Sunday, September 29, 2019

An Alternative Universe in Which the Mets Made the Playoffs

Early last week, I wrote about whether the Nationals should rest Max Scherzer or maximize their chances of retaining home field advantage in the NL Wild Card game. Well, joke is on me, because they easily did both.

With that question centering around the Nationals' top tier pitching options, my esteemed colleague (a suffering Mets fan) posed this: would the Mets be the rare underdog that actually would have a better chance against the Nationals in a 5-game series, as opposed to a one game playoff?

Generally, an underdog wants a shorter time frame to try to pull off the upset - less time equals more randomness, and less of a chance for talent to win out. But the Mets are one of the few teams in baseball that could match the Nationals' pitching staff, with three aces in the top 10 of NL pitcher WAR (plus the NL Rookie of the Year, and still didn't make the playoffs).

So consider an alternative reality in which the Mets make the playoffs, and face the following proposition: they have to face the Nationals in a one game playoff, or a full five game NLDS (in both cases Washington has home field advantage).

Do you want to roll the dice on one game, with your best pitcher (Jacob DeGrom) up against their's (Scherzer)? Or would you rather have your lineup of DeGrom (#1 in NL pitcher WAR), Wheeler (#7), and Syndergaard (#10) go up against Scherzer (#2), Strasburg (#3), and Corbin (#6)?

Over 10,000 simulations, the Nationals win 61.8% of the time in the one game playoff (this also serves as Game 1 of the NLDS). Here are the game-by-game probabilities over a five game series:


GameHomeAwayHome PitcherAway PitcherWSN Win ProbNYM Win Prob
Game 1WSNNYMScherzerDeGrom61.80%38.20%
Game 2WSNNYMStrasburgWheeler69.99%30.01%
Game 3NYMWSNSyndergaardCorbin41.45%58.55%
Game 4NYMWSNDeGromScherzer39.43%60.57%
Game 5WSNNYMStrasburgWheeler69.12%30.88%
SeriesWSNNYM62.30%37.70%

It's actually incredibly close - Washington would only have a 0.5% edge (62.3% favorites) in the NLDS scenario compared with the wild card game.


OutcomeIn...Prob
WSN317.93%
WSN417.39%
WSN526.98%
NYM36.7%
NYM418.9%
NYM512.1%

But, as expected, the Mets would still (barely) prefer the randomness inherent in one game over a five game series.

Thursday, September 26, 2019

Should the Washington Nationals Rest Scherzer, Risking Home Field Advantage?

The Washington Nationals have clinched a wild card spot, but are only one game up on the Milwaukee Brewers for home field advantage in the NL Wild Card game. The Brewers have been red hot, with a 19-4 record in the month of September. 

Meanwhile, the Nationals have one of the best rotations in baseball, with Max Scherzer and Stephen Strasburg ranking #2 and #3 in the NL, respectively, in pitcher WAR this season.

Scherzer is their best pitcher. And he's currently in line to start the last game of the regular season on Sunday. However, that would burn him for the one-game playoff next Tuesday. So which is more valuable? Scherzer pitching in the NL Wild Card game, or the Nationals retaining home field advantage?

My colleague answered this question via FiveThirtyEight's pitcher ratings:
Using 538's Elo rating, they have Scherzer (30) as worth 11 more points than Strasburg (19). However, they have HFA listed at 24 points, which is actually doubled to 48 points if you take the HFA away from WAS and give it to MIL. So Strasburg at home is worth 43 points and Scherzer on the road is worth 6 points, a difference of 37 points! So they theoretically should keep Scherzer on normal rest this week as long as HFA is still on the line.
Per the Elo formula and FiveThirtyEight's ratings, the Nationals would be 54.4% favorites on the road with Scherzer, and 59.7% favorites at home with Strasburg. So they should maximize their chances at retaining home field.

I checked this using my play-by-play MLB simulator, assuming that the Brewers' top arm, Brandon Woodruff, would start in either wild card scenario. I found similar results: the Nationals win 57.5% of the time on the road with Scherzer, and 59.9% of the time at home with Strasburg. Both pitchers are really good! So home field is the deciding factor.

And for the Brewers to maximize their chances of winning, they have to chase either pitcher early: Scherzer's average start in Milwaukee is 6.08 innings in a Brewers win, but 6.49 innings in a Nats win. Strasburg's average start in Washington is 5.79 innings in a Brewers win, but 6.25 innings in a Nats win.

Sunday, September 22, 2019

"What are the odds?" Of Rolling 7 Doubles in a Row (Dice)

In backgammon, each game begins with each player throwing a single die to determine who goes first. In the event that each player's roll is the same, "then both players roll again until they roll different numbers". The following question was posed to me: what are the odds that each player rolls the same number... 7 times in a row?

There are 6 potential numbers on each die, so over two dice there are 6 * 6 = 36 possible combinations. Additionally, there are 6 possible pairs: (1,1), (2,2), (3,3), (4,4), (5,5), (6,6). So 6 / 36 = 1 / 6 ~= 0.167% chance of tying on one roll. But when this is replicated 7 times in a row: (1 / 6) ^ 7 = 0.00036%, or 1 in 279,936.


NChances1 in ...
116.67%6
22.78%36
30.46%216
40.08%1,296
50.01%7,776
60.002%46,656
70.00036%279,936
80.0001%1,679,616
90.00001%10,077,696
100.000002%60,466,176


What if you play backgammon a lot though? How long would you have to play in order to expect to see a streak of 7 dice ties in a row?

