DFS general strategy thread (1 Viewer)

[Bobcat10]

I disagree with Jeff on this. I have noticed in larger tournament on Fan Duel that 120 is cashing most of the time. Last week was the anomaly.

I don't buy into this. If 120 gets you into the top 18% (what most GPPs start paying out at) most of the time then it would get you into the top 18% of large Quintuple ups. If this happened most of the time then the quintuple up EV would make it by far the best play on fanduel. I love me some Quintuple ups but my experience is you need 138-140 to get into the money in a normal week.

I think msud remembers the strange weeks.

When I look in History and click on a GPP, it only shows my lineup and one other person's lineup. Is that person in the last cashing spot? There seems to be no way to look at the field of scores.

edit: scratch that, found one that is lower than my score and I didn't cash in it. Strange you can't see the field...

 

[Maurile Tremblay]

Maurile, based on your number of entries how statistically significant are your results?

I don't know. I've played in an awful lot of contests compared to the average person, but I suspect that you probably have to play in at least about five thousand GPPs (maybe more like twenty thousand) to even approach the long run in those. I haven't come near that. You don't have to play in nearly as many 50/50s to get to the long run, but I haven't broken my results down by GPPs vs. 50/50s or anything else.

I think it's a very good idea to keep track of all that stuff, but I've been lazy about it. I know I've played mostly GPPs, some 50/50s, hardly any H2Hs, and I've roughly tripled the amount I started out with over the course of the season so far. I also know that I've done much better, pretty consistently, at the small sites than I have at FanDuel or DraftKings. I'm actually down a little bit (less than $100 each) at both FanDuel and DraftKings this year.

 

[James Daulton]

How many lineups do most of you typically roll with? I'm having a hard time this week narrowing it down and as a result I have around 8 lineups revolving around largely the same WRs (Gordon, Landry, Still, etc) and either an expensive QB with cheap RBs or a cheap QB with expensive RBs. I have around 40% of my bankroll in play and I wonder if I'm so spread out that I'll have a hard time breaking even. It kind of feels like I should settle on just a few lineups but I don't want to miss what I consider good plays.

Ugh.

 

[snellman]

How many lineups do most of you typically roll with? I'm having a hard time this week narrowing it down and as a result I have around 8 lineups revolving around largely the same WRs (Gordon, Landry, Still, etc) and either an expensive QB with cheap RBs or a cheap QB with expensive RBs. I have around 40% of my bankroll in play and I wonder if I'm so spread out that I'll have a hard time breaking even. It kind of feels like I should settle on just a few lineups but I don't want to miss what I consider good plays.

Ugh.

Typically I will roll with 3 main lineups on Sunday. 1 "Sunday only" lineup and 2 lineups for Sun-Mon. I will roll out other tourney lineups, but my 3 main lineups go in 50/50s and double ups.

 

[pbandy1]

I have to say Maurile, however you do the h-value, you had some real nice picks. For example, J Stewart at 5200 was good value, but you still had him in the top 3 under h-value. You also nailed Roethlisberger, Olsen, and had some real suprising Def and K picks: Panthers, Gano, Josh Brown, Giants, etc

 

[massraider]

How many lineups do most of you typically roll with? I'm having a hard time this week narrowing it down and as a result I have around 8 lineups revolving around largely the same WRs (Gordon, Landry, Still, etc) and either an expensive QB with cheap RBs or a cheap QB with expensive RBs. I have around 40% of my bankroll in play and I wonder if I'm so spread out that I'll have a hard time breaking even. It kind of feels like I should settle on just a few lineups but I don't want to miss what I consider good plays.

Ugh.

I rolled with 12 lineups.

Tell you something I have learned, there is money to be made in those 50/50s. Not as sexy as winning $250,000, but I played ONLY tournaments this week, and while I only lost a few bucks, I would have added 35% to my bankroll if I had ignored them, and just done 50/50.

 

[Bri]

I have to say Maurile, however you do the h-value, you had some real nice picks. For example, J Stewart at 5200 was good value, but you still had him in the top 3 under h-value. You also nailed Roethlisberger, Olsen, and had some real suprising Def and K picks: Panthers, Gano, Josh Brown, Giants, etc

Brown and Giants weren't all that surprising. Titans were top 5 in "points against"(really points for but...) for defense and are starting arguably the worst offensive line ever to be on an NFL field with a rookie QB. JPP did what Watt did the week before. Kicking I'm having a terrible time predicting and took the Titans(thus Josh Brown) off same points allowed for FF chart. I'm not saying it wasn't sweet to get 21 points for the Gmen (dopey penalty would have been 27) but there's not much to learn here.

Stewart was almost misplaced as he was a starter with DeAngelo hurt, going against a meh defense and still listed amongst backups at fanduel. He did better than I thought, but he stuck out to me right away.

I'd like to know more about the best/name skill position players. I have now started Manning and Jimmy Graham when they had their worst weeks in years. I've seen Peyton light up Buffalo for a zillion years. They've had good and bad Ds and it has never mattered til yesterday.

Antonio Brown never seems to have a bad game so I go with him.

