PART 2 — DEFENSE (Attribute Weights / Evaluation) (Offense was Part 1. Same method, same idea: Shapley/dominance for fair “unique value” weights.)
---
DEFENSE
Same setup as the offense post: I’m using a Shapley/dominance-style weighting approach so attributes get credit for their unique contribution instead of correlation “double counting” the same player type.
Same disclaimer too: I’m not posting the exact multipliers. I’m sharing the takeaways and what changed vs my original correlation-driven results.
---
DT
Unfortunately it’s nearly impossible to make a clean distinction between DT and NT with the data I have, so this is basically one “interior DL” bucket.
In my original evaluation, AWR was king here too, with TAK looking like the clear #2.
After recalculating, the result changed more than I expected:
• AWR and SPD sit at the top for DTs
• STR is still important (but it’s not the whole story)
• ACC is in that same “athletic disruption” tier
• AGI matters more than I thought
• TAK ends up being the least useful among the core DT attributes in this disruption-style setup
This doesn’t mean “TAK is useless.” It means for the output I’m rewarding (sacks/TFL/disruption), TAK doesn’t separate DTs as much as awareness + athletic traits do.
---
DE (LE and RE are different jobs in Madden 08.)
For years my assumption has been:
• LE = quick feet (ACC/AGI)
• RE = bigger/stronger profile
My original correlation results supported that strongly (and it matches what we see in the sim).
Recalculating gave a result that both confirmed the idea and clarified the tiers.
LE
• AWR is #1
• ACC is basically right there with it (this is the big “quick feet” signal)
• SPD comes next
• AGI is slightly behind but still relevant
• TAK is a smaller tier after that
• WT is clearly less important than TAK
• STR is the least important in this LE setup
Translation: your left end is the fast-trigger edge.
RE
RE is “more interesting,” but also not — it’s still the other side of the same coin:
• AWR stays at the top
• SPD is a strong next tier
• AGI follows
• TAK is the next level
• ACC comes after that
• STR and Height show up, but more as supporting traits than core drivers
So overall: LE trends smaller/speedier, RE trends bigger/stronger.
---
OLB (LOLB + ROLB combined)
My original assumption was:
• AWR by far #1
• SPD and TAK right behind
After the new method, the structure is similar but the tiers are clearer:
• AWR is still the anchor trait
• SPD is the next tier (notably lower than AWR)
• then TAK
• then ACC and AGI
• STR and WT sit at the bottom of the “important” group
So the scouting message stays consistent:
AWR first, then closing speed, then tackling.
---
MLB
The captain of the defense needs AWR. No method will change that.
Before, I had SPD/TAK/ACC as the clear next tier behind AWR.
With the recalculation:
• AWR is still king
• SPD stays one tier below
• STR shows more relevance than before
• then ACC
• WT / AGI / JMP land around the same tier
• Height trails behind
Personally, I’m not sure I’ll care about Height/JMP much for MLB — but it’s interesting that STR gained relevance compared to the old setup.
---
CB
One of the most important positions in the game, and the results reflect that.
Same headline as always: AWR dominates.
My old method had SPD and ACC right under it.
The big surprise in the new method:
• ACC becomes much less valuable than I expected
• SPD stays very valuable (no surprise)
• CTH rises (if INT/ball skills are highly rewarded, this makes perfect sense)
• Height is useful but not top tier
• then AGI / STR / ACC as a lower tier
• JMP tends to be the lowest priority
So practically: AWR + SPD are still the core, but CTH matters more than correlation made it look.
---
FS / SS
I used CB-like logic for safeties too, so I wanted to see how the new method behaves.
FS
• AWR and SPD are top tier together
• then ACC
• then Height
• then STR
• and the lowest tier is TAK and AGI
Very “range + awareness” coded, which fits FS in sim leagues.
SS
SS is slightly different:
• AWR is the top tier
• then SPD
• WT follows, with Height around the same tier
• then STR
• and surprisingly, ACC and TAK end up last in this particular setup
So SS becomes more “presence/role/profile” than “pure burst” in these weights.
---
K
We all know the kicker story in this engine: there’s an AWR threshold where misses suddenly stop feeling random.
I can’t prove a hard cutoff yet, but based on personal pain (Marler going 50% in two games and then turning into a machine once he hit 72 AWR), I’d guess the threshold is around the low 70s.
Once that AWR baseline is reached:
• KAC is the trait to pump up (especially into the 90s)
• KPW matters for distance, but doesn’t feel like the main driver of “does he miss easy kicks?”
---
P
My previous calculations said:
• KPW is king
• AWR is close behind
• KAC matters, but noticeably less
My newest run basically screamed:
“Only KPW matters.”
I don’t fully buy it, mostly because the punter score definition is tricky (Inside 20 is very context/opportunity dependent, and I weighted Punt AVG at 0.75 vs Inside20 at 1.0). So this is the biggest anomaly on the defensive/special teams side — I’m probably sticking closer to the old logic until I refine the punter score and/or add better usage proxies.
---
TL;DR (Defense)
• DT: AWR + SPD are the real drivers; STR/ACC matter; TAK ends up surprisingly low for disruption output.
• DE: LE is “quick feet” (ACC-heavy); RE trends more “bigger profile” (SPD/AGI/TAK show more; STR/Height support but don’t dominate).
• OLB: AWR is still #1 by far, then SPD, then TAK; STR/WT are low priority compared to the top traits.
• MLB: AWR is king; SPD next; STR shows more relevance than before; Height/JMP aren’t worth chasing.
• CB: AWR dominates; SPD still huge; CTH matters more than correlation suggested; ACC is lower than expected.
• FS: AWR + SPD top tier, then ACC, then Height.
• SS: AWR top, then SPD, then WT/Height; ACC/TAK end up low in this setup.
• K: likely an AWR threshold (feels ~low 70s) where misses drop off; after that, KAC is the trait to push.
• P: new run says “KPW only,” but I’m treating it as an anomaly due to scoring definition + context dependence.
|