The math is explained on this nice "Probability of Runs" calculator, but you would need to play 232,845 games of backgammon until there would be a better than 50/50 chance of seeing a run of ties like this. The average number of games needed until you would see this happen is even larger: 335,922 games of backgammon.


Wednesday, August 28, 2019

Preseason NCAAF Rankings for 2019

As I did in 2018, 2017, 2016, 2015, and 2014, I've compiled my preseason NCAAF rankings that take into account player turnover and recruiting classes.

There's a slightly new wrinkle this year though. As before, I used my final MDS Model ratings from last season as the base, and then factored in ESPN's Preseason FPI and the S&P+ projections (which are now housed over at ESPN with Bill Connelly leaving SB Nation), but this year I've also included Ed Feng's rating system from The Power Rank. Once the season starts, this "preseason" rating will be faded out as the season progresses, carrying less and less weight with each ensuing week.

As I have in the past, I missed Week 0, but got these ratings out before the first full weekend of games.

In the below list, the "Trend" indicates whether the respective team's new ranking rose or fell relative to last year's preseason ratings. No team movement in or out of FBS this year though!

My goodness are Alabama and Clemson evenly matched, and in a class of their own:

Alabama: 0.95035035751657
Clemson: 0.949045941382027


RankTeamPRESEASONTrend
1Alabama0.950UP
2Clemson0.949UP
3Georgia0.851UP
4Oklahoma0.813UP
5LSU0.808UP
6Ohio State0.806DOWN
7Michigan0.793UP
8Notre Dame0.792DOWN
9Auburn0.784DOWN
10Penn State0.743DOWN
11Florida0.740UP
12Washington0.732DOWN
13Texas A&M0.731UP
14Mississippi State0.723DOWN
15Michigan State0.712DOWN
16Oregon0.695UP
17Wisconsin0.691DOWN
18Missouri0.690UP
19Miami (FL)0.669DOWN
20Utah0.667UP
21South Carolina0.660UP
22Florida State0.654DOWN
23Oklahoma State0.653DOWN
24Iowa0.645UP
25USC0.644DOWN
26Texas0.636DOWN
27Stanford0.635DOWN
28UCF0.630UP
29Virginia Tech0.629DOWN
30Tennessee0.627UP
31Iowa State0.627UP
32TCU0.625DOWN
33Boise State0.622DOWN
34Washington State0.622UP
35Baylor0.610UP
36UCLA0.605UP
37Minnesota0.603UP
38Ole Miss0.602DOWN
39Kentucky0.593UP
40Nebraska0.585UP
41North Carolina State0.583DOWN
42Syracuse0.581UP
43Memphis0.577UP
44Arizona State0.575UP
45West Virginia0.571DOWN
46Virginia0.570UP
47Northwestern0.569DOWN
48Texas Tech0.568DOWN
49Purdue0.568UP
50Duke0.566DOWN
51Pittsburgh0.563UP
52Appalachian State0.554UP
53Brigham Young0.552UP
54Indiana0.547UP
55Arizona0.546DOWN
56North Carolina0.544DOWN
57Vanderbilt0.536UP
58Cincinnati0.535UP
59Wake Forest0.532DOWN
60California0.529DOWN
61Arkansas0.524DOWN
62Kansas State0.523DOWN
63Louisville0.515DOWN
64Boston College0.514DOWN
65Fresno State0.508UP
66Houston0.503DOWN
67Temple0.495UP
68Georgia Tech0.495DOWN
69Utah State0.495UP
70Maryland0.490UP
71San Diego State0.489DOWN
72South Florida0.479DOWN
73Marshall0.477DOWN
74Northern Illinois0.469UP
75Toledo0.468DOWN
76Colorado0.464UP
77Western Michigan0.459UP
78Florida Atlantic0.453DOWN
79Army0.452UP
80Ohio0.449DOWN
81Southern Miss0.447UP
82Arkansas State0.440DOWN
83Georgia Southern0.438UP
84Air Force0.437UP
85Southern Methodist0.425DOWN
86Troy0.425UP
87Illinois0.421UP
88North Texas0.417UP
89Louisiana Tech0.412DOWN
90Wyoming0.403DOWN
91Tulane0.400UP
92Middle Tennessee0.397DOWN
93Florida International0.394UP
94Tulsa0.387UP
95Oregon State0.380UP
96Miami (OH)0.380DOWN
97Nevada0.379UP
98Rutgers0.374DOWN
99Western Kentucky0.373UP
100Hawaii0.373UP
101UAB0.363UP
102Eastern Michigan0.360UP
103Navy0.349DOWN
104Buffalo0.348DOWN
105Louisiana-Monroe0.345UP
106Kansas0.341DOWN
107Colorado State0.339DOWN
108Louisiana-Lafayette0.336UP
109UNLV0.325DOWN
110Ball State0.324UP
111East Carolina0.309UP
112Texas State0.307UP
113Liberty0.285UP
114Georgia State0.276UP
115Coastal Carolina0.274UP
116Central Michigan0.274DOWN
117San Jose State0.269UP
118New Mexico0.260DOWN
119Akron0.255DOWN
120Kent State0.250UP
121Bowling Green0.244DOWN
122Charlotte0.234UP
123New Mexico State0.234DOWN
124UTSA0.229DOWN
125Massachusetts0.218DOWN
126South Alabama0.213DOWN
127Old Dominion0.206DOWN
128Connecticut0.201DOWN
129Rice0.162DOWN
130UTEP0.118DOWN