The Daily Spin – GPP Strategies Part III – 2017 Update
Last week, we started our journey into the numbers over the last year for the winning lineups in the $3 GPP events and how they tended to follow certain patterns in terms of the types of players owned and the ownership percentages that give us a better target for which to aim when we begin to construct our teams each week. While noting a few anomalies among some of the events, by and large, our small sample should help us in the coming season to begin to truly think in terms of lineup styles for particular events rather than simply throwing together lineup after lineup that seemingly looks good, but in reality may have little chance to win simply due to particular characteristics at the outset.
Now, these numbers are far from scientific. 36 tournaments is not going to give us anything conclusive to work with, but as the season goes on, I want you to start to consider how you are building your rosters and compare and contrast them with the ownership of winning rosters this season. We certainly do not know what DraftKings is going to do with pricing throughout the season, but my guess would be that they will at least start with something close to what we saw in 2017 and work from there. As the season wears on, try to gauge how difficult the pricing structure is and if there is more or less clustering based upon pricing being too soft or too harsh. If DK sticks with the adjustments that they made to contest sizes last season, the data should be a little easier to interpret earlier in 2018 than it was in 2017. This will not solve all of your lineup building issues, but it should provide a nice framework to start with.
Sign up for an FGI account today to see the rest of this post.
The main takeaways that we have gleaned from the numbers is that lineups should be constructed based on the type of contest entered. For cash games, players with low variance are important and overlapping ownership is not an issue worth worrying about too much. You are looking to minimize mistakes against a field that (hopefully) you believe you are better than average against and you do that by choosing players that tend to be predictable with a low standard deviation from their average performances, something which is easier said than done in PGA DFS from week to week as there are more variables to account for here that can change than any other DFS sport out there. It is what makes this game so much fun and also so wildly upsetting.
For GPP events, in order to have any sort of success, and particularly so in golf where you are only selecting six players, you really need to be able to separate yourself from competitors in finding those hidden gems that people either do not know about or are too scared to put into their lineups. While lineup building is an art for those that play fantasy sports with any regularity, we really think that newer players can benefit when first starting out by adhering to a stricter process that puts parameters in place to work with. It is too easy as a young player to go with the popular plays at each price level, almost assuring that they will lose early and often.
For a lot of you, there is a certain comfort level to doing your research and building your lineups each event. For me, it starts out every Friday afternoon for an hour or two in looking over the field which is announced at 4pm and listed on the PGA Tour page. Saturday and Sunday I spend watching the current tournament and looking over the course and course history for the next event. By Sunday, I usually have a list of 30-40 players that I want to look at more closely and on Monday I spend a few hours evaluating the salaries and odds when they are posted. Monday night, I convene with the FGI team and talk over the players that I like and listen to their input on players that they like and who looks good on our model. After the show, I spend a few additional hours on research and work with our stat guy on the final tweaks to the model and cut my list of players down to about 20. Jeff and I each do our own ownership percentage predictions and then share our numbers and post them to the site. As always, I check over the weather each day and also tee times on Tuesday to see if I can find any early advantage.
Throughout the day on Tuesday, I check in with any number of other players or writers that I respect and share my thoughts and listen to their ideas. It is a tricky process to take in as much as I can, but also not to become overloaded or to be too swayed from my own opinions. By around 8 or 9pm (okay, sometimes closer to midnight or 1am), I am ready to write my column which I typically finish around sunrise. Wednesday is when I typically get the chance to build my lineups.
To avoid having to deal with contests filling up early, I always select the contests I want to enter as soon as they are posted. Before salaries come out, I know how much money I will put into contests (roughly 15-20% of my bankroll) and the breakdown between cash games and GPPs. If you are serious about your results, you should be tracking your lineups and contests with spreadsheets. DraftKings does not allow you to access your old contests after 30 days, but that should be enough time for you to capture all of the data that you need. I want you to start tracking how each of your lineups performs, the percentage that each player was owned and also how the winning lineups were composed and scored.