I went with Beckham and AJ Green, separately, so one of them with Brown. Beckham has been red hot and the Titans pass D is brutally bad so that was easy. The Steelers pass D isn't very good but I did not in anyway think Green would have a monster game. I suppose if you went Calvin instead of Green you were still happy, but Josh Gordon stunk.

What is your stat that makes you say "ooh that guy" when you're looking at the best of the best?

Hopkins had a career game, Beckham about the same. Does that mean you "have to" go with Percy Harvin against the same Titans D this week? During the Giants game, their starting CB's shoulder popped out and their S suffered a concussion so you could argue they are actually worse off now. Harvin (regardless of team) has been terribly unpredictable IMO.

 

[Keerock]

I have to say Maurile, however you do the h-value, you had some real nice picks. For example, J Stewart at 5200 was good value, but you still had him in the top 3 under h-value. You also nailed Roethlisberger, Olsen, and had some real suprising Def and K picks: Panthers, Gano, Josh Brown, Giants, etc

Brown and Giants weren't all that surprising. Titans were top 5 in "points against"(really points for but...) for defense and are starting arguably the worst offensive line ever to be on an NFL field with a rookie QB. JPP did what Watt did the week before. Kicking I'm having a terrible time predicting and took the Titans(thus Josh Brown) off same points allowed for FF chart. I'm not saying it wasn't sweet to get 21 points for the Gmen (dopey penalty would have been 27) but there's not much to learn here.

Stewart was almost misplaced as he was a starter with DeAngelo hurt, going against a meh defense and still listed amongst backups at fanduel. He did better than I thought, but he stuck out to me right away.

I'd like to know more about the best/name skill position players. I have now started Manning and Jimmy Graham when they had their worst weeks in years. I've seen Peyton light up Buffalo for a zillion years. They've had good and bad Ds and it has never mattered til yesterday.

Antonio Brown never seems to have a bad game so I go with him.

I went with Beckham and AJ Green, separately, so one of them with Brown. Beckham has been red hot and the Titans pass D is brutally bad so that was easy. The Steelers pass D isn't very good but I did not in anyway think Green would have a monster game. I suppose if you went Calvin instead of Green you were still happy, but Josh Gordon stunk.

What is your stat that makes you say "ooh that guy" when you're looking at the best of the best?

Hopkins had a career game, Beckham about the same. Does that mean you "have to" go with Percy Harvin against the same Titans D this week? During the Giants game, their starting CB's shoulder popped out and their S suffered a concussion so you could argue they are actually worse off now. Harvin (regardless of team) has been terribly unpredictable IMO.

Harvin likely out the rest of the year with an ankle... so no. But maybe Decker. Problem is he has Geno Smith throwing to him, not Fitz or Eli.

 

[Bri]

But Fitz isn't good and he got 6 TDs against the Titans. I don't like the idea of Geno versus Fitz as options for debate, so I might have to go away from this but (good point) Decker is going to intrigue me now.

 

[Tennessee_ATO]

The problem w/ taking a WR against the Titans is threefold: (1) Titans are horrible against the run too, (2) Titans are so bad in the secondary that every WR is open on every play, and (3) Titans are so bad offensively that games are over before halftime.

Eli and ODB could have combined for infinity points, but a long TD run and defensive score minimized their output significantly.

The NYJ-Titans game might be the worst played game of the season. Offensive ineptitude meets defensive incompetence. What happens when a resistible force meets a movable object? Tune in next week.

 

[David Dodds]

Even though I rarely play GPPs, my gut tells me that they have the highest ROI by a wide margin (for above average players). I say that because I think MOST of the entries going into these GPPs are flawwed.

From my blog this year (October 8th post):

My thoughts on creating a roster that can win FanDuel's Sunday Million (or another large GPP).

1. You need a pretty significant level of uniqueness with 4-5 players (of your 9 roster spots) at a minimum. Ideally, these unique players should be owned less than 5% of the time.

2. You should only have 1 player per NFL team UNLESS one of the players is the QB. This refers to the term negatively correlated players. You actually want to pair your QB with a player from the same team in an attempt to get double points on a correct QB play.

3. Since it will take a score of 180+ to win a giant FanDuel tournament, you need players that will score 3X their salary. This week Peyton Manning costs $10,200 and likely represents good to great value in cash games (50/50s / H2H, etc). But he would need to throw for 350 yards and 4 TDs (with no interceptions) for him to triple his salary. It's certainly possible against the Jets, but in general the highest salaries are not winning plays for a large scale GPP.

With this 3X guide in play, I prefer to build my roster with mid-priced selections that have strong matchups.

4. I want WRs that can get a large percentage of their production on one play (ie deep threats) that play in games that Vegas believe will be high scoring (even if their perceived role is lessened)

5. For defenses and kickers, I want to avoid the 3-4 most popular at each of these positions. This gets me 2 of my 4 unique spots as stated above.

6. With few exceptions, RBs perform way better in wins than in losses.

Of course with any set of rules, someone can point me to a roster that broke all of the above and won a big tournament. Of course any roster can win (especially if it is perfect). The guidelines above though can give you a lot of shots on goal (because of the uniqueness) where it doesn't take a perfect lineup to win (just some of the unique players to come through).

 

[massraider]

Great post, thanks David!

