Trades

There's a lot of coverage all over the place. I mentioned this before, but I have a trade page, that's not as updated as many of the big websites, but it has some useful features, such as grouping trades by team and presenting a few of their season stats on one page. It's a bit of a mess, but it gets the job done.

Canucks 2-1 OTL

So I didn't listen to the whole game and certainly didn't watch, but I found it strange that the Canuck's had 7 minor calls against, while Dallas had one (in the first minute). This meant 7 penalties against the Canucks in a row including a long 5-3 and penalty in overtime. So I was curious what is the worst ratio in the NHL for penalties against each team. The answer is 1:0, happened three times [1 2 3]. However these games had very few calls in total (worst was 4). The game this afternoon had a penalty ratio of 7:1. So a larger picture seemed necessary. I plotted home penalties vs. away penalties [only looking at minors - graph on the right]. Each dot is labeled with the gameid that the NHL uses; you can find the game summary by changing the # in this URL: http://www.nhl.com/scores/htmlreports/20062007/GS020936.HTM. The usual relationship can be seen: as away team gets more penalties the home team does as well. I can't say whether the officiating was fair or not in the Canucks game, but I can say it's a little strange. What's maybe most remarkable is that the Canucks were able to come away with a point.

Betting Rankings

N name BET 1 BUF 59.2% 2 OTT 57.4% 3 N.J 55.8% 4 ATL 54.5% 5 T.B 52.9% 6 CAR 52.4% 7 MTL 52.2% 8 PIT 51.0% 9 NYR 50.2% 10 TOR 50.0% 11 WSH 48.9% 12 NYI 46.4% 13 BOS 45.9% 14 FLA 44.6% 15 PHI 39.8%

N name BET 1 DET 60.3% 2 ANA 59.1% 3 NSH 58.5% 4 S.J 58.0% 5 DAL 55.4% 6 VAN 55.2% 7 CGY 54.9% 8 COL 52.2% 9 MIN 51.7% 10 EDM 50.3% 11 PHX 45.7% 12 STL 43.4% 13 CHI 42.9% 14 CBJ 42.1% 15 L.A 41.8%

Now these change a lot quicker than normal standings would. These come from BCLC - Oddset ratings and should be dated Saturday February 17, 2007. So they're already a little outdated. I did a regression and got a score for each team that's accurate by about 1%, based on the estimated winning percentage that would maximize BCLC revenue. It's not very often you see a regression as good as this one, although there were only about 40 degrees of freedom (30 variables to solve for). I'll see how this develops over time, but some people may find it interesting. If the ratings are off your expected percentages there is a chance you could make money betting on that team (expect Boston to make the playoffs, expect NYI to make playoffs). These are probably reasonably similar to other betting outlets, but I can't speak for them, it's enough work inputing information from one source. As an aside Oddset provides horrible return on bets I've basically determined it's impossible to consistently make money (your expected result has to differ by about 10% of their expectation - so if they say 50% you need to be confident it'll be 60% to make money).

Frequency Stadings Update

It's been a month since I posted these and as always find the data interesting. For example Ottawa has very low CB and a high LW, which partially can explain (or be explained by) their big wins and big losses. Anaheim lost quite a bit of ground in the last month due to the injuries and their CB lost 6 points. A month ago I said Pittsburgh could be a dark horse in the playoffs, and now they've almost secured themselves a playoff spot and could start out on home ice. So without further ado here's the list:

