2016-17 WCHA Week 17 BELOW

2016-17 Week 17 BELOW

TeamBELOWExpWin%Pts/GP %Pts (Max)
7-MTU173980%73%48 (66)
3-BSU164269%79%52 (70)
8-MSU162367%59%39 (57)
4-BGSU153355%53%38 (50)
5-FSU151252%44%29 (47)
2-UAF147747%45%27 (51)
6-LSSU146044%44%29 (47)
9-NMU142640%32%19 (43)
0-UAH135831%36%24 (42)
1-UAA135030%32%19 (43)
Above are the 2016-17 BELOW rankings after Week 12 of WCHA conference play (108 of 140 games played).

Mavericks didn’t throw away their shot

While I’m sure that Minnesota State would have loved a series sweep of Lake Superior, picking up five points on a weekend when Bowling Green split with Ferris State and moving the Mavs into 3rd in the standings table is a good weekend.

MSU has a long trip to the 49th to play Alaska-Anchorage this weekend.  The teams haven’t played this weekend, and we know that the Seawolves have been a tough out even though they’ve usually fallen this year.  Mankato presumably has a bit of an advantage with a short trip to MSP before flying northwest.

They resorted to eating their horses

Alabama-Huntsville showed some signs of life this weekend, but it just wasn’t enough to beat Michigan Tech, whom they’ve never defeated.  A decent third period and a mostly-good regulation left the Chargers with just one point.  Not only were Bemidji State fans in the weird position of rooting for a long-standing rival, Charger fans were looking to keep pace with the FSU/LSSU/UAF crowd.

BG, do you yield?  Yes, they yield

The aforementioned split in Big Rapids had a pretty big effect on the standings and on BELOW.  Not only did the result allow MSU into 3rd, the results let Ferris into a tie for 5th with LSSU.  FSU won the first two games of the season series against the Lakers, so they do have the hammer at A coming down the stretch.

History has its eyes on them

You do not need fancy math to tell you that Bemidji State – Michigan Tech is huge this week.  The teams split in Houghton in a series that told everyone that the Beavers weren’t just going to run away with this one.

You’ll be back, like before

The other two series on the weekend finish out four-game series.  Huntsville travels to Alaska, and BG host Northern Michigan.  The Wildcats are playing for their playoff lives; the Chargers and Nanooks are playing to avoid Bemidji or Tech as much as they are staying in.

They fought the fight and won the war

Mankato and Tech both won their (unlikely) A tiebreaker scenarios with Lake State and Huntsville.  BG and Ferris are tied at A, and that’s not an unreasonable scenario given that the Bulldogs have two games in hand.

2017 WCHA Playoffs Update, 2017-01-28

2016-17 Week 17 BELOW (In Progress)

TeamBELOWExpWin%Pts/GP %Pts (Max)
7-MTU173880%73%46 (67)
3-BSU164269%79%52 (70)
8-MSU1617n66%59%37 (58)
4-BGSU157160%55%38 (53)
2-UAF147747%45%27 (51)
5-FSU147446%41%26 (47)
6-LSSU146645%44%28 (49)
9-NMU142640%32%19 (43)
0-UAH135931%37%23 (44)
1-UAA135030%32%19 (43)
Above are the 2016-17 BELOW rankings after Week 12 of WCHA conference play (105 of 140 games played).

Friday night’s results:

  • Michigan Tech (5-2, including an  ENG) defeated Alabama-Huntsville.  Tech moves to BELOW 1738, Huntsville drops to 1359.
  • Bowling Green (3-2, OT) defeated Ferris State.
  • Minnesota State (4-2) defeated Lake Superior.