If this is purely a recreational activity, then maybe you will not track this data, which is fine. This is after all, supposed to be for fun and a nice diversion to avoid the doldrums that arrive on weekends once the NFL season comes to its conclusion. However, if you want to be a winning player and take your game to the next level, you must exercise this type of discipline each week. The great players of the daily fantasy world are at their computer each and every day doing research, improving their models and becoming more efficient with their process. Fortunately, the world of daily fantasy sports and the popularity of golf is growing fast enough so that the percentage of these players overall in the fantasy universe is still relatively small. If you have a process that works for you, try augmenting it by keeping these types of records and if you find yourself to be struggling, definitely track your results closer as it is much easier to spot what mistakes you are making if you have the data available.
In building my lineups, I typically use 2-3 cash games each week for PGA and a mix of GPP teams. If the field is weak and the cut is tougher to predict, I typically do not play in the more expensive GPP events and expand the number of entries that I have in smaller events. I am looking to maximize the number of opportunities I have to get six players through the cut in weeks where only a small percentage of lineups will make it through intact. The more you track your results, the better you will get at gauging how much you want to play each week and the events you will want to enter. If this is really a long term hobby for you and something you envision doing for years to come, I would suggest sitting down at your computer throughout the weekend and jotting down a few notes here and there about the event and scoring conditions. Was the weather awful? Was the rough left a little higher than normal? Did one wave have much better conditions over another? Were there change to the course this year that made a difference in the scoring? Was someone playing through an injury? Any little factors that stood out can be noted and kept on file for the following year. It won’t always give you a big edge, but it’s useful for me as I read back through all of my old columns from over the years to see what my expectations were going into the tournament and how things actually played out.
Typically, I start building my teams from the most expensive entry down to the least expensive. For the $300 and $1500 GPP events, the first thing that I note is that ownership tends to spike around players much more so than at lower cost events for those most highly owned players. A player like Patrick Cantlay might be owned by 30% in the $3 GPP, 35% in the $27 GPP and 40% in the $400 GPP. Usually, this happens due to the pool of fantasy players that takes part in these higher dollar events. As the level of experience tends to be higher, these players tend to read a lot of the same resources each week and look at similar statistics which makes them more likely to arrive at similar conclusions when assessing the data, thus leading to more overlap in ownership with regards to the most popular picks each week. Use this to your advantage when entering events like this. It might be the best edge for players to exploit against small fields.
Why does avoiding the highest owned players tend to work well in the more expensive events with fewer entrants? It gets back to math and understanding what types of expected outcomes could occur for each player. I think one of the best examples of this from this happened with Kevin Chappell at the RBC Canadian Open in 2015. Chappell entered the event in good form having made seven straight cuts and coming in off of an 8th place finish at the John Deere Classic. Twitter was buzzing about Chappell all week. Most articles and shows gushed over his recent play and potential to play well yet again and at a price of $7,300, he looked like the steal of the week, even if his previous play at the course had been mediocre at best.
In the $1060 clubhouse event that particular week, Chappell was owned by 17 out of 35 entries, essentially half the field. Before the event had even started, the other 18 owners held a big advantage over the others in the field. To understand it a little better, we need to examine the potential outcomes for Chappell and the likelihood of each occurrence. To start, we will give Chappell a generous cut making percentage of 70%, a bit above his career average, but given his form entering the event, a fair starting point. In the five years leading up to the event, Chappell had ten Top 10 finishes out of about 130 events, 7.7% of events entered. About 25% of the time, Chappell finished in the Top 25 during that same time frame. The other 45% of the time, Chappell made the cut, but finished out of the Top 25.
A strategy that blindly ignores a player owned at a level this high in a typical event is profitable more often than not. In 30% of circumstances where Chappell will miss the cut, he will probably post a point total of around 30 points. Roughly 45% of the time, he will make the cut, but finish outside of the Top 25, and score roughly 70 points. 17% of the time, he will finish between 10th and 25th and score 90 points and 8% of the time he will finish in the Top 10 and score around 100 points. All of these numbers are based specifically on the Canadian Open where the scores tend to sore as it is one of the easier stops on tour each year. When you do the math, the average score for Chappell is 63.8 points.