 

[CR69]

Excellent info, thanks for sharing.

To your point on "uniqueness", how do you check the % owned before the game actually starts?

 

[David Dodds]

Excellent info, thanks for sharing.

To your point on "uniqueness", how do you check the % owned before the game actually starts?

Early in the year, you could just enter a Thursday contest to get this info. It's tougher to get now as FanDuel only shows % owned on players you take, but can still be achieved through multiple low priced entries. We have articles on it now at FBG. Additionally we are working on some math that we will debut next year that will calculate this (that we will use in our app and on the website)

 

[ffinmyblood]

Great post David, but I guess the question is what are we considering "mid priced"?? $6,000's?? 7's?? Also, is it rare that you will ever have a player of $9,000 or more on your roster??

Do you have pricing "guidelines" as a general rule per position that you will not spend Over X amount?? For example, you will not spend more than say $15,000 combined at RB??

 

[msudaisy26]

I disagree with Jeff on this. I have noticed in larger tournament on Fan Duel that 120 is cashing most of the time. Last week was the anomaly.

I don't buy into this. If 120 gets you into the top 18% (what most GPPs start paying out at) most of the time then it would get you into the top 18% of large Quintuple ups. If this happened most of the time then the quintuple up EV would make it by far the best play on fanduel. I love me some Quintuple ups but my experience is you need 138-140 to get into the money in a normal week.

I am on my phone right now, but when I get to my computer I will look it up. I can recall at least twice I had teams in the 120's that cashed

I can't figure out how to actually look at other pages of the scores now that the contest is over, but here is what I got. Note these are all 5 dollar tourneys and it starts in week 2.

week 2 - I could only get in to the Sunday rush 2 since I deposited late in the week. Paid out 4732 I scored 144 and finished 900th.

week 3 - I was in the regular Sunday Rush and I didn't cash. Payed out 9680 people. I didn't cash best score was only 100 and 23000 place.

week 4 - Regular Sunday Rush and didn't cash. Payed 9680 people. My best was 121 for 24000 place.

week 5 - Regular Sunday Rush and I cashed. Payed out 9680 people. My closest to the cut line was 113 finishing in about 9400th

week 6 - Regular Sunday Rush and I cashed. Payed out 9680 people. My closest to the cut line was 128 finishing in about 7900th

week 7 - Regular Sunday Rush. Payed out 9680 people. I had one finish at 9978 out of the money with a score 118.

week 8 - Regular Sunday Rush. Payed out 9680 people. I had one score of 143 finish 9100th.

week 9 - Regular Sunday Rush. . Payed out 10544 people. I had one score of 147 that finished in 14800th.

week 10 - Regular Sunday Rush. Payed out 10544 people. I had one score of 153 that finished 5900th.

week 11 - Regular Sunday Rush. Payed out 10544 people. I had one score of 132 that finished 5150th.

week 12 - Regular Sunday Rush. Payed out 16260 people. I had one score of 131 that finished 15500th

week 13 - Regular Sunday Rush. Payed out 16260 people. I had a score of 125 that finished in 41000th.

So far this week I have a team at 110 that is cashing in the Sunday rush again

I would say more weeks than not a score of mid 120's to low 130's is cashing in this tourney

 

[Bobcat10]

Seems like to collect the appropriate data you have to get it when the games of the contest end but before payouts are complete.

 

[ctrlaltdefeat]

I hate to burst bubbles but strategy makes no difference unless you're buying in to the big tourneys anywhere between 50 and 75 times. The guys who win these big gpp tourneys purchase multiple entries and win with totally random lineups. Don't waste your money

 

[Bobcat10]

I hate to burst bubbles but strategy makes no difference unless you're buying in to the big tourneys anywhere between 50 and 75 times. The guys who win these big gpp tourneys purchase multiple entries and win with totally random lineups. Don't waste your money

I only play the single entry GPPs for the most part.

 

[massraider]

I hate to burst bubbles but strategy makes no difference unless you're buying in to the big tourneys anywhere between 50 and 75 times. The guys who win these big gpp tourneys purchase multiple entries and win with totally random lineups. Don't waste your money

My buddy finished first in a $10 Turkey Day GPP. One entry. 25 grand. He's a law student who wagers a hundred a week. What are you talking about?

 

[jeaton6]

I hate to burst bubbles but strategy makes no difference unless you're buying in to the big tourneys anywhere between 50 and 75 times. The guys who win these big gpp tourneys purchase multiple entries and win with totally random lineups. Don't waste your money

My buddy finished first in a $10 Turkey Day GPP. One entry. 25 grand. He's a law student who wagers a hundred a week. What are you talking about?

Nice! So slightly off topic here but at what amount of profit do they send you a W2?

 

[Bobcat10]

Guy winning the $70k Squib has a majorly flawed lineup right?

Stafford, Bell, Calvin

AWilliams, Beckham, Josh Brown

Also picked the Browns defense against the Colts ( .7% ).

 

[Bobcat10]

I hate to burst bubbles but strategy makes no difference unless you're buying in to the big tourneys anywhere between 50 and 75 times. The guys who win these big gpp tourneys purchase multiple entries and win with totally random lineups. Don't waste your money

My buddy finished first in a $10 Turkey Day GPP. One entry. 25 grand. He's a law student who wagers a hundred a week. What are you talking about?

