MASSIVELY UPDATED
(I've run a lot more tests, including how coach strategies affect the draft when you are controlling all 32 teams. Scroll all the way to the bottom if you've already read the first part.)
I've figured out exactly how the CPU logic works for NCAA drafts.
After simming through an entire season, including player progression, resigning, and free agency, the following teams looked like they might have needs at QB.
Bengals (78)
Dolphins (0) - They had NO QB
49ers (78)
Lions (78)
Seahawks (58)
The rest of the teams had at least one QB with an OVR of 80 or higher.
Here were the first ten teams for the draft:
1) Jaguars
2) Bills
3) Seahawks
4) Broncos
5) Browns
6) Raiders
7) Buccaneers
8) Dolphins
9) Giants
10) Titans
Knowing that the Seahawks get the third overall pick, and their starting QB is a 58 OVR, you'd reasonably conclude that they'll be taking a QB with their first pick, right? Might even guess that Andrew Luck is probably heading there, since the Jaguars have Garrard and Gabbert, and the Bills have Fitzpatrick?
Here's how it went:
1) Jaguars - QB Andrew Luck
2) Bills - QB Brandon Weeded
3) Seahawks - HB Tyler
4) Broncos - LT Nate Potter
5) Browns - ROLB Nigel Bradham
6) Raiders - Landry Jones
7) Buccaneers - RT Addison Lawrence
8) Dolphins - QB Ryan Lindley
9) Giants - P Drew Butler
10) Titans - RE Vince Browne
So the Jaguars and Bills take a QB when they don't need one, and the Seahawks who desperately need one take a HB? When they have two HBs already, who are 83 and 81 OVRs? To say nothing of that punter mess...
But I can explain! Here's how the logic works:
Going into the draft, here's the top 13 players listed in order as they appeared on the screen. I've included their OVRs, which I found after the preseason was over:
QB Andrew Luck (84)
QB Brandon Weeded (84)
HB Tyler (84)
QB Landry Jones (83)
LT Nate Potter (86 as a RT, presumably 83 as a LT)
ROLB Nigel Bradham (83)
HB Polk (83)
P Drew Butler (83)
SS Barron (82)
FS Lester (82)
QB Ryan Lindley (81)
FS Smith (81)
RT Addison Lawrence (81)
The first thing to take away from this is the
players are definitely sorted by OVR. This means
there is NO value in scouting, in my opinion.
Now, let's look at why teams drafted like they did. The Jaguars pick first, so they start at the very top of the list: QB Andrew Luck. 84 OVR. Their starting QB has an 83 OVR. So even though it's only a 1 point difference (and they already have two QB's in the 80's), they take Andrew Luck because his OVR is higher than their current QB.
Now it's the Bills turn. Luck is gone, so the top of the list is Weeded. Weeded's 84 OVR beats Fitzpatrick's 82 OVR, so they take Weeded.
Now here's where things get REALLY interesting. The Seahawks are up. They have a QB of 58. Their HB is 83 (with an 81 backup). Here's the list:
HB Tyler (84)
QB Landry Jones (83)
LT Nate Potter (86 as a RT, presumably 83 as a LT)
ROLB Nigel Bradham (83)
HB Polk (83)
P Drew Butler (83)
SS Barron (82)
FS Lester (82)
QB Ryan Lindley (81)
FS Smith (81)
RT Addison Lawrence (81)
They desperately need a QB, and they can have one by dropping just ONE SPOT on the list. But that's not how it works. They start at the top. And 84 HB beats 83 HB, so they take Tyler over Landry Jones!
Now we are up to the Broncos. Here's their list:
QB Landry Jones (83)
LT Nate Potter (86 as a RT, presumably 83 as a LT)
ROLB Nigel Bradham (83)
HB Polk (83)
P Drew Butler (83)
SS Barron (82)
FS Lester (82)
QB Ryan Lindley (81)
FS Smith (81)
RT Addison Lawrence (81)
The Broncos start at the top of the list. An 83 QB doesn't beat Orton at 84, so they move down. Their tackles are 86 and 83, so they keep looking. They have an 82 OLB, so they take Bradham.
Seeing how it works so far?
Now, let's dive into
the Punter issue. By the time it gets to the Giants, 9th pick, the list looks like this:
HB Polk (83)
P Drew Butler (83)
SS Barron (82)
FS Lester (82)
FS Smith (81)
The Giants HB is an 86...but their punter is only a 66. Drew Butler is chosen.
CONCLUSION: The draft choices are sorted by their overall scores, and the CPU KNOWS EVERY SINGLE OVR. There is no scouting involved. Each team starts at the top of the list and takes the first choice they find that is higher than their current OVR at that position. As the Dolphins showed us,
they will take a 1 point increase in starting HB over a 30 point increase in starting QB, if the HB is on the list higher than the QB is.
As for
why the Punter is ranked so high? Because his overall is 83, while the first draft pick (Andrew Luck) is an 84. Position has nothing to do with it...the ONLY thing considered is overall score.
