ezlee2
Junior Member
Posts: 79
|
Post by ezlee2 on Sept 25, 2018 14:03:12 GMT -5
I know this is not a new topic by any means, but after all these years there are still plenty of folks that have no clue what bars are static and what they predict. Dave Hibb gave the best explanation on the old vNFL forums, but unfortunately they are no longer available. Does anyone want to work on a collaborative project to help the rest of the FOF community?
|
|
|
Post by Squirrel on Sept 25, 2018 14:58:17 GMT -5
Rob, I'll get this started.
The table below sets out which of the bars are 'static'.
Static Static Static Static % Pos Bar 1 Bar 2 Bar 3 of OVR C Blk_str - - 20.0% CB Pun_hit - - 1.8% DE PR_str Pun_hit - 29.7% DT PR_str Pun_hit - 28.3% FB P_Ins Blk_str - 15.6% G Blk_str - - 22.0% ILB PR_str Pun_hit - 13.7% OLB PR_str Pun_hit - 12.2% QB Sens_r - - 4.7% RB B_speed P_Ins Spd2Out 30.9% S Pun_hit - - 3.6% T Blk_str - - 19.1% TE Big_p - - 22.4% WR Big_p - - 8.2%
What experienced MP FOF players mean when they say 'static bars' is that some bars you see on a draft prospect correlate closely with the actual bars the prospect will have as a player, others don't.
To give an example, if you take a big sample of DE prospects who are good enough to stay in the league long enough to develop to their potential, and you run the corelations between the midpoint of the draft prospect's bar and the midpoint of the developed player's bar, it will look like this:
Pass Rush Strength: 0.966 (remember 1.000 is a perfect correlation so this is close) Punishing Hitter: 0.957 PR Tech: 0.766 (so a much lower correlation, and arguably the most important bar for a DE) Endurance: 0.798 etc etc...most of the other bars are around 0.8 correlation
Broadening that out to other key bars
RB Hole Rec 0.801 vs Breakaway Speed 0.964, Power Inside 0.961, Speed to Outside 0.965 WR Get Downfield 0.845 and Route Running 0.748 vs Big Play 0.966
And so on and so on.
A few other observations:
1. There is a school of thought that for players with high statics, the other bars are more predictive i.e. the PRT bar on a DE prospect with high PRS is more reliable than it would be if the prospect has low PRS. This effect can be observed clearly across large populations although it doesn't always work with specific players
2. There are quite a few solid drafters therefore who concentrate on statics above pretty much all else. In my experience I can think of a few top-quartile drafting GMs for whom I think this is the cornerstone of their style (not naming names)
3. This effect works better for some positions than others. In the table about I have including a column to show how much of the overall rating is determined by statics. If you're drafting to get high overall ratings, and you're using statics to help you get there, you'll do particularly well at DE, DT and RB
4. Some statics seem to matter a lot in the dice rolls (e.g. QB sense rush), others don't (punishing hitter). I'm not going to start quantifying this but it's fairly clear to be that some positions are much easier to draft than others, and that therefore those positions can be taken down the draft more reliably than others.
|
|
|
Post by Squirrel on Sept 25, 2018 14:59:48 GMT -5
|
|
|
Post by Squirrel on Sept 25, 2018 15:00:37 GMT -5
(Table in second post as I couldn't figure out how to format the text into columns)
|
|
ezlee2
Junior Member
Posts: 79
|
Post by ezlee2 on Sept 25, 2018 16:53:59 GMT -5
This is great stuff squirrel! Even the graph above is a great visual for guys to start to wrap this around their head. For me, when I truly started paying attention to them (per Nick's insistence) I found my drafting to improve 500% almost immediately. This was years ago, but I can remember the "a-ha" moment when after multiple PMs sent between Nick and I that it finally stuck and I started investing the time to really prepare for a draft. That topic in itself is worthy of it's own thread Still, this was a game changer for me in how I participated and enjoyed the draft. I still see guys trade away their top picks for proven players time and time again because they're afraid to waste a draft pick. with a player that could bust.
|
|
ezlee2
Junior Member
Posts: 79
|
Post by ezlee2 on Sept 25, 2018 16:54:22 GMT -5
BTW, you should register squirrel!
|
|
|
Post by nemesis on Sept 28, 2018 11:29:25 GMT -5
Squirrel, do you see any standard deviation correlations between how a combine lines up with say, Blocking Strength/Bench, and Pass Blocking/Agility. To give an extreme example, say a Tackle has a Bench of 25, blue bar range of 75-100. Pass Blocking blue range of 25-50 and an agility score of 7.50. Are you expecting the blue range to be higher and will pass on him, or expect that bar to jump?
