Fantasy Football - Footballguys Forums
This is a sample guest message. Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!
Demeco Ryans (1 Viewer)
- Thread starter huckyoda
- Start date
Jene Bramel
Footballguy
I'll ask Norton to address this when he gets a chance.
I'm not in agreement with some of the players projected to have better raw numbers above Ryans, but I always like to point out that there's really not much difference between the raw numbers of many of the players in the top 25. The #4 and #17 ranked players (by FBG scoring) are separated by two total fantasy points (a little over a single solo tackle), as are the #10 and #33 ranked players. If you asked us to rank players by confidence in the stated projections, I'd be very surprised if Ryans wasn't in the top ten and probably top five. His matchup is fine and there aren't many I'd start over him in any week, including Week 11.
I'm not in agreement with some of the players projected to have better raw numbers above Ryans, but I always like to point out that there's really not much difference between the raw numbers of many of the players in the top 25. The #4 and #17 ranked players (by FBG scoring) are separated by two total fantasy points (a little over a single solo tackle), as are the #10 and #33 ranked players. If you asked us to rank players by confidence in the stated projections, I'd be very surprised if Ryans wasn't in the top ten and probably top five. His matchup is fine and there aren't many I'd start over him in any week, including Week 11.
Islander
Footballguy
2 pt difference is huge. If you were a betting shop and offered your clients to bet on IDP "matchups", the difference between offering a bet on player A vs player B straight up, and offering A+2 vs B-2 is probably the equivalent of giving 7 pts on an NFL game, if not more. When Norton has two players projected virtually equal (less than 0.5 pt difference), in theory that means he would be neutral in taking bets on one player or another (he thinks they have 50/50 chance of scoring higher than the other). I find many projections are not very good and the IDP projections are FBG subscriber content that could be improved. One example is Aaron Ross this week. He's projected DB1 by 0.4 pt over all other DBs and only two other DBs come within 1 pt of Ross. Regardless how good A Ross's matchup is, I could probably pick 10 DBs to match up against A Ross and I would be confident that at least 6 of them will beat Ross this week. Now if in addition you give me +1 on all of them except those that Norton ranked DB2 and DB3, then I really have a strong edge. A Ross might turn out to have a good game and he might beat 8 or 9 of the DBs I pick. But I feel like if I did this every week, picking on the "bad" projections, I would be right 70-75% of the time, which means the projections could be improved (in the perfect world, all players who have equal projections should have 50/50 chance of beating each other). Last offseason I intended to give constructive criticism about the IDP projections but never got around to it in a serious fashion. Maybe next offseason.D Ryans this week is another good example. Calvin Pace is ranked right ahead of him. I am sure Ryans has more than 70% chance of beating Pace this week, so Ryans should be significantly ahead of Pace in the projections.
Jene Bramel
Footballguy
This is a great topic for debate in the offseason.I've seen Norton's method for doing the projections. He has a monster spreadsheet that details how many points have been scored at each position. It's broken up by week, by team and by position (WLB/MLB/SLB and CB/S). He can consider any emerging trends at a glance, he can consider which scheme seems to work best against which team, he can consider changes in supporting cast among other issues.I know I often come off as a Norton apologist in these threads, but it's really not the case. I made this point when I did my yearly quick set of projections for my May dynasty rankings -- you'll be very surprised at what you'll find if you go through the entire process and stay true to what your analysis tells you. It would be (and is) very easy to look at what the final result is and say that you aren't confident that things will shake out that way. But then, which decision making process is better? The one you've spent a long time researching and considering or the sling from the hip changes when the list doesn't look right. And that's for a 16 game season, which significantly reduces the variance, but is still tough to get right by season's end.Should we go through the effort of projecting every player? Is it more accurate than a confidence cheatsheet without details? Hard to say. The projections allow subscribers to use the MyFBG function to customize a cheatsheet for their league, though, so they aren't likely going anywhere. I'm sure John is open to all suggestions on how to improve the process and the result. Also, FWIW, in FBG scoring I'm going to continue to argue that a two point difference (+/- 1pt in my original example) isn't significant for the projections. I'm no mathematician and I'd be interested to see exactly what the number that is statistically significant is, but with all the variables involved in the projections I find it hard to believe that you can reliably project a group of 30 players to within 1.3 solos or less than two assists with all the variables involved. I have no idea how you mathematically prove that a 4pt difference is the equivalent of a TD in a Vegas line, but I'm interested.I'm not sure what the right answer is here. I can assure you that I've had this discussion with Norton for probably ten years now, before FBG and before RedEye. I can also assure you that we're all ears.I'm looking forward to the discussion this offseason. I'm hoping that some of our vets and newly interested posters can start considering some of these questions. The research done here in the past three years on home and road stat crews, how soon a trend is truly a trend, among others is going to help pin these issues down a bit more. I'd love to see some research on consistency scores -- similar to what's been done on the offensive side -- and some more insight on the trend issues that RommelDAK posted earlier (i.e. are preseason DB1s more likely to break out of a funk than DB3s). I'd love to look into the tackle opportunity more and I'm hoping to see still more defensive data from Football Outsiders and others that we can use to make things tighter.I'm just hoping I can find the extra time to undertake some of that myself.