FIXES:
1) Control all teams during the draft (this creates other issues though, including the fact that the computer will not put User controlled team players on IR)
2) EA introduces logic to CPU where they consider serious needs (getting a QB when yours is 58 OVR) over simply improving OVR score at first come, first available (Getting a HB at 84 when you have HBs at 83 and 81.)
3) EA stops sorting the draft by OVR only. Introduce some randomness to it, allow for diamonds and flops, etc.
4) EA blocks the CPU from automatically knowing every player's OVR.
5) EA changes the program so that Punters and Kickers can not be taken before whatever round seems appropriate.
MAJOR UPDATE
Okay folks, I've done some more research, following your suggestions.
Test 1: The original test, controlling only 1 team. CPU chooses a 1 point increase at HB, rather than a 30 point increase at QB. This is covered in detail in my first post here.
Test 2: Controlling all 32 teams, really dived in. Changed each coach's strategy depending on their depth chart. Made very minute tweaks to have a perfect, hands off draft. Findings: Most of my minute tweaks were irrelevant. Say that I've set a team's QB priority at 70 and their HB priority at 80. When it's their turn, the first name on the list is a QB and the second name is a HB. They'll choose the QB. A 10 point priority difference seemed irrelevant.
The Colts chose P Drew Butler with their first pick (half way through first round). I had moved all the punters to priority 10-30, so I was outraged...until I double checked and saw I had never messed with the Colt's sliders. All their priorities, including Punter, were at 100. User error. Whoops!
Test 3. I set up an easier system for determining priorities. If there was an emergency at a position (ex. Dolphins released ALL of their QBs), that priority was 100. 2nd string needs were priority 70. All other priorities were 50, except FB, K, and P. FB's were 30. Needed Ps and Ks were 30. Unneeded Ps and Ks were 10. This system took me about an hour to prepare for, versus the many hours I put into Test 2's system.
Findings: I'll go into detail on this one.
Here is a list of the draft choices in order. I found their overall's after the preseason. I wrote down the first fifty, but I'll only list ten here.
QB Wheeden (85)
QB Luck (84)
QB Jones (84)
LT Stevenson (84)
ROLB Bradham (84)
RE Browne (84)
SS Barron (83)
LE Massaquoi (83)
FS Proctor (83)
QB Lindley (82)
First up is the Raiders. I've put their settings at QB 100, CB 100, SS 70, and the rest 50, except for FB at 30 and K/P at 10. I'll also moved their QB all the way to scrambler, to reflect their Need for Speed. Going off of the priority settings and list (not knowing anyone's speed though), I'd assume they'll go for Wheeden. They do.
Success!
The Buccaneers are up next. They have no dire needs. Their settings are QB 70, WR 70, G 70, T 70, C 70, MLB 70. The rest are at their normal settings (50/30/10). Going off of this, I'd assume they'd go with the first of the 70's available, which happens to be the first choice QB Luck. They do.
Success!
The Broncos are next. HB 70. T 100. OLB 100. MLB 70. CB 100. This is the real test. With QB at 50 and T at 100, they should skip Jones and pick up Stevenson. They do.
Success!
Now it's the Panthers. T 70. G 100. C 70. CB 70. FS 100. SS 70. P 30. Will they jump over QB Jones, RE Browne, SS Barron, and LE Massaquoi to pick up FS Proctor? Yes.
Success!
Let's fast forward a little. It's now the 7th choice of the 1st round. Cowboys up to bat. QB 70. DT 70. MLB 70.
The list:
ROLB Bradham (84)
SS Barron (83)
LE Massaquoi (83)
QB Lindley (82)
The Cowboys should drop to QB Lindley. But they don't.
FAIL. They choose ROLB Bradham. OLB is 50, QB is 70. Is the problem that they won't drop four slots for a 20 difference? Or will they not drop 2 OVR for a 20 difference?
A few picks later, it's the Cardinals turn. WR 100. TE 100. T 70. G 70. C 70. DT 70. CB 70. P 30.
The list:
SS Barron 83
LE Massaquoi 83
HB Tyler 82
FS Smith 82
RT Lawrence 82
FS Lester 81
P Butler 81
FS Tate 81
SS Taylor 81
LE Curry 81
CB Fleming 81
WR Jeffery 80
Will they drop 12 slots and 3 OVR for a 100 need? Or settle for 5 slots and 2 OVR for a 70 need?
Neither. They settle for best available, SS Baron. 50 need.
Immediately after that, the Titans are up. TE 70, T 70, G 70, C 70, OLB 70, CB 70.
LE Massaquoi 83
HB Tyler 82
FS Smith 82
RT Lawrence 82
FS Lester 81
P Butler 81
FS Tate 81
SS Taylor 81
LE Curry 81
CB Fleming 81
WR Jeffery 80
Rather than drop 4 slots and 1 OVR for a 70 need, they take LE Massaquoi (50 need)
Likewise, the Bengals will not drop three slots and 1 OVR for a 70 need, taking a 50 need instead.
