Can this be Done with an Algorithm?

Can daily fantasy baseball be reduced to an algorithm?  I don’t know, but I certainly wouldn’t rule it out.

The Comparison to Stocks

The stock market used to seem random.  To large degree, it is still deemed to be random.  However, there are numerous algorithms that have been invented to handicap the stock market.  Literally thousands of data points are analyzed.  Using the whole history of the stock market, including all securities traded thereon, data is analyzed for patterns.  If history were to predict the future, there’d be a fortune in knowing when and how.

Backtesting

The idea of using past data and it’s patterns

An Algorithm as Daily Fantasy Baseball Strategy

The same concept applies in daily fantasy sports, and particularly in daily fantasy baseball.  What would history reveal?  For instance…

  • If a guy hits a homerun yesterday, are the chance greater or less than standard that he hits one today?
  • If a guy threw a shutout last time out, does he pitch better than his season average this time out?
  • If a guy is hitting better than his season average over his last seven games, what’s the liklihood he hits a homerun tonight?

The examples are numerous.  Last night, I tried something.  I said, I want to find a guy who is hitting better than his season average over his last 7 games, who hit a homerun yesterday and either has a favorable lefty-righty matchup or has hit a homerun and walked off the opposing starter.

Huh?

You have to try something, right?  Turns out, that set of criteria produced a few more guys than one starting lineup.  I used the best salary combination I could find to create a few teams.  I was so proud of my genius algorithm.  I was starting guys I had barely noticed all season.

Probably the worst night of my daily fantasy season.  Nobody hit.  A good night out of a few starters was nullified by a complete lack of offense.  However, I know I’m onto something.

It’s a Huge Amount of Data

I’m not talking about doing something I can do in half an hour just from looking at a few matchup sheets.  I’m talking about the whole history of baseball.  I’m talking about multivariate analysis on every hitter who has ever played the game.  I’m talking about bayesian logic applied to our national pastime.  It’s not crazy.  Ask Billy Beane.

The Biggest Daily Fantasy Baseball Money Leagues are Huge

The amount of money in daily fantasy sports at this point is enormous.  I’m not talking about your fantasy football league that has a $500 season long buy-in with the winner taking home a few grand.  I’m talking about multi-thousand dollar contests running nightly and lasting one day.  And it seems to be growing exponentially every season.  There’s money to be won here.

Maybe It’s Random

Of course, maybe there is no secret to be unlocked.  Maybe in studying history in an insanely deep and statistical way will prove that finding value and consistent performance in daily fantasy baseball cannot be reduced to an algorithm.  Maybe it’s random.  (Not completely random as the good players will be good and the not-so-good will be, well, not-so-good.)  But unlocking the value might be random.

Performance, Salary Value is Secondary

It’s a function of the salary algorithm to some degree too.  Fanduel will create salaries differently than Draftday, for example.  The value is obviously a function of cost.  However, despite the differences between the sites, I don’t think it matters.  The key is finding hitters statistically likely to perform, then constructing lineups around that information with salary as a secondary factor.

Conclusion

You need the data set.  In essence, every box score in history.  Then you need insane mathematical chops.  I mean Ramanujan mathematical chops.  Or at least the ability to program a computer to tap into that kind of data analysis.  But my hypothesis is you find data useful as daily fantasy baseball predictive information.  I also predict it won’t be too long before it’s completely worth it.  These fantasy games are getting huge.  The money at stake is insane.

May 9, 2014

DFB Picks for Friday evening

– also see today’s earlier post – Hot Streaks or Pitcher History?

Pitching

There are quite a few quality options on the hill tonight, including Wacha, Verlander, Tanaka, and Darvish.  There are even more in value names like Teheran, Gallardo, Hammel, Cueto and more.  Ultimately, however, I’m going with the top dog.

  • Jose Fernandez– The guy is white hot.  He’s given up 30 hits in 46 innings with 65 strikeouts (65!).  That means he’s pacing well over 300 strikeouts.  Even when he’s bad he strikes out eight guys.  Wait, he’s never bad. … One day, he won’t have his best stuff and he’ll give up 4 earned in 5 innings.  I’m thinking today is not that day.  He’s pitching in San Diego, against a team that has been shut out three times in the last nine games.  (They’ve scored a grand total of 17 runs in those 9 games.)

Hitting

  • Eric Hosmer  – He started slowly, only hitting his first homerun in his last series at San Diego.  However, last year I think he hadn’t hit a single homerun until post May and he finished with 17.  He’s still a young guy, despite being an anchor in the middle of that KC lineup.  He’s 11 for his last 30 with 4 xbh and 8 rbi.  Strong.  He faces Seattle’s Maurer tonight, with his 6.92 ERA.

Lineup Stack

  • Mariners or Royals– I’m envisioning a slugfest tonight in the Great Northwest!

Value

  • George Springer – If you ever read these picks, you know I love this guy.  Fanduel does not love this guy, though he has moved all the way up from 2200 to 2300 in salary on this Friday evening.  He’s facing the Orioles’s Chen.  The lefty Chen has not been spectacular.  Springer hit his first homerun on Wednesday.  The projected line I saw for him is 23 hr’s, 26 steals.  In five months.  Wah, whhaaattt?  I think he will put his slow start behind him.
  • Mike Zunino – This guy is starting to live up to billing in Seattle.  He’s 24-97 with five homers and 16 rbi.  He has two hr’s in his last 20 ab’s.

Pick of the Day

  • Hosmer– I’m searching for dingers.  Many dingers.  Mucho dingers.

Strategy Tip: What Have You Done for me Lately?

In three years of playing daily fantasy baseball, I’m ready to declare something.  And I must provide the caveat: this is based on feel, not empirical evidence.  I generally tried to provide daily fantasy baseball strategy based on statistics, not just feel.  However, in this case, it’s feel not data.

You will benefit more from hot streak hitters than you will from history versus a particular pitcher.

Spend the bulk of your time studying who’s been hot lately.  It’s amazing how long a hot streak can last.  With a handful of guys, it might last damn near the whole year.  Don’t think a guy can’t keep doing what he’s doing.  Trust recent history.  If your guy is hot, go with it!

You can study the match-ups all day long, but, in the end, if your guy is 0 for his last 14, stay away.  A guy might hit .500 off a particular pitcher, but if he’s in a slump, it won’t matter.  Maybe the last two times your guy faced that pitcher he was hot and that’s why the numbers are so good.  Your guy facing the pitcher at issue tonight is how reversion to the mean happens.

I’m not suggesting batter-pitcher history is irrelevant.  Quite the contrary.  It’s probably second most important behind how the hitter has been performing lately.  Recent performance, however, is number one!  Ie the title of the post – What have you done for me lately?

Most hitters are neither hot nor cold.  Most hitters are run of the mill.  If they’re a .270 hitter, they are 4 for their last 15 or something similar.  If they’re a .320 hitter with power, they are 6 for their last 17 with a homer.  Neither hot nor cold.  With these neither hot nor cold hitters, history becomes crucially important.

But faced with recent history or distant history, I’m going with recent, even if it means ignoring the matchups.