This firms up the playoff picture a bit:

  1. Bemidji State has clinched home ice in the playoffs with the Lake Superior loss.  If Ferris State takes any points from the Falcons tonight, the Beavers will have third clinched.
  2. Michigan Tech has clinched a playoff berth, as none of Alabama-Huntsville, Alaska-Anchorage nor Northern Michigan can pace the Huskies.  The Huskies can finish no worse than 7th, and if the results tonight are like last night’s, the Huskies will finish no worse than fifth.
  3. Bowling Green leads Ferris State 2-0-0-1 in the A tiebreaker between the two teams.  If the Falcons pick up any points tonight, they win at A in a tie with the Bulldogs.
  4. The same is true of Minnesota State and Lake Superior — if the Mavs pick up any points, they win any (improbable) tiebreakers.
  5. Michigan Tech has won the A tiebreaker over Alabama-Huntsville, though this scenario has no relevance.

Some folks have asked for Saturday predictions, but I ran into script-writing problems.  Maybe for next week.

 

ABOVE 2.0b: Randomizing for Awesome

Look at the poor geek, crying in his Red Bull.

Lately, I’ve had a bit of a crisis of confidence.  Week 15 of the 2016-17 WCHA slate was pretty crazy, as a league that hadn’t often gone to overtime saw seven matches in eight games go past regulation time.  I trust the underlying principles of what I’m doing, but even I had some concerns.

One of the things that I did was look at what I call BELOW 2016 — the concept that I’ve been using — versus BELOW 1500, where everyone started at an estimated BELOW of 1500, with adjustments in game play from there.  The third table in the above link shows that there’s not a huge difference, but it exists.

Then I got to wondering about whether the three-goal multiplier was worth the hassle and held any predictive value.  I decided that it probably did.

It’s all really just an estimate!

But both of these got me to thinking more about what BELOW is: it’s an estimate distilled into a rating.  Is there anything special about that one specific number?  What’s the error band on that estimate?

Then I realized what I needed to do: I needed to randomize the range of BELOW.

What?

Think about it like this.  BGSU’s current BELOW rating is 1559, or a team that wins about 58% of the time against any average WCHA team.  When you look at the Falcons’ point percentage, that seems pretty close to the truth: they’ve gotten 53% of the possible points for the games that they’ve played.

But what if that points percentage were right?  We can infer from the table at the bottom of the BELOW explainer that BG could be more like a 1520 or so.  That’s a pretty big swing!  Even those we know that BG has played a schedule that averages at 1516, not 1500.  So the truth is probably somewhere between 1520 and 1559.

Now, I could divine this for everyone, or I could take advantage of the Monte Carlo simulation approach and make things random.

Count on Monte Carlo

If you aren’t familiar with the process, here is a very short explanation:

  1. ABOVE starts with two BELOW estimates, one for each team.
  2. ABOVE calculates an expected value from the Elo family of functions.
  3. Based on that expected value, one team is declared the lesser team and the other the greater, with the break-even point at the EV of the lower team.
  4. Historical data of the number of shootouts, 3v3 overtimes, 5v5 overtimes, one-goal wins, and two-goal wins create bands centered around that expected value, which I consider the basis. Overtime games seem to be generally a toss-up, so the real action is in regulation.
  5. Because of the skew of the basis away from a 50-50 matchup, the stronger team is far more likely to get to a 2+ goal victory.  This is consistent with the data.  If it weren’t, the system would correct and revert teams to the mean.
  6. ABOVE does a dice roll using the Python random.random() function, which generates a pseudo-random floating-point number between 0.0 and 1.0.  This number is checked against those bands, and whenever it falls into a slot, that’s the winner.
  7. Do this lots of times.  For the series-by-series predictions, I run them 1,000,000 times.

So here’s why I’m going to randomize: if the BELOW estimate is wrong for some reason — randomizing will help even out the estimate, just as randomizing the potential results will get you a good estimate.

How Randomizing BELOW Inputs Works

Let’s look at Bowling Green again.  Their BELOW is 1559.  If we say that the error band of BELOW is, oh, 40 points, then ABOVE’s random-integer function could produce a revised BELOW estimate between 1519 and 1599 — a range of being a 53% team and a 64% team.

So?  Bowling Green is playing Ferris State, who has a BELOW estimate of 1486.  That means that their randomized BELOW range could be 1446 to 1526.  Yes, you will very likely come up with a handful of scenarios where the random generator has Ferris State as a favorite!