75% of the time, Kevin Chappell is not a difference maker in helping an owner to win an event in our example. If he makes the cut and has an average performance, half the field received the same small benefit, but failed to distance themselves from the field in a meaningful manner. However, he will miss the cut often enough so that avoiding him and others at ownership levels this high will improve your odds of cashing in events with a small number of entrants. Think of it this way, the advantage gained in a GPP in owning a player like this who performs well is small, while the downside is large when they play poorly. The effects of this grow as the number of entrants in the field is lowered so that the strategy of avoiding these players is most profitable at the highest priced GPPs with fields of 30-100 entries. With cheaper events where there are thousands of entries, your entire lineup is still going to need to perform well in order for you to be competitive so that it is not enough to just avoid the landmines along the way.
Of course, each season, there are players where ownership levels build over time as the results from certain players build with a crescendo that does not shatter until the entire community of players gets taken down in one fell swoop. Brooks Koepka took on this role in 2015, Dustin Johnson followed in stride on 2016 and Luke List did the honors this fall. Each player sustained their success for months on end, producing solid results and exceeding value each tournament. Koepka did it by hammering away eagles and birdies to stack up huge fantasy points, even if he did blow up on enough holes to keep him just out of contention. By the time the FedEx Cup Playoffs came, even most of the early season doubters came aboard the Koepka train….and he crashed and burned badly over the last four events with touts primarily citing fatigue in the post season autopsy of what they had witnessed. Whatever the cause, those few who patiently held out against the bubble were eventually rewarded enormously when some of the bigger contests of the season arrived.
Dustin Johnson was different than both Chappell and Koepka in a few key respects. DJ had far more experience and success under his belt than either of the other two players, but the risk, of course, was that the price of DJ was going to be high even if he was not playing exceptional golf all summer. The risk with Chappell was a bubble around a highly inconsistent player. The risk around Koepka was whether or not a rookie on tour had the stamina to play an aggressive schedule and still finish strong when the big prize money was on the line. The risk with DJ was in paying a high price and needing a Top-5 finish most of the time to make him worth the purchase. Fortunately for the DJ believers who rode their horse through the FedEx Cup Championship, there was only a single landmine at the PGA Championship that proved to be painful along the way when DJ went absolutely off the rails in missing his only cut of the season. In 22 events, DJ finished in the Top-25 in nineteen starts, the Top-10 in fifteen and won three events. If you were all in on DJ every week, even with his high ownership numbers, you enjoyed a very successful season despite the risk of tying up a large chunk of salary with him as his price rose throughout the summer.
In 2017, the most interesting example I could find of ownership hubris took place late in the fall swing season with Mr. Fall Golf himself, Luke List. For those of you not hooked on the NFL drug each fall, Luke List tends to save his best work for the earliest events each season. As the fall events have an almost cult like following among the truly hardcore players due to all of the late night events that take place overseas, it becomes much easier to pick out where these type of crazy ownership clusters are going to spike. Coming into the fall, List had made six of seven cuts. He then went on to make the first four cuts of the fall season and had three straight Top-25’s heading into the OHL Classic at Mayakoba. A quick look at his history there showed a Top-10 finish the year prior so it looked like a great time to be in on List for the bargain price of $7,900.
Unfortunately, none of this information was a secret to anyone playing DFS golf during the fall season. In fact, it was just assumed that he’d be the highest owned player (guilty of know he would be highest owned and guilty of being too weak to fade him) that week and yet, like sheep, everyone flocked to own some List shares. Just as you would expect, ownership climbed with the level of buy-in: 32% in the $4 GPP, 37% in the $33 GPP, 43% in the $400 GPP and, get ready for this…..62.5% in the $1500 GPP for the week. Well, you know what happened next. List blew the cut and wiped away a huge swath of the field.
At those levels of ownership, how well does a player have to finish before you would regret fading him? Top-10? Top-5? If he’s 62.5%, I’d say Top-10 would hurt, but not kill you. Scott Brown was $100 more and posted 99 points for the week to go with his 6th place finish. If I posted their 2016-2017 results next to each other and covered the names up and made you guess who was who, very few would be able to do it. Would you like to try? Here is Player 1) 19 of 29 cuts made, 8 Top-25, 2 Top-10 Player 2) 20 of 31 cuts made, 9 Top-25, 3 Top-10 Both came to Mayakoba having made all four cuts in the fall season, with Brown having two Top-25 finishes and one Top-10 while List had three Top-25 finishes and one Top-10, so better, but not substantially so.
