Kitman Labs’s AI models and metrics projecting success for various players in tomorrow’s NFL Draft
Metrics That Matter: Uncovering Drivers of NFL Success and Avoiding Overreach
Since 1936, winning at the NFL Draft has incentivized teams to scout, tryout and interview college football players in an educated guess attempt to provide smart projections about the players and their likelihood of success, the round in which to draft them on their respective draft boards, and much more. During the 2025 NFL Draft in Green Bay, Wisconsin over the next three days, this cultural phenomenon will see the use of science, historical data, player profiling, NFL Combine results, and now, contextualized AL modeling provided by Kitman Labs, to assist franchises gain a competitive edge in order to win the NFL Draft and also to not miss on picks. Kitman Labs provide the Sports Techie community blog with a deep dive of projections on several of the key skill position players to include, quarterback, running back and wide receiver. Last year, Kitman Labs projected five NFL Draft sleepers using their Artificial Intelligence (AI) Platform.
Metrics That Matter: Uncovering Drivers of NFL Success and Avoiding Overreach
Where NFL draft strategy and real-world performance don’t always align — and why that matters.
In an ideal world, the metrics that drive early draft picks would be the same ones that determine long-term success in the NFL. Teams would select players who not only look promising on paper but also deliver sustained value on the field.
But reality isn’t that clean.
Subjective evaluation during recruitment opens the door to bias, which can lead to mismatches between what gets a player drafted and what actually leads to on-field performance. To investigate whether this misalignment exists, our data science team analyzed key position groups — quarterbacks (QBs), running backs (RBs), and wide receivers (WRs) — and the metrics that shape their trajectories.
Methodology: Drafted vs. Successful
We built two machine learning models for each position:
- First Round Model: Predicts the likelihood a player is drafted in the first round.
- NFL Success Model: Predicts whether a player will succeed in the NFL over multiple seasons.
Both models were trained using publicly available college performance data (averaged across a player’s career) and NFL Combine results. We used interpretable machine learning to identify the most influential features for each model. Comparing the top metrics side-by-side reveals where the gaps lie — and how decision-makers might bridge them.
Where NFL draft strategy and real-world performance don’t always align — and why that matters.
In an ideal world, the metrics that drive early draft picks would be the same ones that determine long-term success in the NFL. Teams would select players who not only look promising on paper but also deliver sustained value on the field.
But reality isn’t that clean.
Subjective evaluation during recruitment opens the door to bias, which can lead to mismatches between what gets a player drafted and what actually leads to on-field performance. To investigate whether this misalignment exists, our data science team analyzed key position groups — quarterbacks (QBs), running backs (RBs), and wide receivers (WRs) — and the metrics that shape their trajectories.
Methodology: Drafted vs. Successful
We built two machine learning models for each position:
- First Round Model: Predicts the likelihood a player is drafted in the first round.
- NFL Success Model: Predicts whether a player will succeed in the NFL over multiple seasons.
Both models were trained using publicly available college performance data (averaged across a player’s career) and NFL Combine results. We used interpretable machine learning to identify the most influential features for each model. Comparing the top metrics side-by-side reveals where the gaps lie — and how decision-makers might bridge them.
Quarterbacks: The Dual-Threat Advantage
In the First Round model, elite passing numbers stand out — Passing Touchdowns, Completion Percentage, and Yards per Attempt are all strong predictors of early draft status.
However, the NFL Success model paints a slightly different picture. While passing remains important, rushing ability becomes a stronger driver of long-term success. This trend has become increasingly evident since 2010, a period that coincides with the rise of more mobile quarterbacks entering the league. The implication? Quarterbacks need to either be elite passers or offer a dual-threat profile — strong passing plus effective rushing. Over time, this dual-threat capability has emerged as a more consistent predictor of long-term success. Being a good-but-not-great pocket passer may not be enough.
Consider recent history. In 2018, Lamar Jackson was the fifth quarterback taken in the draft. In 2020, so was Jalen Hurts. Today, both are standout performers with MVP honors and have left a significant mark on the league. Hurts has already made two Super Bowl appearances, including a Super Bowl victory, while Jackson has redefined the quarterback position with his dual-threat capabilities. Their rushing ability clearly contributed to their team impact, and was likely undervalued at draft time.
Running Backs: Rethinking the Role of Receiving
Interestingly, the First Round model for running backs favors receiving metrics over rushing. Receiving Yards per Reception and total Receiving Yards both outrank traditional rushing indicators.
