2015.

Similarly, Russell Wilson should be expected to have a much better TD rate but that’s being priced into his ADP of 59.8. Similarly, the market is hip to Andy Dalton achieving much greater TD value this year and that’s justified given his 21-point spread between his good YPA and poor TD%.

Dak Prescott is such a public player that there are no bargains with him but his TD rate also should have been between 5% and 6% so let’s call it 26 expected touchdowns. But note that the Cowboys are probably going to have to throw much more with a tougher schedule this year. This is more of a “Buy Dez Bryant” stat though, given his expected market share of Prescott TDs.

The major takeaway: Make Cousins your QB this year. It’s not going to cost you much to see if this model is right.

The sells are similar to last year. But we ignore the variance (higher TD% than YPA) of Rodgers because his offense is absurdly pass-TD friendly. Note he had nine TDs on first or second down inside the opponent’s 3-yard-line last year, easily leading the league. (Next were Matthew Stafford at 7, Drew Brees 6 and Andrew Luck with 5.) Plus Rodgers has been historically a great YPA QB. So fading him for this reason is silly.

And Carr is not going to burn me again. His YPA is slowly moving in the right direction and should be at least average this year at this rate of improvement; he has beaten the model two years in a row regardless.

The one sell again is Bortles, who I understand somehow scored about as well as Winston last year. But he is just a terrible player. You can’t tolerate a YPA+ of 77 where 100 is average. Even though his TD% was below average it should have been horrific — as it would have been had he not gotten all those garbage time TDs when defenses barely cared. That’s likely coming. Bortles should have thrown 16 TD passes instead of 23 last year in 625 attempts (or about the actual rate of the similarly efficiency deprived Carson Wentz). So avoid Bortles — and more importantly Jaguar receivers; paying ADP for Allen Robinson in light of this data is simply absurd.