Or are combine scores pretty independent of each other?
|
|
joel
New Member
Posts: 7
|
Post by joel on Sept 28, 2018 11:45:09 GMT -5
Squirrel, do you see any standard deviation correlations between how a combine lines up with say, Blocking Strength/Bench, and Pass Blocking/Agility. To give an extreme example, say a Tackle has a Bench of 25, blue bar range of 75-100. Pass Blocking blue range of 25-50 and an agility score of 7.50. Are you expecting the blue range to be higher and will pass on him, or expect that bar to jump? Or are combine scores pretty independent of each other? Here is how I look at it, when I draft and it comes to making bar to combine comparisons. 1. If the bars look great, and I know the leagues combine/bar setting is 39, I don't care about the combine. 2. If the bars come in at under where the combine says he would be, I would firstly assume that he is just a combine guy, (and will need an interview) as long as the Static then matches the relative combine score. 3. If the combine scores come in under where the bars say, especially in the static, then I become very intrigued, as I see this a sign that the game is trying to throw you off this players trail. And there is the potential for the player to creep. 4. If the combine matches the static, but the other combines don't match, then I don't see much deviation as to what will happen to bars. However, if the combine that goes with the static is under the bar, then I expect all of the bars to bump up. If the combine for the static is over, then I expect all the bars to go down. This is kind of how I draft. I have been practising with weights recently, as to which bar holds more weight to a position. I really could not care about a WR without RR, but max everything else. And I am becoming very big on 3rd down passing for a QB as that holds 14% of the players overall, the biggest % on any QB bar.
|
|
|
Post by nemesis on Sept 28, 2018 13:59:30 GMT -5
Squirrel, do you see any standard deviation correlations between how a combine lines up with say, Blocking Strength/Bench, and Pass Blocking/Agility. To give an extreme example, say a Tackle has a Bench of 25, blue bar range of 75-100. Pass Blocking blue range of 25-50 and an agility score of 7.50. Are you expecting the blue range to be higher and will pass on him, or expect that bar to jump? Or are combine scores pretty independent of each other? Here is how I look at it, when I draft and it comes to making bar to combine comparisons. 1. If the bars look great, and I know the leagues combine/bar setting is 39, I don't care about the combine. 2. If the bars come in at under where the combine says he would be, I would firstly assume that he is just a combine guy, (and will need an interview) as long as the Static then matches the relative combine score. 3. If the combine scores come in under where the bars say, especially in the static, then I become very intrigued, as I see this a sign that the game is trying to throw you off this players trail. And there is the potential for the player to creep. 4. If the combine matches the static, but the other combines don't match, then I don't see much deviation as to what will happen to bars. However, if the combine that goes with the static is under the bar, then I expect all of the bars to bump up. If the combine for the static is over, then I expect all the bars to go down. This is kind of how I draft. I have been practising with weights recently, as to which bar holds more weight to a position. I really could not care about a WR without RR, but max everything else. And I am becoming very big on 3rd down passing for a QB as that holds 14% of the players overall, the biggest % on any QB bar. Squirrel?
|
|
joel
New Member
Posts: 7
|
Post by joel on Sept 28, 2018 14:09:58 GMT -5
Not squirrel, just Joel, but I thought I’d answer anyway
|
|
|
Post by nemesis on Sept 28, 2018 14:53:38 GMT -5
Not squirrel, just Joel, but I thought I’d answer anyway Ah ok, just making sure.
What you say makes sense.
I also have considered running regression data on seeing which bars hold the most weight for wins. Which will also reveal what position has the highest weight. But I'm far too lazy in my old age now.
|
|
|
Post by Squirrel on Oct 1, 2018 5:44:52 GMT -5
@ nemesis sorry man, not 100% sure I get your Q. Do you mind breaking it down a bit? I'm not following
Process-wise what I used to do is consider the prospect vs other past players with similar bars...but when I do that I pay particular attention to the similarity in the static as it seems to weigh heavier than most. Once I had done that for a while I realised I was basically drafting to prototypes and once I noticed that I simplified my process accordingly down to just looking for guys that fit the description.
I did start to go through the process of trying to work out wins > stats > bars > prospect bars using regression at at one point. I do think it's doable and would make me better at the game but I haven't put in the yards to actually do most of it
|
|
|
Post by bdubbs on Oct 2, 2018 6:01:33 GMT -5
Hey squirrel, I saw your spreadsheet on predicting QB avoid int rate and it was really interesting. What I'm noticing though is that no matter what math people have tried to apply to evaluating players it always comes with the caveat that it doesn't really work on the individual level. I see it like counting cards in blackjack, the change in the odds isn't huge, but the edge reveals itself over time.
I like it though because it gives the rest of us who draft on more of a "feel" than really delving deep into the numbers a chance to compete with luck. It makes me wonder what portion of a draft class actually fits these models and how much of the class is intentionally made to be significantly better or worse than what their bars / combines imply
|
|
|
Post by nick on Oct 2, 2018 9:34:52 GMT -5
That spreadsheet is awesome. We have it working now so that it works for all players. It also works for both current and potential on younger players.
|
|