2018 AL Wild Card

Now that the Cubs are done, let’s handicap the AL Wild Card.

Playoff Horse Race

TeamID W-L Total Hitters Pitchers Starters Relief
NYA 38 43.7 28.2 15.5 4.5 11.1
OAK 32 37.9 21.1 16.7 1.7 15.1

The above view shows a breakdown of team value based upon the 25 man roster for this game.  Since the value measure generated by this model (WAA) has additive properties, the value of a group of players is the sum of player value included in that group.  The sum of an entire team of players should equal their real WAA or W-L number.  The W-L column represents a seasonal value of real wins minus real losses.  The Total WAA represents the value of the current roster which is usually higher.

The above shows OAK has a relief advantage and NYA has a hitting advantage.  Normally team status records would be shown but that just represents seasonal numbers.  NYA is slightly underwater with unearned runs at -4.5 and OAK is almost exactly league average.

Ouija Board

The Ouija Board represents the current betting market from the latest snapshot taken every two hours.  There are many outlets setting lines that may differ and this exercise isn’t about arbitraging that.  The snapshot is just a single random sample to get an idea as to where the market sets an expected break even probability.  it is called a Ouija Board because, like how a Ouija board influences a group of people to text a message, the betting market consists of thousands of strangers independently influence setting a probability that is extremely accurate most of the time.  Sometimes it can meander off course and if you can detect that accurately you can gain an advantage on the house much like counting cards in blackjack.

The break even probabilities in bold on the LINEAWAY and LINEHOME records is what the probability of that team winning where if you made that bet an infinite number of times you would neither make or lose money (i.e. a total waste of time).  If you think the probability of your team winning is greater than that then that’s a betting opportunity.

Tier Combo is a Monte Carlo simulation finding similar lineup/starter and lineup/relief combos from 100,000 games dating back to 1970.  Modern baseball relies far more heavily on relief pitching than pre-1970 or even pre-1980.  Relief pitchers represent almost 1/3 of a baseball roster.

DATE 10_03_8:00_PM OAK NYA

LINEAWAY OAK [ 0.364 ] < 0.385 > +160 $260
STARTAWAY -0.02(0.496) Liam_Hendriks_OAK TIER 3
LINEHOME NYA [ 0.649 ] < 0.636 > -175 $157
STARTHOME 3.30(0.578) Luis_Severino_NYA TIER 1


OAK Lineup 1 ==> NYA Starter 1 / Relief 3 == 0.470 OAK 4.49 runs
NYA Lineup 1 ==> OAK Starter 3 / Relief 1 == 0.530 NYA 4.77 runs

OAK    122 119
NYA    83 85

On the LINEAWAY and LINEHOME records the probability in [] is the starting line and in <> is the current line.  This line hasn’t moved much at all and to bet the Yankees you would need them to win much greater than 64% of the time if they played this game an infinite number of times.  They only play this game once however.

The second to last number on those records is the official line nomenclature which can be confusing.  To bet the Yankees you have to risk 175 to win 275 ( the last number, 100 profit ) which includes your bet.  For the OAK line you only risk 100 to win 260 ( 160 profit ).

Expected Value show how much a 100 wager is worth based upon TC simulation results.  Since TC simulation thinks OAK has a 0.470 break even probability and LINEAWAY has OAK break even probability at 0.385,  the OAK line has an expected value of 122 on a 100 wager.  This means you will make money in the long term — but — but — only if the TC simulation result is accurate.

We have a dataset which include 7 years of daily lines for every MLB game, and 90 years of daily baseball snapshots.   In the off season the last 7 seasons will be simulated and this forecasting will be either proven to work or proven not to work.

The above shows OAK to be a betting opportunity at +160.   A single game does not prove or disprove any of this.  There is a slight problem with roster data (again) which caused NYA relief to be undervalued in simulation.  This should push the above EV below 120.  The dWAA column is for differences in win/loss records which matter in regular season games but not so much in playoff games.

As soon as current lineups come out this post will be updated.  Until then ….

Update: Lineups are posted.

OAK Lineup

WAA Name_TeamID Pos PA 10032018
0.50 Nick_Martini_OAK LF 179
3.09 Matt_Chapman_OAK 3B 616
2.52 Jed_Lowrie_OAK 2B-3B 680
7.75 Khris_Davis_OAK DH-LF 654
2.14 Matt_Olson_OAK 1B 660
3.13 Stephen_Piscotty_OAK RF 605
0.67 Ramon_Laureano_OAK CF 176
0.08 Marcus_Semien_OAK SS 703
-1.07 Jonathan_Lucroy_OAK CR 454
Total WAA=18.82 PA=4727 WinPct=0.576

This is an extremely good lineup but not as good as the one below.

NYA Lineup

WAA Name_TeamID Pos PA 10032018
-0.88 Andrew_McCutchen_NYA RF-LF 682
3.38 Aaron_Judge_NYA RF-DH 498
4.05 Aaron_Hicks_NYA CF 581
4.60 Giancarlo_Stanton_NYA DH-RF-LF 705
3.13 Luke_Voit_NYA 1B 161
4.96 Didi_Gregorius_NYA SS 569
4.09 Miguel_Andujar_NYA 3B-DH 606
2.10 Gary_Sanchez_NYA CR-DH 374
2.35 Gleyber_Torres_NYA 2B-SS 484
Total WAA=27.78 PA=4660 WinPct=0.614

That might be the best lineup of any team this season.  Pitchers on both teams are going to have their hands full tonight.