So how heavily was Brown owned for the week? 7.9% in the $4, 5.29% in the $33, 3.61% in the $400 and….this is the killer, 0% in the $1500. Isn’t it interesting that as ownership rose with List as the buy-in increased, it descended with Brown. This is powerful information and tells us a lot about the psychology of players.
First, it’s two different games between the low end and the high end for GPP events. I’ve discussed previously how as the buy-in rises, you do not need to be as contrarian as you would in $3 or even $33 GPP events. Luke List was in the winning lineup three times in the fall for teams in the $400 GPP. He did not make the winning lineup even once for the $3 or $33 GPP contests this fall, despite his overall strong play during the swing season. As we look at the numbers for aggregate ownership between the low and high dollar buy-ins, you’ll start to see that the approach is much different based on contests with just a few lineups versus events where there are tens of thousands.
Second, we don’t need to go crazy with our pivots to differentiate our teams. Too often I hear people talking about leaving way too much money on the table each week in an attempt to be contrarian or in digging up some unknown player to toss onto a lineup with limited information. In the example above, you literally needed to just pay an additional $100 for Scott Brown, a very reasonable substitute pivot under the circumstances.
Third, these mid-level value plays with clustered ownership are where we are most susceptible to getting out throats cut on a regular basis, especially in the smaller buy-in GPP events. It’s one thing when DJ is in the zone and posting Top-3 finishes week in and week out when he’s not ‘falling’ down the stairs in Augusta, but it’s an entirely different situation when the masses are going all-in on career journeyman Luke List, who has a much higher level of variance in his results for any given season, or hell, from one round to the next. When you see that $7,500 player this season that is garnering way too much attention based upon their historical play, don’t walk, run away from that player and catch the fade train out of town. If that player should happen to win that week, the 40%+ who owned him still had to look elsewhere to find their edge so owning that player did not even result in the advantage they would hope for in rostering the winning golfer! If you can flip the switch in your head to dodge these bullets consistently, you are going to face much less resistance in GPP events when we see massive chalk explosions.
This is where DFS diverges from outright bets. With outrights, we can all win together so if people go piling in together, it’s a blast when they hit and Twitter lights up with a 100 winning tickets. Unfortunately, in DFS, we need to embrace the mentality of the Highlander and realize that there can be ONLY ONE! Now, that’s not exactly true, as I have yet to pout about a 2nd place finish in a big event, but you get my point in that it’s not a place where you want to chase the crowd, though I promise, you and I will both be tempted to do so throughout 2018.
In analyzing the $300-400 GPP events over last season, we used data from 36 different tournaments. There were several weeks where promotional events were run for smaller buy-ins and we elected to focus this portion of the strategy piece on those events that had a cut with large diverse fields. We incorporated the fall events as well to give us a few more events to look at. Again, for all the data wonks reading this, we are certainly not suggesting that a sample size this small can be described as conclusive evidence for our ideas, but many of the trends observed seem to indicate fairly similar patterns in terms of the makeup of winning lineups. These are mere guidelines to start with in 2018 so I would suggest keeping some amount of flexibility available during the season as things evolve. Each year, DK has made small changes along the way so be ready for whatever pops up this year. When I began writing this analysis, I wanted to make it my goal that this series would be accessible to all players. Occasionally, I read papers that others put out in the industry that are at a PhD level and while they are well thought out and expertly presented, the information is going right over the heads of most readers and is not presented plainly enough or in a contextually accurate manner for most of us to glean anything that helps us to sharpen our play. The psychology of both players and owners is a big part of the game so I want to do my best to give you an honest assessment of what we see with the data and how to try to incorporate it into your own game each week.