N name W L T ePTS SW TL OL LW CB 1 ANA 51 20 12 113 2.8 56% 37% 49% 60% 2 N.J 48 17 17 113 2.8 49% 41% 49% 67% 3 NSH 52 23 7 112 3.1 62% 34% 45% 56% 4 DET 53 23 6 112 2.7 55% 41% 58% 60% 5 BUF 48 19 15 111 3.2 49% 42% 46% 69% 6 S.J 53 28 1 107 2.6 53% 46% 66% 59% 7 DAL 47 24 11 105 2.6 46% 46% 57% 59% 8 VAN 45 28 9 100 2.7 52% 42% 50% 51% 9 CGY 46 29 7 99 2.6 52% 43% 53% 51% 10 MIN 39 24 19 98 2.9 43% 46% 44% 59% 11 PIT 42 26 13 98 3.6 50% 44% 38% 63% 12 MTL 42 28 13 96 3.0 43% 49% 48% 61% 13 OTT 46 33 3 96 2.3 53% 45% 62% 44% 14 CAR 42 34 6 90 2.8 47% 49% 52% 54% 15 T.B 39 31 11 90 3.3 47% 47% 41% 58% 16 TOR 39 32 10 89 3.1 49% 45% 42% 52% 17 ATL 38 31 13 89 3.4 47% 47% 39% 58% 18 EDM 42 36 4 88 2.6 50% 47% 52% 46% 19 NYI 38 33 12 87 2.9 45% 48% 44% 52% 20 NYR 37 35 10 84 3.1 47% 47% 40% 50% 21 WSH 36 35 10 83 2.7 42% 52% 49% 50% 22 COL 36 37 9 82 3.5 49% 46% 36% 53% 23 BOS 33 33 16 82 3.6 47% 45% 31% 54% 24 CHI 34 38 10 78 2.9 42% 51% 42% 50% 25 FLA 34 39 9 77 2.9 38% 57% 48% 54% 26 PHX 34 40 9 76 2.8 38% 56% 48% 52% 27 STL 31 39 12 74 3.3 39% 54% 37% 55% 28 CBJ 33 42 8 73 2.6 43% 52% 45% 43% 29 L.A 27 46 9 63 3.2 38% 57% 36% 50% 30 PHI 22 50 9 54 3.5 41% 54% 26% 46%

W - wins, L - losses, T - ties (for 82 game season) - estimated using simulation.
PTS - expected points using this method.
SW - average score switching - high numbers likely indicate difficulty maintaining a lead (bad defense/goaltending)
TL - Percentage the team takes lead.
OL - Percentage the opposition takes lead.
LW - Percentage the lead results in win. (otherwise it is tied again).
CB - Percentage of come backs - # of times team re-ties game. (otherwise opposition wins).

2003-2004 Frequency Standings.

As promised a month ago I have collected the frequency standings from 03-04. I suggested and showed that the 2005-2006 Stanley Cup Playoff teams did better if they had a higher come back score. The 2003-2004 playoffs didn't quite work out as well, with Tampa (57%) and Calgary (55%) going the finals. The Montreal - Boston upset (predicted at 80%) was another strange series. As a result the CB factor was negatively correlated to winning in 2003-2004. When you combine the data (2005-2006 and 2003-2004) you get the positive correlation back, however it is no longer significant (p-value = 30%). However doing a regression on the results showed that predicting wrong generally resulted in a longer series (6 games instead of 5). Either way I still don't have conclusive evidence that the CB stat helps teams win in the playoffs. That being said the other statistics (LW, OL, TL) were less significant than the CB variable so there aren't better alternatives.

N name W L T ePTS SW TL OL LW CB 1 DET 53 18 11 117 2.5 54% 39% 59% 62% 2 N.J 49 21 12 110 2.6 54% 38% 51% 54% 3 TOR 50 22 10 110 2.4 50% 43% 61% 58% 4 S.J 49 22 12 109 2.7 50% 43% 54% 60% 5 BOS 47 20 15 109 2.9 47% 44% 51% 66% 6 T.B 50 23 9 109 2.9 57% 38% 49% 57% 7 DAL 48 22 13 108 2.3 49% 42% 57% 54% 8 PHI 46 21 15 108 2.8 50% 41% 49% 60% 9 VAN 48 25 10 105 2.9 48% 46% 53% 62% 10 OTT 48 24 10 105 2.8 51% 43% 52% 58% 11 COL 46 23 13 104 2.8 51% 41% 48% 57% 12 MIN 39 22 21 99 2.6 43% 44% 43% 56% 13 STL 43 27 12 99 2.9 49% 43% 46% 56% 14 MTL 45 29 8 99 2.5 53% 42% 52% 47% 15 CGY 45 30 7 98 2.7 49% 46% 54% 55% 16 NSH 43 28 11 98 2.7 52% 41% 46% 48% 17 NYI 44 27 11 98 2.6 49% 43% 51% 50% 18 EDM 41 29 12 94 3.1 46% 47% 44% 59% 19 BUF 40 33 9 89 2.5 48% 46% 49% 44% 20 L.A 35 31 16 86 3.0 44% 46% 38% 51% 21 FLA 34 32 15 84 3.0 43% 49% 39% 52% 22 ATL 37 36 9 83 3.0 44% 50% 45% 52% 23 CAR 34 33 15 83 2.6 37% 53% 47% 51% 24 PHX 29 35 18 77 3.2 36% 54% 35% 56% 25 ANA 33 38 10 77 3.0 46% 48% 38% 46% 26 NYR 30 45 7 68 2.7 39% 57% 45% 45% 27 CBJ 28 43 10 67 2.9 42% 52% 36% 42% 28 WSH 27 45 10 65 2.9 36% 57% 40% 47% 29 PIT 27 47 8 61 3.0 37% 58% 38% 46% 30 CHI 24 46 11 60 2.9 39% 54% 33% 41%