I will not be releasing revised estimates of the series-by-series predictions prior to the drop of the puck on Friday — what’s out there is what’s happening.  But this could be a game-changer.

 

2016-17 WCHA Week 17, Series 54 Prediction: Lake Superior (1483) at Minnesota State (1600)

2016-17 Week 17, Series 54: Lake Superior (1483) at Minnesota State (1600)

ResultProbability
MSU sweeps36.50%
Splits27.72%
LSSU Sweeps8.72%
LSSU 5 points (W, OTW)6.26%
MSU 4 points (W, OTL or OTW, OTW)6.04%
MSU 5 points (W, OTW)3.82%
LSSU 4 points (W, OTL or OTW, OTW)2.90%

The teams split in the Soo earlier this season at a time when BELOW was very impressed with Lake Superior — a BELOW rating of 1561 after a blow out win on Friday night was their season peak.  Since then, it’s been a tough slog for the Lakers, who who’ve picked up half of their standard wins against Alabama-Huntsville.

Minnesota State has largely been able to keep its scoring up this season, and it’s lights-out at home, just like always. Gordon Defiel’s squad will be without captain Gus Correale and third-leading scorer Diego Cuglietta, who participated in an extended fracas after the end of Saturday’s game against the Chargers.  This has the hallmarks of a home sweep for the Mavericks.

2016-17 WCHA Week 17, Series 53 Prediction: Bowling Green (1559) at Ferris State (1486)

2016-17 Week 17, Series 53: Bowling Green (1559) at Ferris State (1486)

ResultProbability
BGSU sweeps30.83%
Splits29.89%
FSU Sweeps12.20%
BGSU 4 points (W, OTL or OTW, OTW)7.20%
FSU 5 points (W, OTW)6.53%
BGSU 5 points (W, OTW)3.73%
FSU 4 points (W, OTL or OTW, OTW)2.74%

The teams split a pair in Ohio back on the first weekend in November before Bowling Green had their act together.  The spread of probabilities in the above table feels right to me.  It would make sense for BG to come out well ahead this weekend, but then they’ve played .530 hockey in league play this year.  The Falcons are better on the road this season.  I’d take a split, but I know that no one wearing orange wants that.

2016-17 Week 17, Series 52 Prediction: Michigan Tech (1733) at Alabama-Huntsville (1364)

2016-17 Week 17, Series 52: Michigan Tech (1733) at Alabama-Huntsville (1364)

ResultProbability
MTU sweeps67.67%
MTU 4 points (W, OTL or OTW, OTW)11.57%
Splits5.64%
MTU 5 points (W, OTW)5.63%
UAH 4 points (W, OTL or OTW, OTW)0.76%
UAH Sweeps0.09%
UAH 5 points (W, OTW)0.08%

Only one 2016-17 WCHA series has had a wider starting BELOW differential than Tech-Huntsville (369): Bemidji State at Alaska-Anchorage (387).  As you’ll remember, the Seawolves won that first game by two goals.  In fact, UAA has been involved in all five of the 300+ point differential games this season, and they won three of them, two against MTU.

That said, UAH has been dreadful at home this season (1-0-1-8 in league play), and they’ve lost six straight games.  As an alumnus, I can tell you that the natives are getting restless.  The Chargers desperately need a win this weekend just to keep out of the WCHA cellar, and as you can see, ABOVE expects the worst.

If you’re curious, ABOVE expects two blowouts 13.99% of the time, all% by Tech.  The Huskies can only pick up 12 BELOW points this week because of the wide differential.  The teams would swap 60 points if the nearly-impossible happens and the Chargers sweep.

Is a Three-Goal Multiplier Worth It?