But the NFL Success model flips that script. Rushing Yards per Carry tops the list, followed by total Rushing Yards and Rushing Touchdowns. Receiving remains relevant, but secondary.
This suggests that while receiving skills are viewed as a draft-day differentiator, they may not be the strongest foundation for sustained NFL performance. Teams might be overemphasizing versatility at the expense of rushing dominance.
Another takeaway: Speed isn’t everything. The 40-Yard Dash shows up as a key feature in the draft model but carries less weight in the success model. Instead, the 20-Yard Shuttle — a better proxy for short-area agility — appears to be more predictive of long-term performance.
Wide Receivers: A Rare Alignment
For wide receivers, there’s far more consistency between what gets players drafted and what makes them successful.
Both models rank Receiving Yards as the top indicator, followed by the 40-Yard Dash. Metrics like Receiving Touchdowns and Long Receptions also appear near the top in each model. In this case, front-office strategy appears to be better aligned with real-world results.
Final Takeaway: Closing the Gap Between Draft and Delivery
Our analysis reveals clear gaps between draft criteria and success metrics, particularly for quarterbacks and running backs. These gaps may reflect long-held biases or outdated heuristics that don’t match what actually drives performance in today’s NFL.
By realigning talent evaluation around the metrics that matter most for long-term success, teams can sharpen their draft strategy and make smarter, more sustainable investments in their future.
NFL Draft Meets AI: Predicting Tomorrow’s Stars with Today’s Data
Using AI to Identify the Top Quarterbacks, Running Backs, and Wide Receivers Most Likely to Succeed in the NFL
The NFL Draft is a spectacle of speculation, fueled by projections, hype, and historic bias. First-round picks dominate headlines, while potential stars slip through the cracks. But what if we could cut through the noise? What if we could evaluate prospects using a more objective lens?
At Kitman Labs, we applied AI to do just that—analyzing historical data to uncover the traits most strongly linked to long-term success in the NFL.
Their model leverages college performance metrics, NFL Combine results, and physical characteristics to predict which players have the best chance of thriving at the professional level. This year, we focused on quarterbacks, running backs, and wide receivers.
The Science Behind the Model
Kitman Labs’ AI model doesn’t attempt to predict with certainty who will become a star—that’s still a nearly impossible task. Instead, it identifies which prospects demonstrate traits associated with sustained NFL success. A “successful” player, in our model, is one who contributes meaningfully at the NFL level over multiple seasons.
Notably, even the most promising players score just above a 50% success probability, because success in the NFL is rare. Historically, teams only get it right 19–27% of the time across these positions, with first-rounders reaching success just 56–69% of the time. Their model reflects that reality, offering a grounded perspective rather than inflated expectations.
A striking example? In 2022, the model identified Brock Purdy—the final pick of the draft—as the quarterback with the highest probability of NFL success (51%). While penalized for his smaller frame, Purdy’s exceptional college production, rushing capability, and passing efficiency set him apart. Fast forward to today, and the model looks prescient.
2025 NFL Draft: AI’s Most Compelling Picks
Let’s take a closer look at this year’s draft class—who stands out, who’s undervalued, and which players might just become the next Brock Purdy.
Running Backs
Cam Skattebo, Arizona State
Draft Projection: 3rd Round | AI Success Prediction: 60%
Skattebo leads all running backs in our model with a 60% chance of success. That’s triple the average for his draft range. His dual-threat production—averaging 416 receiving yards per season—and 219-pound frame make him a high-upside value pick.
Ashton Jeanty, Boise State
Draft Projection: 1st Round (6th Overall) | AI Success Prediction: 59%
The top-ranked RB in traditional mock drafts also fares well in our model. Jeanty boasts elite production, including 16 rushing touchdowns per season and a solid 211-pound build. Not a sleeper, but certainly validated.
TreVeyon Henderson, Ohio State
Draft Projection: 2nd Round | AI Success Prediction: 32%
Despite explosive stats—like 6.28 yards per carry—Henderson’s lighter build (202 lbs) lowers his projection. Still, his athleticism and playmaking offer upside beyond what the model alone can capture.
Quarterbacks
Dillon Gabriel, Oregon
Draft Projection: 5th Round | AI Success Prediction: 54%
An eyebrow-raiser: Gabriel earns the highest success prediction of any QB in this class. Historically, 5th-round QBs have fared poorly, but Gabriel’s passing efficiency (65.4% completion, 8.77 YPA) and rushing ability make him a strong outlier.