A quick breakdown of the $300 GPP events that we tracked, which includes 33 events (non-majors, with a cut):
1) The aggregate ownership level for the winning lineup ranged between 72.2% at The CareerBuilder Challenge up to 170.9% for the Dell Technologies Championship. There were really only two events on the high end that I would qualify as outliers for the higher dollar events. The Dell and the Shriners. In both cases, one specific player played a massive part in creating these outliers: Patrick Cantlay. Cantlay became the feel good story of the year on the PGA Tour with his inspiring return to the PGA Tour that we covered very closely from his first tournament back all the way to his victory in Las Vegas. He has yet to miss a cut since returning and this has led to the bandwagon being overloaded to a point we have rarely experienced. As we’ve seen in other years, the trend in aggregate ownership does tend to rise a bit during the year as favorite players are anointed, but it was less pronounced this year.
This trend interested us as we tried to think about the reasons behind the numbers. Remember, we excluded small field, no cut events which wipes away the WGC events and the last couple of FedEx Cup events which we would naturally believe would have higher concentrations of ownership. My thought is that it indicates just how much the top players are getting their information from similar sources. Early in the year, there is always a little mystery about which players will emerge to play at an elite level. This is the time of year when the models produced are probably the least reliable and the pool of players that most owners are interested in owning is at its highest point, thus causing additional dispersion of ownership among those early fields. However, as the year unfolds, the data being used throughout these models becomes much more uniform, player form is easily observable, and pricing discrepancies are easier to spot as the Vegas odds tend to tighten up as the year wears on. Look for this trend again this year. If it comes to fruition, you will be able to make some small adjustments in how you build lineups in these larger buy-in events to exploit your competitors. Remember, the goal is to be just contrarian enough in building rosters so that you are not taking on too much risk. We are not going to have a Patrick Cantlay type of player who just keeps making cuts from the beginning until the end of the season as a young player. More likely, what we will see is the type of performance that Brooks Koepka had at the end of 2015 where the masses were burned again and again by a young player that had been so dynamic for most of the season, but ran out of gas by the end of the playoffs.
The other idea I thought about when I observed this trend was that players grew tired of searching for darkhorses last year as so many random players found their way into the winner’s circle week after week. Owners who got too creative in the early months were consistently burned by missed cuts. In 2017, getting six players through the cut was a monumental task, particularly when you were only able to enter a couple of teams. By the end of the season, many owners in the higher buy-in events got away from building contrarian style lineups and instead tried to fill a roster with cut makers. If only 3-5% of teams had the full compliment of golfers remaining over the weekend, it did not make as much sense to chase after sleepers, thus we end up with players like Luke List being 30%, 40% or even 60% owned for certain GPP events as confidence rises not so much that he will win, but that he just will not miss the cut. While both of these explanations do make sense to some extent, the limited sample size should keep you from thinking that this is gospel. However, being aware of these trends during the year will be helpful in knowing how to attack each event as the season wears on.
2) While the winning lineups in the $300-400 GPP were more chalky than those that won the $3 GPP events, there were still a fair amount of players that were owned on these rosters that were owned by a small percentage of the field:
– Number of players per winning lineups (0-6) who were owned between 0-5% in the field: 0.97
– Number of players per winning lineups who were owned between 5-10% in the field: 1.14
– Number of players per winning lineups who were owned between 10-20% in the field: 1.63
This indicates that in most cases, 1-2 of your players each week are going to need to come from that pool of underowned players, but with a bit less emphasis on the lowest range than the $3 GPP events.
3) Of the 74 players who were owned by less than 10% of the field on winning lineups, the categories for them were somewhat similar to those in the $3 GPP events with just a few small differences. There were still plenty of young players in that group, particularly during the fall season when there are plenty of new faces joining the tour for the first time. There was also a mix of veteran players that either had mixed results at the course, but in good form, or a solid track record at the course, but struggling with current form. This group also included a few international players that played sparingly on the US Tour and thus, did not attract the attention that other names would that play here every week.