Current ABOVE Goal Margin Multipliers

When I calculate BELOW, I use the following goal-margin point multipliers:

  • Win in a shootout: 10 points
  • Win in 3v3 overtime: 20 points
  • Win in 5v5 overtime: 30 points
  • Win by one goal in regulation: 40 points
  • Win by two goals: 50 points
  • Win by three goals: 60 points

For the regulation wins, I deduct any empty-net goals from the margin, including Friday night’s 5-2 Ferris State win over Bemidji State.  This keeps things fairer in my mind.

I think that it should be obvious that any game going past 60:00 minutes should have fewer BELOW points awarded than a game decided in regulation — if you have to go longer, you’re not demonstrably better than the other team.  This is the same argument that I made in previous seasons about overtime games: winning in overtime got you 0.75 in actual value, not 1.0.

A review of the 2016-17 outcomes to-date

Now that we’re 73% of the way into the season, I thought that I’d look at matters.  Here are when games have been decided so far:

  • Shootout: 8
  • 3v3 overtime: 7
  • 5v5 overtime: 8
  • One goal: 30
  • Two goals: 22
  • Three goals: 27

Why did I decide to spend a couple of hours looking at this?  I was running some fake simulations on Sunday night to test some new wrinkles in my by-series script and saw this:

For teams 200 BELOW points apart or more, the weaker team never can get to the two- or three-goal margin.  (You can see that a handful of one-goal wins by the weaker team on the first night gets a handful of two-goal wins.)

Is this borne out by what we see in reality?  After all, BELOW is designed to be an estimate of reality, and if it’s not doing that job, I have to make an adjustment.

The overtime results, in order, by bracket

Let’s consider all 102 results and the differential between them in each of the outcomes.  I’ll highlight any that are interesting.  Negative point differentials mean that the weaker team won.  I’ve ordered these from lowest to highest.

  • Shootouts: -364, -184, -101, -89, -58, +92, +233, +301

There doesn’t seem to be any real order to that set.

  • 3v3 overtime: -335, -224, -89, -62, +20, +53, +257

There’s not much order there, too — it seems like getting to the new, funky overtimes gets us to randomness.  That fits with how ABOVE is currently modeling these event.

  • 5v5 overtime: -89, -38, -38, +56, +69, +70, +172, +239

 

Overall, it looks like the BELOW-favorite is a better pick in overtime, but it’s not by much.  Overtime appears to be as much of a toss-up as ever, which reinforces the concepts that BELOW is an estimate and improbable things do happen.

Mean and Median BELOW through 102 games

Now I want to note that the average differential of BELOW so far this season is 128 points.  The median — as many results lower as higher —is 110.  That’s pretty consistent with the spread that we’ve seen in the standings to-date.  Before we go on to regulation wins, let’s carry on to a simulation of a 110-point margin through ABOVE:

The BELOW-underdog wins a respectable amount of time, but this one is pretty heavily-weighted to the BELOW-favorite.  But what I want to note are the line for Home: they did not win by more than two goals on the first night, and they barely won by 3+ on the second night.  Does this match what we see?

One-goal margins

  • One-goal wins: -237, -154, -143, -141, -134, -119, -111, -84, -59, -47, -46, +4, +9, +21, +22, +35, +38, +38,+108, +113, +114, +143, +187, +233, +237, +272, +297, +354

The one-goal wins are strongly skewed to the stronger teams, and there are 12 games where the margin is smaller than the median, and the stronger team won eight of those.  That seems to validate — with a small sample size — that the model is working.  This is without doing anything like root-mean-square or other statistical fun that I will not bore you with (other than to show results).

Two-goal margins

  • Two-goal wins: -387, -192, -164, -143, -114, -39, -33, +34, +41, +65, +87, +90, +98, +107, +124, +145, +159, +165, +215, +204, +209

Of the 22 two-goal wins, 15 were won by teams favored by BELOW, with eight of those by teams with a differential greater than +100.  This is pretty much what you would expect.  There have been five big upsets, but that’s going to happen.

