The Great American GPP: The Home Run Concordance: aEV+ and Coming into Form
In my last post we discussed the use of advanced stats compared to traditional stats, and introduced the Statcast measurements of Exit Velocity (EV) average EV (aEV) and Exit Velocity Against. We traced the correlation between EV and a the probability of a home run, and proposed a simple method for quickly adding this to our daily batter research. This week, we’ll take a deeper look at EV and other Statcast data in order to build our home run concordance.
Over the weekend there were a few exciting batted ball events and some amazing fantasy point explosions to go with them. I hope that a few readers added the EV method to their research, and decided to chase the HR upside and play a few batters with increasing aEV. Personally, I enjoyed Corey Seager performance on June 3rd, when he hit three solo dingers and collected 42 Draft King points. He gave an encore performance with 2 homers and 4 RBI’s on June 5th for another 35 DK points. I enjoyed Seager’s personal game of HR derby partly because I played him that night, but also because there had been so much dim witted chatter about him recently. Seager was hailed in spring training as one of the best young players in the game, and major part of the Dodger’s offense. He had a strong first week of the season, prompting more of the same praise and an increase of $600 in his price. He cooled down for the rest of April, and just as quickly, sections of the baseball press/blogosphere turned on him. This silliness was fueled by his long power outage, as he hit only two home runs in April and had only 4 by mid May. This drought included stretches of 10 games/48 at-bats, and 16 games/64 at-bats between home runs. By mid May, there were calls for sending “one of the games best young players” back to the minors, and he was generally abandoned in fantasy play. I included this narrative about Seager’s season and the media because for the majority of the baseball world (dfs owners included) the model they use to evaluate players is little more than a mix of current box score performance,‘streaks’, media narrative and perception.
Seager had some good games mixed in during this downswing, but the cause wasn’t a loss of his talent or the other odd things that were said about him, but some bad luck on balls in play, and being out of form for periods of time. Around the middle of May, Seager began to come back into form, as his aEV, avgDIST and Hard Hit%* began increasing. On 5/28 the salary for Seager had dropped to $2,900 on DK ( a day he scored 17 points btw); his salary was 3,800, with an ownership of 6.6% when he hit is three homers on June 3rd; his cost had jumped to $4800 two days later when sent two more out of the yard and collected 35 DK pts. As of June 7th, Seager cost over $5,000 on DK. I can’t think of much better evidence for the concrete value of taking your research a bit further to model the batter’s current form.
Bloggers don’t have permission to embed here, but please take a second and click to see a chart of Seager’s aEV through the 2016, charted with his home runs. This is the perfect example of ‘coming into form.’
Indeed, Seager saw an increase over his season averages in EV, aEV, Distance, and HH percentage in the 15 days leading up to his homer happy weekend, which just happened to coincide with the end of his early season power outage, as 11 of his 14 home runs occurred in this span. As late as May 11th, Seagers aEV was 89 mph (an Exit Velocity at which there were only 5 out of 3,995 batted ball events as of June 7, 2016).
The progress of aEV (or HH% if that is all you have available,as aEV is a major factor of how HH% is calculated—just remember they are not the same thing) over a period of a few days can help you make a calculated guess in the direction of the quality of at-bat performance, an even luck, of a hitter. If the aEV or HH% trends down, this may be a danger sign the player is regressing, but if EV is trending up, there is a good chance the player is ‘heating up’ and may start to produce offense. Indeed, we can even say that as the EV of bated balls trends up, the batter is putting themselves in a position to get lucky, and a little bit of luck can turn EV+ balls into doubles, triples, and HR’s. I’m going to call balls hit during an upward trend in Exit Velocity, EV+, because it reminds me of ‘pos EV’ in poker, and hopefully the result of noticing EV+ in batters is the same as making a ‘pos EV’ play in poker: Profit.
