DFS Players EV as it Relates to BR

[mlbmodel]

A recent thread has sparked some interesting conversation on the DFS players’ EV (expected value) and the quantification as it relates to bankroll management. I have some ideas though admittedly I have not been able to come to a conclusion.

Here is the basic problem: Most DFS touts will say that bankroll management is a key to success…You can’t win if you bust out…yada yada yada. They then will offer several methods of maintaining what they would consider proper bankroll management techniques. Play cash games, grow bankroll by playing triple ups, don’t shoot your wad playing GPP. These are all fair things to say but mostly bullshit. If you were to inquire further some logical questions would be:

1. What is my EV (expected value) on a given slate? The answer would lay somewhere between what the Hell is EV and no one knows.
2. If you cannot calculate EV how does a player know if they are +EV? An EV player, meaning even value would expect to basically lose the rake, +EV would expect to profit some amount over the rake.

In essence without the ability to truly calculate your EV a player has no idea how much they should be wagering on a given slate. All slates would not be created equally and all contest types would offer varying EV on a given day. Price, teams, even the players in the contest pool would greatly effect daily EV. A pro in your pool of 10 players in a contest would greatly decrease an average players EV and thus should change the amount risked.

When I wager on sports I know exactly what my price is (the line -130). I then calculate what I believe the price to be (my line -140). The difference in the 2 tells me how much I should wager based on the Kelly Criterion. As you can see it is much simpler to know if I have an edge, how big is my edge and how much I should wager.

Given the fact that some very smart people are in DFS I am sure others have tried to quantify a daily edge and wager accordingly. If you have any ideas on how to do this please share.

[znmeb]

I’ve got code that analyzes your DraftKings contest history to determine your *past* EV for contests with fixed odds, like a Triple-Up. The same algorithm should work for any site that lets you download our history as a CSV file. Projecting it for a given slate is a lot harder – you pretty much have to simulate the other contestants’ skill levels and the games that will be played.

[rdtaylor9]

I’ve given it thought, but have yet to actually attempt to calculate a concrete EV. My sense is it would entail something along the lines of:

1. Projecting ownership %s for all players, and from that deriving an ‘average contest lineup’
2. Use your player projections to calculate the ‘average lineup score’ for opponents from that slate
3. Compare the ‘average lineup score’ to your expected lineup score on the given slate (you’d also have to have some sense about what distribution the scoring follows for that given sport and slate)
4. Finally, obtaining what % of the time you cash given xScore and xOpponentScore (and the scoring distribution)

Obviously, from there you have P(Success) and P(Failure), as well as contest payout (at least for non-GPPs), so it’s an easy plug into the Kelly formula.

This is obviously not ideal, but for now I make a simple and conservative estimate of roughly a 2% (varies depending on sport) edge and play slates at 1/2 Kelly. Definitely interested in other ideas.

[hagrin]

TL;DR – @radtaylor9’s thoughts mirror my own on the issue of quantifying edge and then applying it to bankroll management + utilize opponent tracking software / scrape opponent lineups to create a power ranking + a weighted recent results ranking to combat players now purchasing good tout plays + contest entering times for non-H2Hs.

Long Version
@rdtaylor9 – I think your post is closest to the correct answer and basically what I concluded using my past experience (+ the insights from @mlbmodel). It looks like @znmeb touched on something I tried in the past in his first response.

The only thing other thing I tried doing to some success was scraping the lineups of all the contests I could (against the ToS so gets tricky/technical) and evaluating the strength of the plays chosen by each contestant and creating an opponent power ranking. This actually led me to finding an RG Top 10 TPOY player who was a negative EV player in cash games. I ended up scooping almost all of his 2013-2014 NFL H2Hs until he finally stopped posting them. However, after this one player stopped posting H2Hs, I noticed something with several players the more they played – their results drastically and rapidly improved. One would assume that a player would gradually get better at DFS with more experience, but there were several players that went from hugely losing players to winning players seemingly overnight. My hypothesis is that they started using a site like this one to get better projections / plays which made all my old data useless and any potential edge calculation “impossible”. It was this obvious finding that led me to my question of diminishing edge in the other thread.

The above, in conjunction with all the key pieces @rdtaylor9 already pointed out, IMO, is the only way to truly determine a H2H edge in any predictive manner and potentially apply it to bankroll management. Tools to track results are absolutely something every player should have, but they aren’t predictive and will lead you to overestimating your actual edge – especially in a static priced DFS environment.