W - wins, L - losses, T - ties (for 82 game season) - estimated using simulation.
PTS - expected points using this method.
SW - average score switching - high numbers likely indicate difficulty maintaining a lead (bad defense/goaltending)
TL - Percentage the team takes lead.
OL - Percentage the opposition takes lead.
LW - Percentage the lead results in win. (otherwise it is tied again).
CB - Percentage of come backs - # of times team re-ties game. (otherwise opposition wins).

Making the playoffs

The west is settled except for Edmonton and Colorado. Edmonton has to make up about 9 points in 26 games and Colorado has to make up 12 points in 27 games. Colorado has one advantage: 15 divisional games vs. Edmonton’s 10. Winning a lot of these games can guarantee a playoff birth, for example if Colorado wins 5/6 of the games with Calgary they get +10 on Calgary (6*2-2 = 10), theoretically tying Calgary and Colorado. Edmonton still has 5 games against Minnesota. Colorado has 4 against Vancouver. Meaning any of these teams can easily fall out of playoff contention quickly. Other than the complications in the North West the West is settled with Dallas, Anaheim, San Jose, Detroit and Nashville fighting for positions in the top 8.

The East is a little more complicated: beyond the fact we know that Boston, Washington, Florida and Philadelphia are out and Buffalo and New Jersey are in. So I’m going to make some interesting predictions and explain why:

Atlanta [3^rd]

It’d be nice for Atlanta to be guaranteed a spot, but considering in their last 18 games they’ve got a goal differential of -13 and a Pythagorean win percentage of 39% and 3 regulation wins in those games I would say Atlanta is a great candidate for falling out and it could happen quickly (Calgary, Ottawa and Carolina are up next). It appears Atlanta had a great finish last year and couldn’t make it they better hope the opposite can holds true this year: that is to say, a terrible finish, but still make it.

Pittsburg [4^th]

In the last few weeks Pittsburg has taken off and almost caught up to New Jersey. Pittsburg is 8 points from falling out of the playoffs and considering the drive of Crosby to join the post season it’s hard to imagine them not being able to pull it off. They have a lot of games to play in March so their lead could dwindle slightly, but they should make it.

Ottawa [5^th]

Ottawa has had a strange year so far with such a terrible start they appear to have everything under control in fact their Pythagorean win percentage over the last 17 games is sitting at 73% with 11/17 games won in regulation. They’d need some pretty poor luck to miss the playoffs.

Tampa Bay [6^th]

As the only team with a positive goal differential in the South East division they should win the division. Both of the games against Atlanta are at home, however, which if Atlanta gets things going again would most likely seal the division for Atlanta. They’ve won a lot of shootouts lately boosting their point totals. They’ve been the most consistent team in the South East all season and they should be able to get in, but they could easily slip out. My number crunching pegs them at positive odds (69%) chance of making the playoffs.

Montreal [7^th]

They better make it. I always enjoy watching and cheering for Montreal in the playoffs their series (most notably the one vs. Boston a couple years ago) are some of the more entertaining ones. However since January they’ve played at a Pythagorean win percentage of 35% and won 6/19 games in regulation. If they want to make it they need to get the ball rolling, the game against Florida on the 13^th should help (or really hurt). My numbers say they’re slightly favored to make the playoffs (57%).

Carolina [8^th]

Based on who’ve they’ve played and won against, Carolina is the better team in the South East division and should make it in the playoffs. They have to do a bit better than they’re currently doing, but it’s perfectly feasible.

Toronto [9^th]

There’s two ways for Toronto to make it in: Montreal falls out, or that the Atlantic and South East divisions are so bad they allow 4 teams from the North East in. I often consider the possibility of 4 teams in the playoffs from one division as not possible as it requires the better division to be a lot better than the bad divisions. Toronto still has 3 games against Montreal that could change things significantly including a game on April 7.