Nice! So slightly off topic here but at what amount of profit do they send you a W2?

$600 net profit

 

[jeaton6]

I hate to burst bubbles but strategy makes no difference unless you're buying in to the big tourneys anywhere between 50 and 75 times. The guys who win these big gpp tourneys purchase multiple entries and win with totally random lineups. Don't waste your money

My buddy finished first in a $10 Turkey Day GPP. One entry. 25 grand. He's a law student who wagers a hundred a week. What are you talking about?

Nice! So slightly off topic here but at what amount of profit do they send you a W2?

$600 net profit

Oof. So I'm up about about $700. Sounds like I'm best off playing tourney only and hoping to hit big and if not get just below the $600 Hundo and play elsewhere. Am I better off losing $100 just to get below that threshold? Not that anyone here would support tax evasion

 

[Ignoratio Elenchi]

Seems like to collect the appropriate data you have to get it when the games of the contest end but before payouts are complete.

This should work for recently completed contests:

http://o.ffense.com/fanduel/contests.php

Just go on FanDuel and click on one of your live or completed contests. If the URL looks like this:

fanduel.com/contest/NUMBER/scoring/lineup/OTHERNUMBER/

Just copy the NUMBER and paste it into the box on my website. If it doesn't look like that (and instead it's something like fanduel.com/entry/ABUNCHOFLETTERS) it's not going to work.

Note, I just whipped that together in the past 20 minutes or so and have done extremely little testing, so no guarantees. If it works it should make it easy for you to see what score was needed to cash (this is the "Last Winning Score").

 

[BigJohn]

Seems like to collect the appropriate data you have to get it when the games of the contest end but before payouts are complete.

This should work for recently completed contests:

http://o.ffense.com/fanduel/contests.php

Just go on FanDuel and click on one of your live or completed contests. If the URL looks like this:

fanduel.com/contest/NUMBER/scoring/lineup/OTHERNUMBER/

Just copy the NUMBER and paste it into the box on my website. If it doesn't look like that (and instead it's something like fanduel.com/entry/ABUNCHOFLETTERS) it's not going to work.

Note, I just whipped that together in the past 20 minutes or so and have done extremely little testing, so no guarantees. If it works it should make it easy for you to see what score was needed to cash (this is the "Last Winning Score").

Contest ID 8133185 Name $80K Sun NFL Snap #3 Sport NFL Entries 45977 Entry Fee $2 Prizes $80,000.00 GPP Yes High Score 201.48 Last Winning Score 127.98 Last Winning Rank 7310

 

[Maurile Tremblay]

Guy winning the $70k Squib has a majorly flawed lineup right?

Stafford, Bell, Calvin

AWilliams, Beckham, Josh Brown

Also picked the Browns defense against the Colts ( .7% ).

Looks good to me. I mean, any lineup that scores a lot of points looks good to me, but that lineup looks good even according to conventional tournament-lineup snobbery as I understand it.

Stafford-Calvin is a good stack. Williams-Brown and Beckham-Brown are also positively correlated pairs.

Bell-Calvin and Williams-Beckham are slightly negatively correlated pairs, but the positive correlations outweigh the negative ones by a decent margin.

(Stafford-Bell is neutral.)

Browns D is lightly owned. Dodds has been saying for a long time that defense and kicker are good positions to go for lightly owned players.

 

[Wadsworth]

For GPPs I start by looking at every game. Check the Vegas line and get an idea how the general public expects the game to go. I try to find 2-3 games that I think could go a completely different direction. In these game I look for clear players that would benefit from the unexpected result. If Carolina goes off against the NO last week, it is safe to predict that Cam, K-Ben and Olsen will be the primary beneficiaries. Next I try to predict game flow for that game. If Carolina starts thumping NO, will the Saints be able to keep it close enough for Carolina stay aggressive throughout the game. If you think so, then consider adding a player from the Saints to that line up since for this line up to work the game will need to stay competitive so why not double up on it. At that point I fill out the rest of the roster with value pick.

Look for creative pairings, not just QB/WR. RB/Def, WR/WR or TE without the QB is he is very expensive, Def/WR who also returns punts(double points on a return TD). Try to predict the unlikely and build a roster that will put up a huge number if the unlikely happens.

That is the Cliff notes version, but it has worked pretty well for me so far.

 

[Wadsworth]

Guy winning the $70k Squib has a majorly flawed lineup right?

Stafford, Bell, Calvin

AWilliams, Beckham, Josh Brown

Also picked the Browns defense against the Colts ( .7% ).

Looks good to me. I mean, any lineup that scores a lot of points looks good to me, but that lineup looks good even according to conventional tournament-lineup snobbery as I understand it.

Stafford-Calvin is a good stack. Williams-Brown and Beckham-Brown are also positively correlated pairs.

Bell-Calvin and Williams-Beckham are slightly negatively correlated pairs, but the positive correlations outweigh the negative ones by a decent margin.

(Stafford-Bell is neutral.)

Browns D is lightly owned. Dodds has been saying for a long time that defense and kicker are good positions to go for lightly owned players.