High aEV, and high aDIST only occur when a player is in full control of their mechanics, have their swing and timing tuned, and are consciously performing well at the plate. Add in an evaluation of a player’s plate discipline (an upcoming topic), and you have a fairly accurate model of if a batter is in full control of their skill and playing very well, (and thus getting results they should or getting unlucky) or if a batter is up there hacking away, swinging at everything, with their timing and mechanics a mess (and getting the poor results they deserve, or getting lucky and playing over their head). The really fun stuff, for both the batter and the DFS player, happens when the locked, in full control player (HH, aEV etc) starts getting lucky. This is when we start getting the 2 or 3 HR games, the 4 for 5 nights with a HR, 3 doubles, and a deep sac fly—you know, those 35, 45, 50 fantasy point nights that are necessary to win the big money in a GPP.
These tools are also another way to identify pricing errors on players. We know recreational DFS’ers, and a good many regular ones too, avoid “cold batters” like the plague no matter what their from and discipline at the plate looks like, or how BABIP is effecting the results. Since it is believed that DK includes player ownership (which is really public perception of a batter’s hold or cold production) as well as past performance, the DFS pwner armed with a true model of the batter’s performance at the plate no matter the offensive outcome can spot players who are priced too high or too low. But perhaps the greatest asset created by using this data is the ability to be ahead of the general public in identifying when periods of poor production are most likely to end, or in other words…when a cold player is about to get hot, or when a producing batter has a good chance of exploding.
This use of our model is similar to one of the key parts of edge betting at the race track. In addition to knowing how to correctly read the Daily Racing Form, and finding value in the pari-mutuel odds after takeout and breakage, any serious, edge horse better needs to be aware of the current form of a thoroughbred racing horse . Horses are known to be ‘coming into form’ if they are getting back to their ‘true form’. True form means the horse is running at its full capacity in every race. Much like a batter, a winning horse may need several races to get ‘hot’; perhaps they are coming off a layoff, or an injury, or are upset after being moved a long distance from track to track. Likewise a horse can go ‘out of form’ for periods of time, when they run far worse than their ability, usually after a long winning streak or a period of hard, close races. In the past, edge betters had to either go to the track every day (or pay an old-timer or rail bird a small fee) to watch the horse’s practice trials and races to see if the horse was coming into or falling out of true form. The horse’s current form was often the last factor considered before making a bet because form was not fully detectable just by looking at the numbers. In other words, the horse may have been in better form (or performed better during the race) than the outcome of the race suggests. Horse bettors needed to see how the horse actually ran during a past race to truly understand the past performance data. Nowadays, a smart horse player can watch past races on their computer, just like daily Statcast data gives us the ability to recreate a model of a batter at the plate in past games.
It makes sense to think the same way when putting your money on a player in dfs. The traditional stats and the box score do not tell the entire story of the value of the player. Only the numbers and a sense of the player’s current form combined gives you the whole picture. We all know that watching a player on game day gives us a better sense of their performance than just the numbers. Like the edge horse better, the edge dfs player can get a sense of the state of a batter’s true form—positive or negative—no matter what numbers are in the box score. The owner aware of the need to evaluate form will have an edge over the owner only aware of the box scores.
The point of the story isn’t to treat Corey Seager and other batters like horses, but rather to remember that all of us, including baseball players and thoroughbred horses, have our moments when we are in our best form and it seems like nothing in the world can stop us. We all also experience coming out of form, times when we just don’t feel like ourselves, as nothing seems to go right. DFS experts, correctly, always tell us to take a player’s BABIP into consideration when evaluating the performance and numbers of a player, and to look for spots for regression (which may be the most used word in MLB DFS). However, we also need to remember to check a player’s current form. As the first post in the HR Concordance series argues, the aim of advanced stats and game/batted ball/ heat map type data is to find correlations and construct the most accurate models of players possible; in other words the form of the model should match the form of the real player, and the information derived from aEV, aEV+ and other Statcast and plate discipline data is central to this task. The reward for all this work, when combined with our normal, solid research method: more multi-hr monster days on our GPP rosters, and a much better sense of the true value of the MLB players than others.