But the strangest one thus far is the Bears. HB 70. TE 70. DT 70. OLB 100. MLB 70.
FS Smith 82
RT Lawrence 82
FS Lester 81
P Butler 81
FS Tate 81
SS Taylor 81
LE Curry 81
CB Fleming 81
WR Jeffery 80
DT Crick 81
They probably won't drop 9 slots and 1 OVR for a 70 need, so they'll take best available at FS Smith, right?
No. But they also don't go all the way to Crick. They choose SS Taylor. They bypass 4 50 needs...including two that are 1 OVR higher...for no apparent reason. WHat the?
Meanwhile,
P Butler is not chosen in the first round. He falls all the way to the
TOP OF THE SECOND ROUND. Where he is chosen by the Chargers. QB 70. FB 70. WR 70. TE 70. T 70. C 70. DT 70. OLB 70. MLB 70. I don't have the available list in front of me, but while I do know the P was at the top of it, you're telling me there were NO OPTIONS with that many 70 needs? The P need, by the way, was 10.
After the P was taken, I simmed the rest of the draft.
Mildly related side note: I checked the draft grades. ALL TEAMS HAD A D+, with the exception of the Steelers, who had an A+. The Steelers had QB 70, DT 70, MLB 70, CB 70. Here were their draft picks, from 1st round to 7th.
1) FS Lester (81 A). Notably not on their list.
2) HB Ballard (78B). Not on their list.
3) DT Howard (73C). On the list.
4) QB Foles (71C). On the list. Listed as their "Best Value" pick
5) LE Taylor (70?). Don't know potential - he was cut in the preseason.
6) ROLB Baker (68?) Cut in preseason
7) TE Byrne (58D)
So that's an A+ draft. For comparison sake, let's compare it to a randomly-chosen D+ team, the Falcons. QB 70, WR 70, TE 70, OLB 70, SS 70. They chose 21st. The Steelers were 20th.
1) FS Tate (81A) Same position and OVR, taken immediately after Steelers
2) RT DeChristopher (77B). One point below Steelers.
3) LOLB Williams (74B). One point ABOVE Steelers, chosen after Steelers.
4) LE Coples (73C). Two points above the Steelers' best value.
5) WR Rodgers (69C). One point below.
6) QB Thomas (68D). Equal.
7) TE Paulson (63D). 5 points higher.
So the Falcons got an OVR of 6 points higher than the Steeelers, and beat their best value, despite choosing AFTER the Steelers...and the Falcons have a D+ while the Steelers have an A+.
Test 4. I had saved immediately before the last draft (after changing all the rankings). Decided to run the draft again without changing a thing, just to see if it was all predetermined or not.
The entire first round went the exact same way...the Chargers still chose Butler at the beginning of the second round...and almost all teams still had the exact same 7 players at the end.
Almost. I had written down the outcomes of 15 teams after the draft from test 3. Almost all teams ended up with the exact same 7 guys,
with the exception of the Browns and the Steelers. They each had a different guy in their 7th slot. The other 13 teams were the exact same. I don't know about the 17 teams I didn't track.
Here's the most interesting point, in my mind. ALL TEAMS now had a D+ grade, even though the only difference for the Steelers was their 7th round draft pick (who, when they had an A+, was a 58D overall). How much worse than a 58D did they do this time, to end up with a D+ instead of an A+???
The world will never know.
The game corrupted while I was trying to access the coaching strategies.
I'm noticing a trend where, if I do anything unusual immediately before/after a draft (close out of a franchise, load a saved game, etc), the game explodes. Draft classes will not load, screens become distorted, etc. I have to restart my PS3 for everything to work again.
Major Conclusions:
1) Changing drafting strategies for each team DOES alter the CPU's drafting logic. It no longer takes the first player that increases it's overall. In Test 3 above, they choose QB Andrew Luck, who's OVR is lower than their current QB, Freeman. If they dropped just two slots, they could've improved their OLB by two OVR, but they chose Luck because QB was one of their top priorities.
2) At this time, it is impossible to predict HOW the CPU's drafting logic changed. A Punter was not taken in the first round...it fell to the beginning of the second round, where it was taken by a team that had a punter priority of 10. The rest of it's priorities were all 50-70 (except for FB and K).
3) A 10 point difference in priority is irrelevant. A 20 point difference is sometimes irrelevant. Even a 40 point difference will not guarantee a specific outcome.
4) I THINK there is a pattern I'm not seeing yet. X Difference between priorities means the CPU will be willing to drop X slots and X OVR for a position. If it can't find one there, it will take best available...even if that's a punter in the second round.
5) Not always the case though. Look at the Bears above. They couldn't find a position they wanted, yet they still didn't take best available. This may be falling back to the original logic of taking the first player who increases your OVR. Can't say for sure, but I'll keep poking it with a stick and find out.
08-31-2011, 02:28 PM
The CPU Is Cheating: Draft Logic Completely Explained (Including Punters!) UPDATED
