Retriever Analytics Lab

23 posts

Retriever Analytics Lab banner
Retriever Analytics Lab

Retriever Analytics Lab

@RetrieverLytics

Bringing a fresh wave of sports data science solutions to the world 🌎 check out our foundational product at https://t.co/gwYnHy6Tjt

参加日 Haziran 2023
38 フォロー中4 フォロワー
固定されたツイート
Retriever Analytics Lab
Retriever Analytics Lab@RetrieverLytics·
We built an NFL draft analytics platform from scratch and it's finally live at ready-set-draft.com! Most draft boards are built around consensus. Where is this guy mocked? What round does the industry think he goes? We don't use any of that. Our model doesn't care where you're projected. It evaluates prospects against what actually correlates with NFL success at each position, then matches them to the teams where they'd fit best. The foundation is a proprietary Draft Value Score model. Position-specific weights across four pillars (measurables, athletic testing, production stats, and PFF grades), tuned through correlation analysis against 6,498 historical NFL careers dating back to 2005. Every position group has its own weight structure because what predicts success for a quarterback is completely different from what predicts success for an offensive lineman. On top of that, every prospect gets an archetype classification based on their trait profile, historical comps through percentile-based weighted Euclidean distance matching, a confidence-adjusted grade (A+ to F), and a risk label. Then there's the team side. We built a 6-layer pipeline that constructs roster profiles for all 32 NFL teams, identifies trait gaps weighted by scheme, and produces fit scores for all 19,488 prospect-team combinations. Scheme classification uses empirical clusters derived from actual playcalling data, not assumptions about what a coach "probably" runs. Some of our grades diverge hard from consensus but we consider that a feature not a bug. When your model is anchored to historical outcomes instead of draft position, the board looks different. Go see for yourself now!
English
1
0
0
117
JP Finlay
JP Finlay@JPFinlayNBCS·
These WR rankings from @ChrisTrapasso are very very different than what the consensus has become, though the separation of each prospect is minimal
JP Finlay tweet media
English
14
4
57
17K
Cleveland Browns
Cleveland Browns@Browns·
breaking down 3 quarterback prospects in the draft
English
23
2
49
20.4K
Retriever Analytics Lab
Retriever Analytics Lab@RetrieverLytics·
@Seanptheghost Are you comfortable with this TE room? People were surprised when the Bears took Loveland even though they had Kmet, that turned out pretty well. Definitely plenty of holes to fill though not denying that
Retriever Analytics Lab tweet media
English
0
0
0
22
Good Luck Chuck
Good Luck Chuck@Seanptheghost·
He's not stupid..you don't dump that type of money and resources into TE then draft another one when you have other pressing needs
Retriever Analytics Lab@RetrieverLytics

@ShawnRicks9 @JPFinlayNBCS Fans would hate it but don’t be surprised if they sneak someone like Sadiq or Stowers for the new Ben Johnson-esque offense. Chig provides some new juice but I’m not convinced he is the answer everyone thinks he is. These two match the athletic freak profile Peters loves too

