NHL Situational Plays

[nash13]

First we can eliminate the smileys if we go to reply and go advanced, there is a box, where you can disable smileys.

I have two good ones:
p:streak >= 3 and p:L and site = away and p:site = home

108-155 (-0.45, 41.1%)	avg line: 110.0 / -121.3	on / against: -$4,189 / +$3,025	ROI: -14.2% / +8.8%
102-134-27 (-0.06, 43.2%)	avg total: 5.5

p:streak >= 3 and p:L and site = home and p:site = away

166-161 (0.16, 50.8%)	avg line: -149.7 / 137.0	on / against: -$6,073 / +$4,303	ROI: -12.1% / +12.5%
159-146-22 (0.29, 52.1%)	avg total: 5.5

Seems like switching sides after a winning streak ends will cause a letdown.

[emceeaye]

Courtesy of Ross Benjamin 4/20/14

NHL Betting System
Play against any money line road underdog of +100 to +150 that had a 0.3 or more goal per game differential on the season, and had allowed 3-goals or more in each of their previous 3-games. When fading the road team in that exact situation you would be 29-5 (85.3%) during the last 5 seasons.

This is my version of it, but unclear exactly what the "differential is between...

D and 150 <= line >= 100 and tA(goals@team and season and 43<=game number<=84) - tA(goals@team and season and 1<=game number<=42) >= .3 and po:goals >= 3 and ppo:goals >= 3 and pppo:goals >= 3 and season >= 2008

Here's another version of it that's pretty good:

D and 150 <= line >= 100 and tA(goals) - p:goals >= .65 and po:goals >= 3 and ppo:goals >= 3 and pppo:goals >= 3 and season >= 2009

[emceeaye]

Another one explained by Ross Benjamin in post on 4/20/14:

NHL Betting System
Play on any money line home favorite during the 2nd half of the season that’s lost 3 or more games in a row, and has a winning percentage of .510 to .600. By playing on the home favorite in that exact scenario you would be 156-60 (72.2%) during the last 18 seasons.

Here's the SDQL I think. Either way, the win% is nice:
HF and streak<=-3 and 60>=WP>=51 and game number>=43

[nash13]

I like the last two a lot. Good work.

[emceeaye]

Hey Nash13,
Do you happen to know the terms to express power play conversion rate? In particular, power play goals / power play chances?

I'm unable to find anything but power play goals and power play minutes...I suppose we could express it through power play minutes....

[otunj]

Hey Nash13,
Do you happen to know the terms to express power play conversion rate? In particular, power play goals / power play chances?

I'm unable to find anything but power play goals and power play minutes...I suppose we could express it through power play minutes....

Not sure if there is one for the traditional conversion rate, but I have an interesting idea of looking at power play success at a different angle using power play minutes. Hopefully I can explain this in a way that's not confusing.

Officially, a power play is any time a team is up one man, despite the amount of time they were on it. For example, let's say the Blackhawks were on their first pp of the game and, 29 seconds into it, they got their own penalty which ended their pp. According to the official stats, they are 0-1 on the pp, despite only being on it for 29 seconds. This is where the new angle comes in. Instead of defining a power play opportunity as it's currently done, why not define a power play opportunity as the accumulation of 2 minutes of total power play time? So, if the Hawks were only on a pp for 29 secs, they are really 0 for 0.24 for the power play (29 secs/120 secs).

Using the Blackhawks example. Let's say the Blackhawks had 3 official power plays in the game:

1) 1st power play (above) - 29 seconds of total power play time (0 for 0.24 on the pp)
2) 2nd power play - unsuccessful for the full 120 secs (new game conversion rate - 0 for 1.24 on the pp)
3) 3rd power play - GOAL!! 56 seconds in (new rate - 1 for 1.7083 on the pp)

The Blackhawks had 205 seconds of total power play time. So, based on the new definition of a power play (accumulation of 2 total minutes), the Hawks had 1.7083 power plays (205 total power play seconds/120 secs=1.7083). Using minutes, which is what SDQL uses.... 3.416 total power play minutes minutes/2 minutes. So, whereas the official stat states that the Blackhawks averaged 1 goal per 3 power plays for the game (33% conversion rate or .33 goals per pp opportunity), based on the new definition of a power play opportunity, the Hawks averaged 1 goal per 1.7083 power plays (58.5% conversion rate or .585 goals per pp opportunity) for the game.

What if we apply this to the whole season? Would a team's power play strength on the season be better measured based on the actual time spent on the power play rather than the number of times they have one (whether they're on it for 2 minutes, 34 secs, 57 secs, 1 min 11 secs)?