NY Islanders [10^th]

Of all the teams that shouldn’t make it in, the Islanders might be the team that does. However, they’ve been losing ground since the beginning of the year when they did really well. Over the last 19 games they’ve won 6 in regulation and went to overtime a lot. However they’ve maintained an even goal differential and are pegged to play a bit above average for the rest of the season. I have at 57% chance of making it.

NY Rangers [11^th]

The most important thing for the Rangers in the last stretch is that they play the Islanders 4 times, the winner of that series will likely be the one closest to making the playoffs. Too bad for Rangers their record against the Islanders is rather poor this year. I don’t see them making it, but they do have a 50-50 chance statistically speaking.

Basically it’ll work out like [best guess]:

1. Buffalo [114]
2. New Jersey [108]
3. Tampa Bay [96]
4. Ottawa [102]
5. Pittsburg [99]
6. Carolina [93]
7. Atlanta [93]
8. Toronto [91]
9. Montreal [91]
10. New York Rangers [91]
11. New York Inlanders [89]

I could easily see those tie braking rules coming into effect this season in the East.

Trade Deadline

Trades in general make both parties involved better off. Often in the NHL a trade deadline deal is a swap of high present value for high future value. Draft picks are the best representation of high future value and Stuart or Nagy represent high present value. For a team making the playoffs the present is extremely valuable (each additional playoff round is worth millions) and the present is almost worthless for a team not making the playoffs as the only thing they’re going to lose is a few ticket sales in the last 30 games.

I always find the trades that go down to provide significant insight into the value of draft picks and players. For example prior to the new (2006) CBA draft picks were often swapped for good players. Currently it appears almost mandatory to include a prospect in the other direction rather than a draft pick. This is because the new CBA makes the players RFAs after they turn 27 rather than 30 and considering that most players (after the first 20 picks or so) don’t become regulars until they turn 23 or 24, which leaves only 3-4 years of value. The salary cap also shifted the value of the RFAs in such a way that is too complicated to discuss here.

In order to look at the statistical value of the trades I added a trade feature on my website, trades up to this date are displayed here as well:

F T Name GP NET Salary EV EV+ EV- VAL PP PPS SH SHS C. KOBASEW 40 $2.3M $1.2M 10:50 2.1 1.1 0.2 2:20 4.3 0:03 5.8 A. FERENCE 54 $4.2M $0.8M 13:14 2.4 1.9 -0.09 2:03 5.1 3:12 3.7 B. STUART 48 $-1.4M $2.4M 16:44 3 4.5 -0.19 2:13 2.6 3:58 6.9 W. PRIMEAU 51 $0.4M $1.2M 11:02 1.5 3.1 -0.42 1:07 4.4 2:56 6.5 J. VASICEK 38 $2.7M $1.3M 11:45 2.7 2.6 0.05 0:44 5.2 0:42 6.2 E. BELANGER 56 $0.8M $1.3M 10:40 2.2 2.3 0.16 2:40 2.8 1:31 5.5 V. VISHNEVSKI 52 $3.5M $1.6M 16:08 2.7 3 -0.19 0:06 4.7 3:03 4.5 A. HALL 49 $0.4M $1M 8:32 1.2 3 -0.61 1:40 4.9 2:15 6 P. DUPUIS 48 $0.3M $0.8M 11:40 1.2 2.1 -0.33 0:47 4.1 2:40 7.1 S. AVERY 55 $4.1M $1.1M 12:35 2.2 3 0.21 2:31 6.6 1:45 5.8 J. WARD 46 $5.9M $0.7M 9:06 2.4 3.3 0.09 0:11 5 3:02 2.9 J. LUNDMARK 39 $1.9M $0.6M 7:46 1 1.8 -0.27 0:48 5.4 0:03 6.8 C. CONROY 52 $3.7M $2.4M 11:47 1.8 2.8 0.1 2:03 6.2 1:38 5.2 2008 2nd $1.5M 2007 4th $0.5M J. MOTZKO 7 $2.8M - 6:16 1.4 1.4 0 0:08 0 0:02 0 M. HARTIGAN 6 $4.5M $0.5M 10:12 2 0 0.17 3:38 6.8 0:00 5.5 2007 4th $0.5M 2007 1st $2.7M M. TJARNQVIST 18 $2.1M $0.5M 8:00 1.9 3 -0.23 0:10 0.1 0:04 -3.7 L. NAGY 55 $4.1M $3M 13:15 2.7 2.6 0.17 3:58 6.4 0:40 5.6 2007 1st $2.7M