Three-goal margins

  • Three-goal wins: -118, -98, -48, -23, -20, 0, +4, +17, +46, +64, +66, +71, +80, +89, +98, +103, +116, +130, +148, +156, +172, +193, +194, +251, +265, +286

Only one team — Lake Superior over Minnesota State in Week 6 — with a differential greater than -100 won a three-goal game.  Ferris State (-98) defeated the Mavs in Week 11.  In fact, just five BELOW underdogs won by three or more goals.

In general, these results are what we would expect.  Only that Laker win would be missed by ABOVE, and for me, a 1% improbability is okay in my book.  The improbable is often possible!

Is the 3+ differential worth it?

In our highest BELOW differential-turned-blowout, Northern Michigan at Bemidji State, the Beavers got all of 10 points.  Reducing it would give the Beavers just eight.  Remember: BELOW assigns points based on who wins and what their expected value of winning was.  If NMU had won that game, they would’ve been +50, not -10.

Here’s a table of BELOWs with and without the 3+ differential multiplier:

2016-17 Week 16 BELOW, Revisited Again

TeamBELOWBELOW
7-MTU17331724
3-BSU16421634
8-MSU16001605
4-BGSU15591552
5-FSU14861489
6-LSSU14831474
2-UAF14771483
9-NMU14261423
0-UAH13641370
1-UAA13501366
Above are the 2016-17 BELOW rankings after Week 12 of WCHA conference play (102 of 140 games played). The left column has three-goal multipliers; the right stops at two goals.

In eight of the ten cases, BELOW reverts to the mean a bit, as expected; the outliers are Lake Superior and Minnesota State, who were both involved in the three-goal upsets.  Alaska and Lake Superior flip places.

In short: having the 3+ goal multiplier doesn’t make a huge change, but I’m happy to keep it in there out of momentum if nothing else.  If I do drop it, I’ll probably go to something like 5-10-20-30-40. That recalculates everything and will generally draw teams closer in the ratings, as you can see above.  I will consider this after the season unless there’s some huge reason to make the change.

2016-17 WCHA Week 16 BELOW

2016-17 Week 16 BELOW

TeamBELOWExpWin%Pts/GP %Pts (Max)
7-MTU173379%72%43 (67)
3-BSU164269%79%52 (70)
8-MSU160064%57%34 (58)
4-BGSU155958%53%35 (53)
5-FSU148648%43%26 (50)
6-LSSU148348%47%28 (52)
2-UAF147747%45%27 (51)
9-NMU142640%32%19 (43)
0-UAH136431%38%23 (47)
1-UAA135030%32%19 (43)
Above are the 2016-17 BELOW rankings after Week 12 of WCHA conference play (102 of 140 games played).

Home Ice Almost in Hand

With the weekend’s results, Bemidji State has 52 league points.  Alaska (51), Ferris State (50), Alabama-Huntsville (47), Alaska-Anchorage (43), and Northern Michigan (43) cannot match the Beavers’ total, which means that they can finish no worse than fifth.

Lake Superior can max out at 52 points.  If they’re tied, each team ends up with 16 regulation conference wins, but six Bemidji losses puts them with 10, and there they lose the C tiebreaker (LSSU his nine losses to date).

Bemidji is off this weekend, but all they need is for Minnesota State to pick up one point this weekend at home against Lake Superior.  It would also work for Ferris State to beat Bowling Green once to limit their maximum.

An Update at A

If you’ll remember from late last month, the A tiebreaker is unchanged: play four times and get the better points percentage to win.  I found 25 two-way A tiebreakers this year, and some of them have concluded.  The winners so far are below:

  1. Alabama-Huntsville over Ferris State (3-0-0-1)
  2. Lake Superior over Alabama-Huntsville (4-0-0-0)
  3. Alaska and Bowling Green remain tied (2-0-0-2)
  4. Minnesota State over Alaska (2-0-1-1)
  5. Bemidji State over Alaska-Anchorage (3-0-0-1)
  6. Michigan Tech over Alaska-Anchorage (1-1-2-0)
  7. Northern Michigan over Alaska-Anchorage (3-0-0-1)
  8. Bemidji State over Bowling Green (3-0-0-1)
  9. Bemidji State over Northern Michigan (4-0-0-0)
  10. Lake Superior and Northern Michigan remain tied (1-1-1-1)
  11. Minnesota State over Michigan Tech (2-1-0-1)