I'm gonna try to experiment with this over the next week. Query wise, this is what I'm thinking:

1/((tS(minutes power play)/2)/tS(power play goals)) --this will get you the average number of goals per power play opportunities (or conversion rate).

(tS(minutes power play)/2) ----- finding the number of power play opportunities (new definition) for the season. Total number of minutes on the power play divided by 2 mins

divided by....

tS(power play goals) ------ explains itself

This gets you the the average number of pp opportunities per goal.

1 / average number of pp opportunities per goal = average number of goals per pp opportunity

Hope that all made sense

[emceeaye]

Nice work, Otunj!

I like that idea. Practically speaking, though, the database times out when using that version of conversion rate:

1/((tS(minutes power play)/2)/tS(power play goals))

Anyone else have better luck?

[emceeaye]

Bruins ML

Active today:

H and p:margin>=4 and op:margin<=-4 and season>2006

[nash13]

The main problem here is, if the minutes of power play are counted full or only the real time spent on power play?
any info on that?

[emceeaye]

Here's an offshoot of another Ross Benjamin described situation:

A and tA(o:goals)>=3.2 and po:goals>=3 and ppo:goals>=3 and pppo:goals>=3 and ppppo:goals>=3 and season>=2009

It's active today on Blackhawks.

Thoughts?

[nash13]

Here's an offshoot of another Ross Benjamin described situation:

A and tA(o:goals)>=3.2 and po:goals>=3 and ppo:goals>=3 and pppo:goals>=3 and ppppo:goals>=3 and season>=2009

It's active today on Blackhawks.

Thoughts?

The sample size is ok, but the ROI is very low, if you look on the odds offered i believe these are too risky to take.
Btw where do you get he Ross Benjamin situations from?
Maybe we can share the load.

[otunj]

any one know anything about True and False? I have 2 queries that contradict one another. When I do "(Query 1) is True and (Query 2) is False" the the game & play is NOT active. However, when I do "(Query 1) is False and (Query 2) is True", the game IS active. Would that indicate that Query 2 would be the play? Thanks!

[emceeaye]

Ross Benjamin situation description:

4/10/14 post
Any road team which is -100 to -150 on the money line in the 2nd half of the season that’s coming off of 4 or more games in a row which have gone under the total, and they’re averaging 3.0 goals or more per game on the season, has gone 47-17 (73.4%) during the last 18 seasons. Bet Colorado over Vancouver for Hockey Picks.

My rendition of the query of his description:

A and -150<=line<=-100 and ou streak<=-4 and tA(goals)>=3 and game number>=42

Nice

[emceeaye]

The sample size is ok, but the ROI is very low, if you look on the odds offered i believe these are too risky to take.
Btw where do you get he Ross Benjamin situations from?
Maybe we can share the load.

Agreed, the juice is way too high for a play...

Find NHL showcased daily picks here on SBR. Look for a Ross Benjamin prediction, and then on the bottom of the page click on his name and then you'll have access to all of his posts. Here's the SBR page link:

http://www.sportsbookreview.com/handicappers/ross-benjamin-16/

I've already gotten to 4/10/14 working backwards in his prediction posts...I think a lot to go. There will sometimes be two "systems" showcased in one post, just in case you miss it.

Thanks for the help and for your z-score formula.

[Mr BBall]

Good thread... keep it up guys!

[nash13]

any one know anything about True and False? I have 2 queries that contradict one another. When I do "(Query 1) is True and (Query 2) is False" the the game & play is NOT active. However, when I do "(Query 1) is False and (Query 2) is True", the game IS active. Would that indicate that Query 2 would be the play? Thanks!

I read about that in the GoogleGroup, could you share your code?

https://groups.google.com/forum/#!topic/sportsdatabase/NgdepBR29Wo

[otunj]

I read about that in the GoogleGroup, could you share your code?
https://groups.google.com/forum/#!to...se/NgdepBR29Wo

No problem:

WP < 45 and o:WP > 45 and op:L and month = 10

96-70 (0.34, 57.8%)	avg line: 111.2 / -122.7	on / against: +$3,999 / -$4,867	ROI: +21.0% / -21.9%
89-65-12 (0.45, 57.8%)	avg total: 5.5

A and 12 > tA(penalty minutes) > 11 and month != 11 and month != 12 and game number >= 5

536-505 (0.04, 51.5%)	avg line: 123.9 / -135.7	on / against: +$14,118 / -$19,574	ROI: +12.5% / -13.3%
463-474-104 (0.20, 49.4%)	avg total: 5.4

Love the sample size on the 2nd one and any ROI above 10% is beauty. The ROI only in the month October since 2006 is 36.7%.