I almost never run a QB/RB/WR stack but actually did it this week with Cam, J-Stew and K-Ben. If the Saints had kept that came close I might be a lot richer today. That line up had 105 points before the end of the first half of 1pm games on Sunday. it just died after that. Whatever wins is the right line up.

 

[msudaisy26]

Seems like to collect the appropriate data you have to get it when the games of the contest end but before payouts are complete.

To get exact data yes, but I think this shows you should throw your 5050 lineup in the GPPs.

 

[Keerock]

Has anyone figured out whether or not multiple entry Double-Ups/Triple-Ups are a better/worse/equal play to single entry?

 

[ajm27501]

Has anyone figured out whether or not multiple entry Double-Ups/Triple-Ups are a better/worse/equal play to single entry?

I haven't done any math, but my gut would say that it's probably a better play than a single entry, because in theory the second/third/etc. entries will be less than optimal entries. But there are many people in this thread way smarter than me.

 

[Bobcat10]

Guy winning the $70k Squib has a majorly flawed lineup right?

Stafford, Bell, Calvin

AWilliams, Beckham, Josh Brown

Also picked the Browns defense against the Colts ( .7% ).

Looks good to me. I mean, any lineup that scores a lot of points looks good to me, but that lineup looks good even according to conventional tournament-lineup snobbery as I understand it.

Stafford-Calvin is a good stack. Williams-Brown and Beckham-Brown are also positively correlated pairs.

Bell-Calvin and Williams-Beckham are slightly negatively correlated pairs, but the positive correlations outweigh the negative ones by a decent margin.

(Stafford-Bell is neutral.)

Browns D is lightly owned. Dodds has been saying for a long time that defense and kicker are good positions to go for lightly owned players.

I almost never run a QB/RB/WR stack but actually did it this week with Cam, J-Stew and K-Ben. If the Saints had kept that came close I might be a lot richer today. That line up had 105 points before the end of the first half of 1pm games on Sunday. it just died after that. Whatever wins is the right line up.

I totally get whatever wins wins in the end, but I thought that went against some of the common themes I've read around here. A kicker with two players you choose to score tds and a qb-rb-wr stack that I don't even think was a high Vegas line. Who knows I'm far from the expert on these matters.

 

[Tennessee_ATO]

Guy winning the $70k Squib has a majorly flawed lineup right?

Stafford, Bell, Calvin

AWilliams, Beckham, Josh Brown

Also picked the Browns defense against the Colts ( .7% ).

Looks good to me. I mean, any lineup that scores a lot of points looks good to me, but that lineup looks good even according to conventional tournament-lineup snobbery as I understand it.

Stafford-Calvin is a good stack. Williams-Brown and Beckham-Brown are also positively correlated pairs.

Bell-Calvin and Williams-Beckham are slightly negatively correlated pairs, but the positive correlations outweigh the negative ones by a decent margin.

(Stafford-Bell is neutral.)

Browns D is lightly owned. Dodds has been saying for a long time that defense and kicker are good positions to go for lightly owned players.

I almost never run a QB/RB/WR stack but actually did it this week with Cam, J-Stew and K-Ben. If the Saints had kept that came close I might be a lot richer today. That line up had 105 points before the end of the first half of 1pm games on Sunday. it just died after that. Whatever wins is the right line up.

I totally get whatever wins wins in the end, but I thought that went against some of the common themes I've read around here. A kicker with two players you choose to score tds and a qb-rb-wr stack that I don't even think was a high Vegas line. Who knows I'm far from the expert on these matters.

I've long maintained that some RB are very good stacks w/ their QBs. If a RB projects to catch 4+ balls, I think he should be considered in a stack. 6+ and I think he should be stacked. Bell is often the #2 recreiver in terms of targets for the Lions. Stacking him w/ Stafford makes a lot of sense to me.

 

[Tennessee_ATO]

Posted this in the week 14 thread last night. Some data for you number crunchers.

Here are the FD cash games I did and the score required to win for those tracking such things:

Thursday --

2x ($5) 115.5 (1818 entries)

2x ($10) 122.22 (112)

3x ($10) 128.5 (100)

5x ($5) 129.2 (223)

Sunday --

2x ($2) 1116 entries in each 4 total: 112.74, 113.4, 111.7, 112.64

2x ($5) 670 entries in each 7 total: 112.2, 110.4, 113.5, 113.4, 112.1, 110.2, 113.16

3x ($5) 402 entries in each 8 total: 119.1, 118.8, 120.6, 120.76, 117.8, 119.3, 120.6, 121.82

5x ($5) 335 entries in each 5 total: 126.32, 125.22, 125.66, 126.44, 126.9

 

[Tennessee_ATO]

Has anyone gathered enough data to have reliable information about the relative value of spending above the minimum salary at each position? For example, to roster a team (assuming you have a starting QB) this week, you have to spend $42,400 ($4500 for everything but QB, $6400 for Bortles). Obviously we know that a $4500 RB is typically useless while a $4500 PK is not. I wonder at one point we tend to see diminishing returns on salary above the minimum at each position.