What was less prevalent, were those players that seemingly were chosen for no apparent reason at all. In 2016, play was dominated by golfers who won events out of nowhere, but very few of these players made it onto winning lineups for the more expensive GPP events. In 2017, there were plenty of surprises as well. Si Woo Kim was such a surprise in winning The Players Championship that he did not make the winning roster for even the $3 GPP contest that week. DA Points was the random winner that stood out in 2017. He was under 1% owned, but found his way onto the winning roster for both the $3 and $33 GPP contest that week, but was owned by just one person out of 474 in the $300 GPP contest. In the $300 and up events, the risk/reward factor just is not there. Though there were a few noteworthy cases where golfers that were barely owned helped teams to win $300 GPP events, on the whole, it did not happen that nearly as often as the smaller buy-in events.
4) In terms of players that had higher ownership percentages for winning $300 GPP lineups:
– Number of players per winning lineups who were owned between 20-25% in the field: 0.77
– Number of players per winning lineups who were owned between 25-30% in the field: 0.71
– Number of players per winning lineups who were owned between 30-40% in the field: 0.46
– Number of players per winning lineups who were owned above 40% in the field: 0.31
What you notice when you compare the numbers between the $3 GPP and the $300 GPP are a few important differences. One similarity through all buy-ins is the inclusion of players between 5-20% where regardless of cost, winning lineups were made up of three players in this range with two being in the 0-10% range and 1-2 being drawn from 10-20%. The biggest change occurred when comparing the number of players owned by less than 5% of the field.
The real leap in the numbers came from the overall average for aggregate ownership which pushed all the way up to 106 for the big dollar GPP events versus in the low to middle 80 range for the lower dollar buy-ins. Much of this just comes from the fact that the fields are vastly larger for the small dollar events which allows for players to be a little more liberal in multi-entering a lot of teams where they are not worried about keeping every team intact. The standard deviation was also higher at just under 22 giving us a range of 84 to 128 for a single standard deviation in either direction. As there were no results below 72% and only a few below 80% all season, I would suggest using a low end target of around 85% as your starting point for building your lineups. On the high end, I would try to keep yourself from exceeding 125%. While there were a couple of winners that eclipsed that mark, much of this on the extremes was discussed above.
What the data is suggesting is that while some calculated risks are important for all GPP types of events, that the fewer the number of entrants that there are in the field, the less risk you need to take in finding low owned players. This may seem intuitive to a lot of fantasy players out there, but it needs to be stated here since there is still a lot of misunderstanding when it comes to lineup construction. It is not simply a matter of trying to build the best lineup, but a process that fits within a certain framework depending on the dollar entry and number of entries in the contest.
Here are a few helpful guidelines to get your started as you approach the $300 GPP events that we have discussed:
– Use a target of 85-125% ownership for your GPP teams, move towards the higher end as the season progresses
– Select 1 players you would predict to be in the low sinlge digit percentage of ownership
– Select 3 players you would predict to be owned by 5-20% of the field, one in the 5-10% range and one or two in the 10-20% range
– In most cases, avoid owning more than 2 players that will be owned by 25% or more of the field
– For players you select in the single digit range, utilize the categories mentioned above (young/new to tour or poor form/good history) to use around your core players. Wildcard plays can be avoided unless you have extremely good knowledge of a specific player and course that others will not know.
Obviously, the basics of lineup building still apply, but can be relaxed to some extent. With the $3 GPP events, it will rarely be a good idea to start your roster with three core players that will all be at least somewhat highly owned. However, in the $300-400 GPP events, this is not a problem and should be used most of the time. Since many times there will only be 5% of the field that gets six golfers through the cut, putting yourself in position against just a handful of others with full lineups gives you a chance to win, even if a couple of your players only do moderately well over the weekend. You only really need one or two players in a lineup that surprise to the upside to win a higher buy-in GPP so do not get too cute with your teams by trying to play a team full of overlooked players.
Making the necessary adjustments to building your lineups is the biggest change when moving up to more expensive GPP events during the season. You need to understand not just how certain players are performing or simple course histories. You need to understand that there are certain guidelines for any event and that they change based upon the number of entrants involved. If you need a crutch to help you out, start with the parameters in the last two articles and work from them until it becomes easier to transition smoothly from one event to the next. If you can get yourself to understand the components that go into building a winning lineup, you are going to be far ahead of most of your competitors each week.
Follow Us