Jaxson Dart, Ole Miss
Draft Projection: 1st Round | AI Success Prediction: 48%
The youngest QB in the class offers strong dual-threat upside—averaging 385 rushing yards per season—with solid passing stats to match. The model dings him for missing Combine data but still rates him highly.
Seth Henigan, Memphis
Draft Projection: Undrafted | AI Success Prediction: 51%
Henigan may be this year’s hidden gem. A four-year starter with impressive passing and rushing production, he matches Brock Purdy’s 2022 projection. With traits the model values and a strong statistical profile, he’s a high-upside UDFA candidate.
Wide Receivers
Jayden Higgins, Iowa State
Draft Projection: 2nd Round | AI Success Prediction: 56%
Our model’s top-rated wideout, Higgins checks every box—974 average receiving yards, 8.3 touchdowns per season, and elite combine numbers (128” broad jump). His size (6’4”, 214 lbs) and explosiveness are ideal.
Elic Ayomanor, Stanford
Draft Projection: 2nd Round | AI Success Prediction: 50%
Ayomanor’s historic 294-yard game vs. Colorado made headlines, but our model likes his full body of work—long receptions, elite broad jump (127”), and a sturdy frame. He’s a well-rounded prospect with serious potential.
Conclusion: Rethinking the Draft
There are no guarantees in the NFL Draft. But objective analysis rooted in data can help cut through the noise and surface valuable insights—especially when it comes to later-round or overlooked talent.
From Cam Skattebo’s top-tier projection to Dillon Gabriel’s surprising upside and Seth Henigan’s potential “Mr. Irrelevant” story arc, AI gives us a unique lens into which players have the underlying traits of future NFL contributors.
As franchises look for every edge, we believe the smart use of historical data, contextualized AI modeling, and player profiling can be the difference between a draft-day win or a missed opportunity.
Sports Techie, I personally wonder about the player no one wanted to recruit out of high school, quarterback Cam Ward, and whether he is worthy of the number one draft pick by the Tennessee Titans? The other highly ranked QB is Sheduer Sanders, who is projected to be selected anywhere from number three by the NY Giants to the early 20’s by the Pittsburgh Steelers, so why is he such an unknown commodity? The 2024 Heisman Trophy winner Travis Hunter is expected to go as high as number two by the Cleveland Browns or three by the GMen, however, is he best equipped to play defensive back, wide receiver or both?
Day one of the NFL Draft in nearly here, with 32 picks raring to go, all selected on stage by the NFL Commissioner, Roger Goodell. Seven rounds make up the entire draft.
Does anyone remember the virtual 2020 NFL Draft because of COVID?
This year, as always, there will be a ton of educated guessing since the NFL Draft is not an exact science.
One thing for certain, using AI conceptualized modeling provided by Kitman Labs along with historical data and player profiling to project and select players, has to be better than the old methods used in 1936 of simply trusting the old school coaches on whom to draft and when.
Right?
See you later sportstechie in Seattle, Atlanta and around the world!
Sports Techie Social Media NetworksSports Techie Twitter: @SportsTechieNET: http://twitter.com/SportsTechieNETTHE #SportsTechie Twitter: @THESportsTechie: https://twitter.com/THESportsTechieSports Techie Facebook Fan Page: http://www.facebook.com/SportsTechieSports Techie YouTube Channel: http://www.youtube.com/user/sportstechieSports Techie LinkedIn Group: https://www.linkedin.com/groups/2958439/Sports Techie LinkedIn (Robert Roble): https://www.linkedin.com/in/robertrobleSports Techie Instagram: http://instagram.com/sportstechieSports Techie Pinterest: http://pinterest.com/SportsTechie/
Sports Techie Skyword: http://bobroble.skyword.com/Sports Techie Snapchat, Skype: sportstechie
Sports Technology Podcast with Bob Roble.
Sports Techie Quora: https://www.quora.com/Bob-RobleSports Techie Moby Picture: http://www.mobypicture.com/user/sportstechieNETSports Techie Flickr: http://www.flickr.com/photos/86791607@N04/Sports Techie Vimeo: https://vimeo.com/user36773456Sports Techie Guy: https://twitter.com/sportstechieguy
Sports Techie Alignable: https://www.alignable.com/atlanta-ga/sports-techie-llc
THESportsTechie Is A Consultant For Videon: https://www.videonlabs.com/