I realize on any given week that value will be higher or lower based on specific salaries and specific performances. Variance is relentless after all. That said, there has to be value in knowing that generally spending more on RBs than WRs (or vice versa) results in higher scores, or typically the return on a dollar flattens at QB (or RB or WR) at $x.

 

[dochavok]

I considered starting a new thread on H-value, but figured I'd drop it in here.

I've been struggling to identify high scoring value plays, and I like the new H-value indicator on the value charts. Any chance you can post a link to a writeup by Dan Hindery where he describes the process used to come up with the metric (or even just more info on it)?

There is one part of the description on the Interactive Value Charts that I think is a little misleading:
"H-value, a measure inspired by Dan Hindery, is often a better indicator of true value than points per dollar."

My specific issue is the wording "a better indicator of true value". I would argue that technically, H-value generally identifies a (marginally) worse value, but a better play. I'll work with a week 15 example at QB (using Maurile Tremblay's early values):

Andrew Luck, for $10.1K, with a projection of 23.5 points (429.78 $/pt) [H-Val 46.9]
Cam Newton, for $8.1K, with a projection of 20 points (405 $/pt) [H-Val 44.3]

In this case, the better value is Newton. He's simply projected to get more points per dollar spent. However, Luck has a slightly better H-value. I think what the H-value is trying to indicate is that there's a lost opportunity cost in taking Newton over Luck that's not fully evident in looking at the pure value ($/pt) numbers.

In playing Newton over Luck, theoretically a DFSer would need to "make up" the missing 3.5 points somewhere else. Naturally, the $2000 saved by selecting Newton would be spent on another player, and the 3.5 points would be made up by being able to spend up. Those "lost" points will essentially cost 571.43 $/pt ($2000 / 3.5pts), which is more expensive than either of the QB options cost on their own. However, $571/pt isn't necessarily that hard to make up somewhere else.

As I write this, I'm realizing that the example I've selected really highlights the confusion that comes with relying solely on the H-value when trying to determine the value of a player. I'm currently looking at 18 WRs and 20 RBs under the $571/pt mark, which means that by selecting Newton I can put my saved dollars to better use elsewhere (maybe upgrading from Justin Forsett [7700/17.1pts] to Le'Veon Bell [9600/26pts]), and using that 2000 to more than make up for the 3.5 points I'd lose in selecting Newton. Luck/Forsett (17.8K for 40.5pts) is clearly not as good value as Newton/Bell ($17.7K for 45.9pts).

Where I think that the H-value functions best is by using it across all positions. Sorting all positions by H-Value would have immediately identified Bell as solid play this week. Starting there might have bypassed the whole discussion, but my goal was to find two players at the same position who had similar values, where the H-value numbers were contradictory.

I think that the main takeaway is that the H-value certainly has value in identifying potential plays, but should not be used alone to find the best value play. Specifically, I think that the H-value would be best used when comparing two plays with equal values ($/pt), and identifying which player would be better to roster. Or, to identify a player across position types where their "bang for the buck" might not be optimal, but certainly would help in setting a lineup. Sometimes the best approach is not necessarily finding the cheapest play per dollar; sometimes finding the second or third best value but spending a higher percentage of the cap can be a better strategy. Maybe it just comes down to semantics on what one considers the "true value". However, at the end of the day the goal is to pack as many DFS points into a roster using a limited amount of dollars. H-value on its own doesn't do this, but it can identify where to start.

 

[nastynate]

Amount of Entries:

To start the year, I enter 2 main lineups into cash games, and diversify with 5-6 GPP lineups.

At this point, I just don't feel like I have the time, or the ability some weeks, to put together more than 1 really solid lineup. So, I enter my best, then diversify off of that for 2-4 additional GPP lineups. I've had much better luck this way, although the weeks are a bit more volatile. My main lineups have been 113-148 and I've had GPPs between 89-174.

Probably limiting my upside a bit, but with 30-40 minutes to spend on DFS / week, I feel this is my best strategy to consistently win.

Also, been avoiding H2H more and more and spending more on 2x, 5x and 50/50 games.

 

[Maurile Tremblay]

I've been struggling to identify high scoring value plays, and I like the new H-value indicator on the value charts. Any chance you can post a link to a writeup by Dan Hindery where he describes the process used to come up with the metric (or even just more info on it)?

I used to just have points and value listed on the interactive value charts, where "value" was defined as points per dollar (multiplied by a constant to improve readability).

Dan's observation was that the guys who were the best values by that metric were not necessarily the ones who would help your team the most. To construct a high-scoring team, you need to (a) find players who are good values, and (b) use up most or all of your salary cap.

If you just took the nine guys who were the best values (in terms of fantasy points per dollar), you'd spend 74% of your salary cap and score 85% of the points scored by a team that uses its whole cap. [Numbers pulled from my rear, meant only to be illustrative.] You scored more points per dollar, but fewer points. Ultimately, it's total points that matter, not points per dollar.

H-value is an attempt to help you accomplish both (a) and (b) above, whereas value is only about (a). H-value, therefore, is a better indicator of which players are most likely to have a place in an optimized lineup.