Update: Looks like the 1st query will be 3-0 tonight (Bruins & Leafs win; Stars winning 6-1 over canucks). The 2nd query has the play as the canucks. Looks like i misspoke on the previous post though. Doesn't matter which query is True and which is False...both will show active plays that contradict each other

[nash13]

No problem:

WP < 45 and o:WP > 45 and op:L and month = 10

96-70 (0.34, 57.8%)	avg line: 111.2 / -122.7	on / against: +$3,999 / -$4,867	ROI: +21.0% / -21.9%
89-65-12 (0.45, 57.8%)	avg total: 5.5

A and 12 > tA(penalty minutes) > 11 and month != 11 and month != 12 and game number >= 5

536-505 (0.04, 51.5%)	avg line: 123.9 / -135.7	on / against: +$14,118 / -$19,574	ROI: +12.5% / -13.3%
463-474-104 (0.20, 49.4%)	avg total: 5.4

Love the sample size on the 2nd one and any ROI above 10% is beauty. The ROI only in the month October since 2006 is 36.7%.

Update: Looks like the 1st query will be 3-0 tonight (Bruins & Leafs win; Stars winning 6-1 over canucks). The 2nd query has the play as the canucks. Looks like i misspoke on the previous post though. Doesn't matter which query is True and which is False...both will show active plays that contradict each other

i have two different angles on that.
the first one is to look up if the roi and profit varies a lot if i have one query true and the other one false.
the second one is harder. i decided to give points of success for each of my trends.
these go from lowest 1 up to 3. if two trends collide i subtract the ones from another and the bigger one gets the bet.
the amount is linked to the strength of each trend.
since the first trend is only for the month of october it may worth a look how the second trend is doing then.
in my ranking system second query has 0.5 points more than the first one. so it would be 0.5 units on it.

i had the canucks active in one of my strongest trends. so sometimes it is really hard to decide which one to take.

4 < po:goals < 7 and p:FL

[nash13]

Here is one more of Ross Benjamin
NHL Betting System
Any road team that’s -100 to -150 on the money line in the month of April, and they’re coming off a road loss, has gone 68-30 (69.4%) over the last 18 seasons.

I tweaked it a bit

A and -100 >= line >= -150 and p:AL and (month = 3 or month = 4)

97-59 (0.62, 62.2%)

avg line: -121.4 / 112.0

on / against: +$2,517 / -$3,112

ROI: +13.3% / -19.7%

[otunj]

i have two different angles on that.
the first one is to look up if the roi and profit varies a lot if i have one query true and the other one false.
the second one is harder. i decided to give points of success for each of my trends.
these go from lowest 1 up to 3. if two trends collide i subtract the ones from another and the bigger one gets the bet.
the amount is linked to the strength of each trend.
since the first trend is only for the month of october it may worth a look how the second trend is doing then.
in my ranking system second query has 0.5 points more than the first one. so it would be 0.5 units on it.

i had the canucks active in one of my strongest trends. so sometimes it is really hard to decide which one to take.

4 < po:goals < 7 and p:FL

Thanks! Will you please check your private message, I sent you a question.

That said.....here's one of my favorite ones. Don't get scared about the length. If you're familiar with hockey sabremetrics, I was able to figure out a query related to a team's PDO. For those that don't know about that, a real quick explanation is the sum of a team's save % and their shooting percentage. Over the course of a season, even if a team starts real strong at the beginning of the season (let's say PDO 104 or 105), their PDO tends to regress towards the mean (100) as the season progresses. The opposite is true as well. Most team's that really suck at the beginning, will get somewhat better with their save % or scoring % which will eventually increase their PDO

(tS(goals) * 100 / tS(shots on goal)) + ((tS(o:shots on goal-o:goals) / tS(o:shots on goal)) * 100) > 100 and (oS(goals) * 100 / oS(shots on goal)) + ((oS(o:shots on goal-o:goals) / oS(o:shots on goal)) * 100) < 100 and tA(shots on goal) < oA(shots on goal) and month = 10 and game number >= 6

93-123 (-0.28, 43.1%)	avg line: -116.0 / 104.6	on / against: -$5,052 / +$4,015	ROI: -18.0% / +15.7%
107-99-10 (0.32, 51.9%)	avg total: 5.5

The part in red means: PDO>100 and o:PDO<100

[JMon]

otunj..do you think I played around with your query too much?

