The last few years I’ve been working on applying statistical models to the NHL playoffs. Last year I seemingly made a breakthrough as my model did a far better job of predicting the teams that had a good shot at the conference finals. Armed with this information I was able to win all of the playoff pools I entered (something that I thought was nearly impossible). The main league I play in has already drafted and the others don’t read this blog so I’m free to share what my research has uncovered for the 2010 playoffs.

I use three methods to calculate out what teams I need to target. The first is a Rizer Rating. I took the stats from the playoffs for the last four seasons and noted which statistical categories set the top teams apart from the rest of the pack (it’s actually very odd how certain areas are key markers to success and they aren’t what you think). I take these key areas and based on the standard deviation within these categories apply a rating to each team based on the current regular season stats. I sum this up to produce a rating that ranges from 0 to 20. I’d caution against very high numbers indicating supremacy, anything over 10 is a solid rating.

The second is based on Poisson Distributions and applying this to determine how likely a team is to win each game of a series. I didn’t invent this method, but rather applied an article I read that did a similar thing determining what odds sports bookies should offer and what they actually offer. The premise was that sports bookies shift the odds such that they will come away with a profit, but if you know which way the odds are shifted and you bet accordingly you should come away ahead in the long run. I would note that it does skew the percentage towards the home team a little too much for my liking so I’d probably bump the percentage 5% back towards the underdog.

The third is head to head record. There’s something to be said for certain matchups. Some teams can make life miserable for others for no other reason than the makeup of their rosters. Therefore I never discount this information no matter how ridiculous it may seem.

Round 1

The East

Washington (Rizer 14.9, Poisson 71.2%, H2H 2-3)
Montreal (Rizer 5.3, Poisson 28.8%, H2H 3-2)
My Prediction: Washington 4-1

New Jersey (Rizer 10.3, Poisson 59.2%, H2H 2-6)
Philadelphia (Rizer 10.4, Poisson 40.8%, H2H 6-2) HIGH
My Prediction: Philadelphia 4-3

Buffalo (Rizer 10.4, Poisson 58.7%, H2H 2-4)
Boston (Rizer 8.6, Poisson 41.3%, H2H 4-2) HIGH
My Prediction: Boston 4-3

Pittsburgh (Rizer 10.4, Poisson 61.6%, H2H 2-2)
Ottawa (Rizer 6.0, Poisson 38.4%, H2H 2-2) MEDIUM
My Prediction: Pittsburgh 4-2

The West

San Jose (Rizer 14.9, Poisson 62.6%, H2H 2-2)
Colorado (Rizer 5.6, Poisson 37.4%, H2H 2-2)
My Prediction: San Jose 4-2

Chicago (Rizer 16.1, Poisson 66.9%, H2H 4-2)
Nashville (Rizer 4.4, Poisson 33.1%, H2H 2-4)
My Prediction: Chicago 4-1

Vancouver (Rizer 16.4, Poisson 60.1%, H2H 3-1)
Los Angeles (Rizer 8.3, Poisson 39.9%, H2H 1-3) LOW
My Prediction: Vancouver 4-2

Phoenix (Rizer 8.4, Poisson 56.8%, H2H 2-2)
Detroit (Rizer 9.8, Poisson 43.2%, H2H 2-2) HIGH
My Prediction: Detroit 4-3

Round 2 (Can’t find H2H records except for the first round)

The East

Washington (Rizer 14.9, Poisson 68.2%)
Philadelphia (Rizer 10.4, Poisson 31.8%)
My Prediction: Washington 4-2

Pittsburgh (Rizer 10.4, Poisson 57.0%)
Boston (Rizer 8.6, Poisson 43.0%) MEDIUM
My Prediction: Pittsburgh 4-3

The West

San Jose (Rizer 14.9, Poisson 61.7%)
Detroit (Rizer 9.8, Poisson 38.3%) LOW
My Prediction: San Jose 4-3

Chicago (Rizer 16.1, Poisson 57.8%)
Vancouver (Rizer 16.4, Poisson 42.2%) HIGH
My Prediction: Vancouver 4-3

Round 3 (Can’t find H2H records except for the first round)

The East

Washington (Rizer 14.9, Poisson 67.1%)
Pittsburgh (Rizer 10.4, Poisson 32.9%)
My Prediction: Washington 4-2

The West

San Jose (Rizer 14.9, Poisson 55.0%)
Vancouver (Rizer 16.4, Poisson 45.0%) HIGH
My Prediction: Vancouver 4-2

Round 4 (Can’t find H2H records except for the first round)

Washington (Rizer 14.9, Poisson 60.9%)
Vancouver (Rizer 16.4, Poisson 39.1%)
My Prediction: Washington 4-2

As a final note I must mention that in a pool where you draft players you have to be a bit reactive. Other people might have similar teams pegged to go far so when it comes to your turn it may not make sense to pick a Capital because the top 4 or 5 are gone and it’s time to move onto the next team. Oddly enough in the main draft I do (14 teams, 12 players each) the first few guys targeted Vancouver and Washington making my strategy a bit difficult to accomplish. I had to shift gears to a few second teams and ones that based on my stats have the next best chance at an upset. Here’s my roster (7th pick):

1 - Semin WSH
2 - Sharp CHI
3 - Franzen DET
4 - Theodore WSH
5 - Seabrook CHI
6 - Filppula DET
7 - Kronwall DET
8 - Corvo WSH
9 - Fedetenko PIT
10 - Ponikarovsky PIT
11 - Grabner VAN
12 - Rask BOS