Dan's original formula was to multiply value by projected points. Since value is projected points divided by salary, Dan's formula was really projected points squared divided by salary. I found that squaring projected points put a bit too much emphasis on points, in my view, and raising to a power of 1.5 didn't emphasize points quite enough, so I'm currently splitting the difference and raising to a power of 1.73. (Why 1.73? That's the square root of three, which just sounds cooler than 1.75.)

Dan's thoughts on this topic are not published anywhere, so far as I know, so I cannot give you a link to his write-up. He offered his observation in a casual discussion among the staff.

There is one part of the description on the Interactive Value Charts that I think is a little misleading:

"H-value, a measure inspired by Dan Hindery, is often a better indicator of true value than points per dollar."

My specific issue is the wording "a better indicator of true value". I would argue that technically, H-value generally identifies a (marginally) worse value, but a better play.

Yes, I struggled with the wording there. "Value" is a bit ambiguous. I defined it, for purposes of the charts, as projected points divided by salary -- but in that sentence I'm using "true value" to mean helpfulness to your team. A player with a high H-value is expected to help your team win. So will a player with a high regular value, but I think H-value is a more reliable indicator of that than regular value.

I will try to re-write that part with less ambiguity when I get a chance.

 

[jeaton6]

I disagree with Jeff on this. I have noticed in larger tournament on Fan Duel that 120 is cashing most of the time. Last week was the anomaly.

I don't buy into this. If 120 gets you into the top 18% (what most GPPs start paying out at) most of the time then it would get you into the top 18% of large Quintuple ups. If this happened most of the time then the quintuple up EV would make it by far the best play on fanduel. I love me some Quintuple ups but my experience is you need 138-140 to get into the money in a normal week.

I am on my phone right now, but when I get to my computer I will look it up. I can recall at least twice I had teams in the 120's that cashed

I can't figure out how to actually look at other pages of the scores now that the contest is over, but here is what I got. Note these are all 5 dollar tourneys and it starts in week 2.

week 2 - I could only get in to the Sunday rush 2 since I deposited late in the week. Paid out 4732 I scored 144 and finished 900th.

week 3 - I was in the regular Sunday Rush and I didn't cash. Payed out 9680 people. I didn't cash best score was only 100 and 23000 place.

week 4 - Regular Sunday Rush and didn't cash. Payed 9680 people. My best was 121 for 24000 place.

week 5 - Regular Sunday Rush and I cashed. Payed out 9680 people. My closest to the cut line was 113 finishing in about 9400th

week 6 - Regular Sunday Rush and I cashed. Payed out 9680 people. My closest to the cut line was 128 finishing in about 7900th

week 7 - Regular Sunday Rush. Payed out 9680 people. I had one finish at 9978 out of the money with a score 118.

week 8 - Regular Sunday Rush. Payed out 9680 people. I had one score of 143 finish 9100th.

week 9 - Regular Sunday Rush. . Payed out 10544 people. I had one score of 147 that finished in 14800th.

week 10 - Regular Sunday Rush. Payed out 10544 people. I had one score of 153 that finished 5900th.

week 11 - Regular Sunday Rush. Payed out 10544 people. I had one score of 132 that finished 5150th.

week 12 - Regular Sunday Rush. Payed out 16260 people. I had one score of 131 that finished 15500th

week 13 - Regular Sunday Rush. Payed out 16260 people. I had a score of 125 that finished in 41000th.

So far this week I have a team at 110 that is cashing in the Sunday rush again

I would say more weeks than not a score of mid 120's to low 130's is cashing in this tourney

I agree that typically a score in the mid 120's will cash. You'd be amazed at how much that extra 5 points from 120 to mid 120's makes a difference though in standings.For what it's worth I've compiled my GPP entries through week 13 and here's what I've found.

I've cashed in 65/150 GPPs. My average cashing score was 148.28. 24 of these GPPs I cashed in the lowest payout tier (not including contests that had an overlay). The contests I've tracked range from paying out anywhere from 13.7% (mini rush) and 18.9% (Squib). Of the 24 I cashed in lowest tier my minimum score was 116.2, max was 174.2 with an avg of 132.5.

In 6 of these I finished in top 15.2-16.6% which would have put me very close to cut line and my avg score was 129. Far from statistically significant but at least in my case it looks like mid to upper 120's is about right.

 

[ceejee]

When referring to a player hitting "value", what is the $/pt ratio people look for? 400? 350?

 

[Keerock]

When referring to a player hitting "value", what is the $/pt ratio people look for? 400? 350?

Generally, you like for a player to hit 2pts per $1000 in cash games and 3pts per $1000 in GPP (this is using FanDuel pricing)

 

[ceejee]

When referring to a player hitting "value", what is the $/pt ratio people look for? 400? 350?

Generally, you like for a player to hit 2pts per $1000 in cash games and 3pts per $1000 in GPP (this is using FanDuel pricing)

Thanks. And for DK?

 

[Keerock]

When referring to a player hitting "value", what is the $/pt ratio people look for? 400? 350?

Generally, you like for a player to hit 2pts per $1000 in cash games and 3pts per $1000 in GPP (this is using FanDuel pricing)

Thanks. And for DK?

I don't use DK, so someone else will have to answer...

 

[Tennessee_ATO]

When referring to a player hitting "value", what is the $/pt ratio people look for? 400? 350?