(tS(goals) * 100 / tS(shots on goal)) + ((tS(o:shots on goal-o:goals) / tS(o:shots on goal)) * 100) > 100 and (oS(goals) * 100 / oS(shots on goal)) + ((oS(o:shots on goal-o:goals) / oS(o:shots on goal)) * 100) < 100 and tA(shots on goal) < oA(shots on goal) and (month!=1 and month!=6 and month!=4 and month!=11) and -180<line<140 and 60>=wins>=2 and o:losses<40 and game number>4

507-624 (-0.17, 44.8%)	avg line: -114.8 / 104.5	on / against: -$21,212 / +$15,756	ROI: -15.5% / +12.6%
481-561-89 (0.08, 46.2%)	avg total: 5.5

[nash13]

something wrong with the query or am i missing something?
and -180=wins>=2

if i exclude this part, everything is fine

[JMon]

something wrong with the query or am i missing something?
and -180=wins>=2

if i exclude this part, everything is fine

damn sbr interface. Click "reply with quote" on my post and you will see it in full.

[nash13]

thx a lot

[emceeaye]

non obvious one with good record

Here's one I came up with whose logic isn't immediately obvious to me but which increases in record as you increase the total goals scored in previous game plus the game before that. It also appears to systematically increase as the differential between a team's goals scored away and at home increases in favor of away...of course as you increase these values, sample size decreases to an unacceptably low level.


AD and tA(goals@team and season and A)>=tA(goals@team and season and H)+1.02 and p:goals+pp:goals>=7 and season>=2007

Thoughts?

[emceeaye]

Ross Benjamin situation posted on 4/9/14 on SBR hockey picks:

Any money line home favorite of -200 or less that’s playing with exactly 2-days of rest, versus an opponent that’s playing with 3 or more days of rest, has gone 124-53 (70.1%) over the last 18 seasons

HF and line>=-200 and rest=2 and o:rest>=3

[JMon]

Ross Benjamin situation posted on 4/9/14 on SBR hockey picks:

Any money line home favorite of -200 or less that’s playing with exactly 2-days of rest, versus an opponent that’s playing with 3 or more days of rest, has gone 124-53 (70.1%) over the last 18 seasons

HF and line>=-200 and rest=2 and o:rest>=3

H and -110 >= line >= -200 and rest = 2 and o:rest >= 3 and total > 5 and month!= 11

[emceeaye]

H and -110 >= line >= -200 and rest = 2 and o:rest >= 3 and total > 5 and month!= 11

nice tweak JMon!

[mdunlap3]

Any way to see PL results like we see RL for bases?

[mdunlap3]

Very simple...but essentially a mini-streaking(hot) team playing at home against a team with no rest who lost their last game.

H and p:W and pp:W and ppp:W and o:rest=0 and op:L and NB

[otunj]

Any way to see PL results like we see RL for bases?

Nope. Only thing you can really is run queries with margin parameters

[otunj]

otunj..do you think I played around with your query too much?

(tS(goals) * 100 / tS(shots on goal)) + ((tS(o:shots on goal-o:goals) / tS(o:shots on goal)) * 100) > 100 and (oS(goals) * 100 / oS(shots on goal)) + ((oS(o:shots on goal-o:goals) / oS(o:shots on goal)) * 100) < 100 and tA(shots on goal) < oA(shots on goal) and (month!=1 and month!=6 and month!=4 and month!=11) and -180<line<140 and 60>=wins>=2 and o:losses<40 and game number>4

507-624 (-0.17, 44.8%)	avg line: -114.8 / 104.5	on / against: -$21,212 / +$15,756	ROI: -15.5% / +12.6%
481-561-89 (0.08, 46.2%)	avg total: 5.5

Looks good on total games and ROI. My fear is that you picked and chose to include and exclude too many things. I'd observe it first see how it does, which is what I'll be doing

[emceeaye]

Here's today's situation by Ross Benjamin on the under for Canucks/Blues game tonight:

Any road team with a total of 5.0 or less, coming off 1 or more losses in a row, and has a winning percentage of .510 to .600 on the season, versus an opponent with a winning percentage of .250 to .400, has seen 32 of those 41-games (78%) go under the total since 1996.

This is the query that expresses it, but the results of the limited years database we use points to the over. Am I missing something here with this query?

A and total<=5 and streak<=-1 and 60>WP>51 and 25 < o:WP <40

[otunj]

Very simple...but essentially a mini-streaking(hot) team playing at home against a team with no rest who lost their last game.

H and p:W and pp:W and ppp:W and o:rest=0 and op:L and NB

Too many games for only a ROI if 2.2% and only 5 unit profit. If your unit was $5, you would have to wager, well over $1500 just to win $25. For me, I try to use queries that account for 5% of the games per season (around 61 per season) and an ROI of over 10%

[nash13]

Too many games for only a ROI if 2.2% and only 5 unit profit. If your unit was $5, you would have to wager, well over $1500 just to win $25. For me, I try to use queries that account for 5% of the games per season (around 61 per season) and an ROI of over 10%

i mix it a bit more. will share some examples later.