English
1
0
0
53
Retriever Analytics Lab
Retriever Analytics Lab@RetrieverLytics·
@ShawnRicks9 @JPFinlayNBCS Fans would hate it but don’t be surprised if they sneak someone like Sadiq or Stowers for the new Ben Johnson-esque offense. Chig provides some new juice but I’m not convinced he is the answer everyone thinks he is. These two match the athletic freak profile Peters loves too
Retriever Analytics Lab tweet mediaRetriever Analytics Lab tweet media
English
1
0
0
478
JP Finlay
JP Finlay@JPFinlayNBCS·
If you want to consider the reports about a Commanders/Jets draft trade, the draft value chart numbers add up. Commanders No. 7 = 426 points Jets 16 (305) & 44 (135) = 440 points drafttek.com/NFL-Trade-Valu…
English
67
5
191
58.6K
Retriever Analytics Lab
Retriever Analytics Lab@RetrieverLytics·
@Tiller56 These are the only true blue chip talents at positions of need. Could argue Downs although our model isn’t quite as high on him (still has elite potential)
Retriever Analytics Lab tweet mediaRetriever Analytics Lab tweet media
English
0
0
1
292
Mark Tyler (Hogs Haven)
Washington lacks blue-chip talent. This draft lacks blue-chip talent. Mendoza is not a blue-chip QB. He’ll go #1. None of the OTs are what I’d consider blue-chip talent. One, maybe two, will go in the top 6. If we are sitting at 7 and a blue-chip talent is there, YOU TAKE HIM!
English
29
18
294
19K
Retriever Analytics Lab
Retriever Analytics Lab@RetrieverLytics·
@Chiefs DB is the 4th highest need according to our roster analysis. Will the chiefs look to fill that early or lean towards Edge or WR? Check out ready-set-draft.com to explore all the options yourself!
Retriever Analytics Lab tweet media
English
0
0
0
174
Kansas City Chiefs
Which top five CB prospect would you want to see in Chiefs red? 🤔
English
39
38
379
31.6K
Washington Commanders
Washington Commanders@Commanders·
What you've all been waiting for... Wednesday, 4/15 at 10am 🤩
Washington Commanders tweet media
English
139
363
2.6K
324.5K
Baltimore Ravens
Baltimore Ravens@Ravens·
A preview of your reaction Thursday night 👀
English
163
563
5.4K
565.6K
Philadelphia Eagles
Start your week with the greatest draft pick announcement
English
45
411
4.2K
1.1M
Wake Up Barstool
Wake Up Barstool@wakeupbarstool·
"Did you hear that Masters Jets stat? Rory has won the Masters twice before the Jets have gotten one interception? That's an incredible stat." - @stoolpresidente
English
8
15
423
106.9K
Retriever Analytics Lab
Retriever Analytics Lab@RetrieverLytics·
The GM performance page at ready-set-draft.com/gm-performance answers a simple question: How well does your team's general manager actually draft? We graded every draft pick from 2018 to 2025 using VOTA: Value Over Tier Average. The idea is straightforward. Instead of asking "did this pick become a good player," we ask "did this pick outperform what you'd typically expect from this draft slot at this position?" A solid starter taken in the first round might be meeting expectations. That same player taken in the fifth round is a steal. Context matters, and most draft grade systems ignore it completely. Every pick gets classified into one of seven quality tiers: Grand Slam, Steal, Solid Value, Expected, Underperformer, Bust, or Too Early (for 2025 picks that haven't had enough time yet). Each GM gets a stacked bar showing their full distribution across those tiers, plus a VOTA score (cumulative value over expectation) and a hit rate (percentage of picks that met or exceeded expectations). You can sort GMs by VOTA score or hit rate, and the display adapts. Click into any GM and you get their full draft history with every pick listed, their best picks (the ones that outperformed their slot the most), and their worst misses. We also show the round for every pick correctly, including compensatory picks, which sounds obvious but a lot of tools just divide pick number by 32 and call it a round, which breaks for comp picks. One methodology note: picks from 2024 and earlier that still haven't produced meaningful NFL stats are classified as Busts. They had their window. 2025 picks get the Too Early label because one season (or less) isn't enough data to judge fairly. That gray segment in the bar chart is an honest acknowledgment that we don't know yet, rather than pretending we do. Every GM card is exportable as a PNG, same as prospect and team cards across the site. Whether you want to see if your GM has earned your trust or just want ammunition for your group chat, it's all here!
English
1
0
0
49
Retriever Analytics Lab
Retriever Analytics Lab@RetrieverLytics·
We built an NFL draft analytics platform from scratch and it's finally live at ready-set-draft.com! Most draft boards are built around consensus. Where is this guy mocked? What round does the industry think he goes? We don't use any of that. Our model doesn't care where you're projected. It evaluates prospects against what actually correlates with NFL success at each position, then matches them to the teams where they'd fit best. The foundation is a proprietary Draft Value Score model. Position-specific weights across four pillars (measurables, athletic testing, production stats, and PFF grades), tuned through correlation analysis against 6,498 historical NFL careers dating back to 2005. Every position group has its own weight structure because what predicts success for a quarterback is completely different from what predicts success for an offensive lineman. On top of that, every prospect gets an archetype classification based on their trait profile, historical comps through percentile-based weighted Euclidean distance matching, a confidence-adjusted grade (A+ to F), and a risk label. Then there's the team side. We built a 6-layer pipeline that constructs roster profiles for all 32 NFL teams, identifies trait gaps weighted by scheme, and produces fit scores for all 19,488 prospect-team combinations. Scheme classification uses empirical clusters derived from actual playcalling data, not assumptions about what a coach "probably" runs. Some of our grades diverge hard from consensus but we consider that a feature not a bug. When your model is anchored to historical outcomes instead of draft position, the board looks different. Go see for yourself now!
English
1
0
0
117