All three series this weekend — Michigan Tech at Alabama-Huntsville, Bowling Green at Ferris State, and Lake Superior at Minnesota State — finish out A tiebreakers.  Six more are decided in Weeks 18, 19, and 20, with the league’s rivalry weekend finishing off the final five.  It’ll be two more weeks before we know the result of the high-profile three-way tiebreaker, BSU-MSU-MTU.

A Look Ahead

2016-17 Week 16 BELOW, Revisited

TeamBELOWRBELOWExpWin%Est. Pts.
3-BSU1642156660%63
7-MTU1733148081%63
8-MSU1600147567%50
4-BGSU1559142568%47
2-UAF1477140660%41
6-LSSU1483147851%40
5-FSU1486153144%36
0-UAH1364160320%28
1-UAA1350151228%26
9-NMU1426159228%26
Let's look at what's left

That table is not ABOVE-generated.  It’s pretty simple:

  • RBELOW is the average of the opponents’ BELOW ratings for the rest of the season, as they stand today.
  • ExpWin% here is the expected winning percentage based on that team’s BELOW and their opponents’.
  • Est. Pts. is an extrapolation of the EW%, the number of games left, and the current total.

Bemidji has the lead, but Tech has the easier schedule.

Hopefully I’ll have time this week to knock out a to-completion script.  Working one job while training for another is wearing me out.

2016-17 WCHA Week 16, Series 51 Prediction: Alaska-Anchorage (1412) at Northern Michigan (1364)

2016-17 Week 16, Series 51: Alaska-Anchorage (1616) at Northern Michigan (1697)

ResultProbability
Splits26.42%
UAA sweeps23.34%
NMU Sweeps12.45%
UAA 5 points (W, OTW)10.63%
UAA 4 points (W, OTL or OTW, OTW)9.56%
NMU 4 points (W, OTL or OTW, OTW)8.31%
NMU 5 points (W, OTW)7.55%

Alaska-Anchorage is 1-1-2-4 on the road in the WCHA this season, which is to say that it’s kept is close half of the time.  They should expect to do better against Northern Michigan, who have failed to turn mediocre fundamentals into mediocre results.  Consider:

NMU has:

  • The third-best power play in league play.
  • The fourth-best penalty kill (leaving them -1 net so far).
  • They’re seventh in league scoring (86 GF) and league defense (100 GA).

But also:

  • The three teams behind them in league scoring and defense are all above them (10th, 13 points) in the standings: UAH at (t-6th, 23), LSSU (8th, 22), and UAA (9th, 19).

It has to stick in the Wildcats’ whiskers.  Their road back to the playoffs starts this weekend.

2016-17 Week 16, Series 51 Prediction: Minnesota State (1616) at Michigan Tech (1717)

2016-17 Week 16, Series 50: Minnesota State (1616) at Michigan Tech (1717)

ResultProbability
MTU sweeps30.23%
Splits24.29%
MTU 4 points (W, OTL or OTW, OTW)12.97%
MTU 5 points (W, OTW)12.23%
MSU Sweeps7.66%
MSU 5 points (W, OTW)5.92%
MSU 4 points (W, OTL or OTW, OTW)4.96%

BELOW accounts for momentum in the sense that a great run (see Bemidji State’s climb to the top) gets you a high rating but a drop-off does the same (see Bemidji State’s drop of the past few weeks, or Michigan Tech’s similar drooping).

At the turn of the year, the gap in BELOW between the two was 1732-1589.  It’s quite narrower now.  This prediction feels somewhat Tech heavy, but not too much.  It would be easy to overreact to Tech’s struggles with Alaska-Anchorage and overvalue the Mavericks’ efforts against Alabama-Huntsville and Bowling Green, but I’m not sure that this is all that out of line.