Generally, you like for a player to hit 2pts per $1000 in cash games and 3pts per $1000 in GPP (this is using FanDuel pricing)

But be careful getting too focused on that # tho. That's what you're "aiming" for as a whole. Specific players and positions will vary. If I can get 2.5x from a high-salary guy in a GPP I'm usually thrilled. A $5000-6000 guy I'm actually looking for 3.5-4 times. Think about it: a 25 point day is pretty good even tho you paid $9700 to get it.

 

[donkey]

Question for the FBG staffers....considering that historical salaries are maintained in the weekly content, do you have the resources to write a script to look at the optimal score for each week? It would be very interesting to see a sample of what the top available score each week was. I would be interested in looking at:

• What % of salary was spent at each position each week?
• Does the optimal lineup normally contain a QB/Catcher stack (we all play them because of the upside)?
• I target positions to reach 3x value like everyone else for tournaments, but do optimal lineups consistently have a position that over or underperforms that target?

Result may be completely random, but I think it would be awfully cool to get that type of look for FanDuel.

Posting the optimal lineup for the previous week as early week content would be pretty sweet - at least to me.

 

[snellman]

Ignoratio Elenchi, on 11 Dec 2014 - 09:12 AM, said:

Can someone walk me through how to use the Export Lineup feature at FD. Lets say I wanted in enter 25 $2 50/50s using the same lineup
TIA

If you already know what lineup you want to use before you enter all the contests, you'd use the Import feature. Enter one contest and put in the lineup you want. Then when you enter each subsequent contest, just import that lineup.

If you've already entered the contests and you want to change the lineup in all (or some) of them, go to the Upcoming Entries page. Go in to edit an entry that you want to update and put in the new lineup you want. When you submit, you can then click a button to export that lineup to other entries. You'll be taken to a page with a list of all your entries, and you can select which ones you want to export the new lineup to.

Also can you place the same lineup multiple times in the same 50/50 and if so is it better to do that or enter the same lineup in 25 different 50/50s

To my knowledge none of the 50/50s allow you to submit multiple entries. A similar contest format to the 50/50 is the Double Up, and some of those allow multiple entries. You'll see these marked with an "M" in an orange circle to indicate that they allow multiple entries.

IMO the primary advantage of playing in low-stakes 50/50s is that the field is larger. In a $5 50/50 you're playing against 99 other people. In a $50 50/50 you're only playing against 19 other people. If you're just entering the same lineup I wouldn't see any real purpose to putting it in the same 50/50 even if you could, then you'd just effectively be reducing the size of the field you're competing against.

Entering the same lineup in multiple contests serves some purpose to reduce variance. Let's say you're wagering $50 on a single lineup, and it ends up scoring 120 points. If you enter a single $50 contest, or put 10 identical $5 entries in the same contest, either way you're going to win $90 or win $0. Nothing in between. But if you enter the same lineup in 10 different contests, you might win 7 and lose 3 (because even though your lineup will score the same exact amount in every contest, the score required to cash in any given contest can vary). So you results will be a little moderated. If you're just getting started and trying to steadily build up a bankroll, this is probably what you want to do. If you're just playing for the thrill and don't care about money, then a different approach might be fine.

Perhaps even better than entering the same lineup in 10 $5 contests, you might want to enter similar, but slightly different lineups, in 25 different $2 contests or something. As hard as we try to predict things, football is unpredictable and you may not want to be all in on a kicker if this is the week he puts up 0 points or something. Pick 2 different QBs you like, 3-4 RBs, etc. and mix and match a little bit. That way even if you guess wrong on one guy, you can still win money with your other lineups.

 

[Bobcat10]

I see the same 4 or 5 $2 $2K double ups today that I did days ago. These start on Sunday. Do they start turning over more after we get past tonight? Or do I need to dip into Sunday 50/50's to increase my $2 plays?

 

[Tennessee_ATO]

I see the same 4 or 5 $2 $2K double ups today that I did days ago. These start on Sunday. Do they start turning over more after we get past tonight? Or do I need to dip into Sunday 50/50's to increase my $2 plays?

You'll see an increase in contests after tonight I suspect.

 

[Megla]

Am I playing too many different games?

$5.00 H2H [SIZE=medium]3[/SIZE]

$1.00 50/50 [SIZE=medium]2[/SIZE]

$2.00 50/50 [SIZE=medium]5[/SIZE]

$5.00 50/50 [SIZE=medium]2[/SIZE]

$10.00 50/50 [SIZE=medium]2[/SIZE]

$25.00 175K THUR NFL BOMB [SIZE=medium]1[/SIZE]

$5.00 600K SUND NFL RUSH [SIZE=medium]1[/SIZE]

$5.00 40K SUN NFL SPIKE [SIZE=medium]1[/SIZE]

$10.00 450K SUN NFL KICKOFF [SIZE=medium]1[/SIZE]

$25.00 2MILLION SUNDAY MILLION [SIZE=medium]1[/SIZE]

[SIZE=medium]It seems those that are doing well simply load up on low level 50/50s and some leagues. Do you guys also play some GPPs just for the fun of it?[/SIZE]