Speaking of H2H, game selection also will determine this edge quantification process. IMO, if you’re playing 50/50s or double/triple/quadruple ups, if you do maintain an opponent tracking ranking through scraping, you’re best severed by joining contests that are as close to full as possible so you can identify the strength of the field. For instance, even in as low as $2 contests, you may sometimes find that the first 10 entrants in a 100 field 50/50 are condia, 1ucror, 00oreo00, etc. and you’re already playing at a far smaller edge than you would in a normal 50/50. I have actually seen from some of the data I scraped from some qualifiers that is almost the opposite – that if a respected “pro” puts in the max amount of entries early, you’ll see the participation from the other pros fall below their normal participation number and the edge is actually greater than normal qualifiers. A good example of this was 2015 MLB I believe when BeepImaJeep would put his entries in early and you’d see other “pros” put in a much smaller amount of lineups than normal. Only after he maxed out his live final seats did you see some of those same pros start entering their normal amount of lineups again.

Obviously, the sooner you join a contest, the more limited the opponent information is, the harder it is to even approximate field strength / edge.

It’s just so super complicated and so hard to actually quantify that it’s a really great thought experiment / learning exercise, but far more complicated than sports betting. Really great discussion though, this is quickly becoming a daily stop of mine.

[znmeb]

@rdtaylor9:
“I’ve given it thought, but have yet to actually attempt to calculate a concrete EV. My sense is it would entail something along the lines of:

“1. Projecting ownership %s for all players, and from that deriving an ‘average contest lineup’
2. Use your player projections to calculate the ‘average lineup score’ for opponents from that slate
3. Compare the ‘average lineup score’ to your expected lineup score on the given slate (you’d also have to have some sense about what distribution the scoring follows for that given sport and slate)
4. Finally, obtaining what % of the time you cash given xScore and xOpponentScore (and the scoring distribution)

“Obviously, from there you have P(Success) and P(Failure), as well as contest payout (at least for non-GPPs), so it’s an easy plug into the Kelly formula.

“This is obviously not ideal, but for now I make a simple and conservative estimate of roughly a 2% (varies depending on sport) edge and play slates at 1/2 Kelly. Definitely interested in other ideas.”

Honestly I think that’s probably way too much effort for a very small potential edge. There are some easy things you can do that give you a lot of insight. For example:

1. Download your contest history CSV. I know this works on DraftKings and FanDuel – I haven’t played enough elsewhere.
2. The three columns you care about are total number of entries, the fantasy points you got, and the rank.
3. Now, compute the quantile for the rank. This is (1 – (Rank – 1) / (Entries – 1)). If you won, it’ll be 1.0. If you cashed or would have cashed a DK triple-up against the same competition, it’ll be greater than or equal to 0.7. If you cashed or would have cashed a 50-50 it would be greater than or equal to 0.5, etc.
4. Make a scatterplot. The quantile goes on the X-axis and the fantasy points on the Y-axis. If you fit a smooth curve through the points, you’ll see a funny “Lazy S” shaped curve and a bunch of points all over the map. But that curve tells you how many points you need on the average to cash a given contest type. And the points will show you how much this varies from slate to slate and sample of contestants to sample of contestants.
5. Cluster the points – this is where the magic happens. When I did this for NBA I got four clusters. Three of them were where you’d expect them to be – along the S-curve. But the *fourth* is high in score but low in quantile – I got lots of points and didn’t cash.

If you can tell whether it’s the slate, the competition or both that tends to give these off-curve points, you can just not play those contests. I haven’t done the grunt work of building the database to answer this yet. 😉

[ycmmat]

As far as saying “I have an advantage today (or on a given game) so I’m going to play more” like you can with wagering? I don’t think that concept exists in DFS. There are too many variables that any reasonable assumptions about ownership, scores etc quickly get out of hand.

However, I’ve found success with consistency 1) using my projections consistently. 2) playing consistent contests 3) playing for consistent dollar amounts per day/slate. It’s more taking the long view approach.

I’m more of a NHL/MLB guy so there are lots more contests to play than NFL.

[znmeb]

As a general rule, the more complicated a game is, the more advantage a good player has over bad ones. So DFS should reward “handicapping” skill more than money line, point spread, over-under or parimutuel wagering do.

IMHO for all “skilled chance” games with a house take, like poker, sports betting / prediction markets, racing and DFS, there’s a certain equilibrium point where the “good” players just barely profit over the rake. If the rake is too high, nobody plays the game, so there are no fish, whales or sharks. If the rake is too low, the house does not make a profit.

[Author]

Posts