Islander
Footballguy
I will make an effort to make sure we revisit this issue in the offseason with a good discussion.
For the moment, here is a simple example:
When looking at an individual player, you would tend to think that +/- 1 pt is not significant because if player A averages 10 pts per game, in reality he's scoring between 5 pts and 15 pts in each individual game, so 9, 10, or 11 both seem "ok" as projections. That's why I think my argument can be better understood by comparing player A to B. If B who also averages 10 pts per game, and his scores are always between 5 and 15 like player A, then you could say his expected value for the current week is equal to A (both have a projection of 10 assuming matchups are "average"). Each has a probability distribution that might look like this:
5 pts 1%
6 pts 4%
7 pts 8%
8 pts 12%
9 pts 15%
10 pts 20%
11 pts 15%
12 pts 12%
13 pts 8%
14 pts 4%
15 pts 1%
By projecting 10 pts for each, then if you have to bet on A or B it does not matter, they both have equal chance of beating the other. Without going into the details, in that scenario there is 43.5% chance that A wins, 13.0% chance of tie, and 43.5% chance that B wins. It's a fair bet.
However if you give me A+1 (if A scores 5 and B scores 5 I win, if A 5 and B 6 it's a tie, if A 5 and B 7 or more I lose, etc.), suddenly I have a huge edge. If I get A+1, I have 56.5% chance of winning, 12.3% chance of tie, and 31.3% chance of losing. If you give me a fair bet on this, I will clean you out in the long run. Just one point is a significant advantage.
The link to NFL games can be done by looking at the distribution of final scores. If a game is pick'em based on Vegas, historical results of all "pick'em games" might tell you something like this:
Team A wins: 50%
Team B wins by 1-6: 20%
Team B wins by 7: 5%
Team B wins by 8+: 25%
(numbers are hypothetical)
You can calculate the equivalent bias by giving me +7 instead of pick'em, I go from 50/50 chance of winning/losing to 70% chance of winning, 5% chance of tie, and 25% chance of loss.
That's how I can translate an x pt difference in projections for IDP into y points on NFL games.
Now the conversation gets a lot more complicated for IDPs if we break down projections into sacks, INTs, etc. but let's keep this for another day
For the moment, here is a simple example:
When looking at an individual player, you would tend to think that +/- 1 pt is not significant because if player A averages 10 pts per game, in reality he's scoring between 5 pts and 15 pts in each individual game, so 9, 10, or 11 both seem "ok" as projections. That's why I think my argument can be better understood by comparing player A to B. If B who also averages 10 pts per game, and his scores are always between 5 and 15 like player A, then you could say his expected value for the current week is equal to A (both have a projection of 10 assuming matchups are "average"). Each has a probability distribution that might look like this:
5 pts 1%
6 pts 4%
7 pts 8%
8 pts 12%
9 pts 15%
10 pts 20%
11 pts 15%
12 pts 12%
13 pts 8%
14 pts 4%
15 pts 1%
By projecting 10 pts for each, then if you have to bet on A or B it does not matter, they both have equal chance of beating the other. Without going into the details, in that scenario there is 43.5% chance that A wins, 13.0% chance of tie, and 43.5% chance that B wins. It's a fair bet.
However if you give me A+1 (if A scores 5 and B scores 5 I win, if A 5 and B 6 it's a tie, if A 5 and B 7 or more I lose, etc.), suddenly I have a huge edge. If I get A+1, I have 56.5% chance of winning, 12.3% chance of tie, and 31.3% chance of losing. If you give me a fair bet on this, I will clean you out in the long run. Just one point is a significant advantage.
The link to NFL games can be done by looking at the distribution of final scores. If a game is pick'em based on Vegas, historical results of all "pick'em games" might tell you something like this:
Team A wins: 50%
Team B wins by 1-6: 20%
Team B wins by 7: 5%
Team B wins by 8+: 25%
(numbers are hypothetical)
You can calculate the equivalent bias by giving me +7 instead of pick'em, I go from 50/50 chance of winning/losing to 70% chance of winning, 5% chance of tie, and 25% chance of loss.
That's how I can translate an x pt difference in projections for IDP into y points on NFL games.
Now the conversation gets a lot more complicated for IDPs if we break down projections into sacks, INTs, etc. but let's keep this for another day
I projected him at 7-2 with .7 PD, .2 int and 14.1 fantasy points. I'm not sure how that can be called low
Or why Pace shows up ahead of him. Pace was at 13.9 points on the early projections, just behind Ryans. I'll have to check into this and see if there is a glitch. At any rate, early in the week there were some concerns with the health of Reggie Bush (concussion) and Ryans (ankle) so I knocked his tackle numbers down a couple until I could get a read on those 2 issues. Both guys seem GTG and I bumped Ryans back up a little on the update. Good luck everyone
Last edited by a moderator:
