Estimated NHL Over/Under Push Frequencies

**Munson15** · 02-18-08, 11:03 PM

Are you saying that playing for a side with an under 5.5 goals and an over 5 goals would be profitable? Assuming that the juice is less than what %?

**Ganchrow** · 02-19-08, 03:06 AM

Originally posted by Munson15

Are you saying that playing for a side with an under 5.5 goals and an over 5 goals would be profitable? Assuming that the juice is less than what %?

It could be profitable ... it depends on the payout odds offered on each bet (as well as your estimates of the probabilities of the game going over or under 5 -- unless you structure your bets irrespective of these probabilities). The juice is irrelevant.

Assuming you bet to equalize losses if the total lands other than 5 (thus rendering further probability estimates moot), then for a profitable betting opportunity to exist the decimal odds on the over and under must satisfy the following inequality:

<table><tr> <td rowspan=2 valign=middle> d_u5½  >  </td> <td align=center>d_o5</td></tr> <tr><td align=center style="border-top: 1px solid;"> d_o5 - 74.398% </td> </tr> </table>

or equivalently

<table><tr> <td rowspan=2 valign=middle> d_o5  >  74.398% × </td> <td align=center> d_u5½ </td></tr> <tr><td align=center style="border-top: 1px solid;"> d_u5½ - 1 </td> </tr> </table>

where d_o5 and d_u5½ correspond to the decimal odds on the over 5 and under 5½, respectively.

So, for example, if the odds on the over were -110, the odds on the under would need to be at least / ( - 74.968%) ≈ for a profitable half-middle to exist.

Conversely, if the odds on the under were -110, the odds on the over would need to be at least 74.968%* / ( - 1) ≈ for a profitable half-middle to exist.

**chemist** · 02-19-08, 06:06 AM

Originally posted by Ganchrow

Code:

Total   N  Std. Err.  Freq.
5      789    1.556%   25.602%
6    2,296    0.630%   10.148%
7      847    1.424%   21.960%

I know nothing of NHL but it seems strange that 5 and 7 should be so much more common totals given the spread total than is 6. What is the explanation for this?

Your text doesn't seem to quote very well.

**Ganchrow** · 02-19-08, 06:16 AM

Originally posted by chemist

I know nothing of NHL but it seems strange that 5 and 7 should be so much more common totals given the spread total than is 6. What is the explanation for this?

With the advent of the shootout NHL games can no longer end in a tie score.

Originally posted by chemist

Your text doesn't seem to quote very well.

It's actually an HTML table. For future refernce you can just copy and paste the table data into a pair of [CODE][/CODE] tags.

**3put** · 02-19-08, 07:20 AM

Ganchrow, I would like to know if it is possible to calculate the expected total given the o/u lines.
Examples from NHL today:

Boston vs. Carolina 5.5 -109/-101
Philadelphia vs. Ottawa 6 -131/121

**Ganchrow** · 02-19-08, 07:33 AM

Originally posted by 3put

Ganchrow, I would like to know if it is possible to calculate the expected total given the o/u lines.
Examples from NHL today:

Boston vs. Carolina 5.5 -109/-101
Philadelphia vs. Ottawa 6 -131/121

No, not without further defining the probability distribution beyond those values given for the 5, 6, and 7.

But just to be clear, is it really the expected total for which you're looking?

**3put** · 02-19-08, 09:45 AM

I am interested in the connection between the puckline and the total in a game.
Obviously in a highscoring game the favorite has a greater probability of winning by at least 2 goals.
Ottawa is the favorite today and the o/u line is 6 -131/121 indicating that the book expect that there will be scored ca. 6.2 goals on average were this game played many times.
That is what I mean by expected total.

I wondered if the push probabilities and the o/u line in combination was enough to calculate this expected total.

**Ganchrow** · 02-19-08, 10:19 AM

Originally posted by 3put

Ottawa is the favorite today and the o/u line is 6 -131/121 indicating that the book expect that there will be scored ca. 6.2 goals on average were this game played many times.
That is what I mean by expected total.

Actually an over/under defines a percentile (which would be the median in the case of a balanced market) not an expectation. A line of 6o-131/6u+121 implies that about 55.62% of non-pushed totals will go over 6 and 44.38% will go under, but does not speak to the goals scored on average. In general the average total should be higher than the median total due to the fact that totals can never be less than 1 but have no theoretical maximum value (in other words, scoring distributions are positively skewed).

Originally posted by 3put

I wondered if the push probabilities and the o/u line in combination was enough to calculate this expected total.

As I said, you'd first need to fully define the set of outcome likelihoods (which can be approximated with the push probabilities adjusted for the current line -- this is how the HPC estimates these figures). Once you have a set of outcome likelihoods with which you're satisfied, calculating the expected total becomes a trivial matter of taking the dot product of the vector of totals and the vector of associated outcome probabilities.

**rjp** · 02-19-08, 11:16 PM

There's a reason why this kind of info isn't released...

**Obioha** · 04-23-08, 08:31 PM

Just Checking

So, Sir, if I have a sportsbook that likes to post "unconventional" totals on Playoff Hockey games, according to your push frequency calc, I just need better than 3-1 in order to make money?

**Ganchrow** · 04-23-08, 08:33 PM

Originally posted by Obioha

So, Sir, if I have a sportsbook that likes to post "unconventional" totals on Playoff Hockey games, according to your push frequency calc, I just need better than 3-1 in order to make money?

Please illustrate your question with a specific example.

**Obioha** · 04-23-08, 08:36 PM

If one sportsbook hangs up over 4' at -195 and another posts under 5 at -105, that is still a profitable wager?

The reason I ask is because I only take the wager when the odds are approx 4-1, but you might be giving me a fine tuning.

**Ganchrow** · 04-23-08, 08:40 PM

Originally posted by Obioha

If one sportsbook hangs up over 4' at -195 and another posts under 5 at -105, that is still a profitable wager?

The reason I ask is because I only take the wager when the odds are approx 4-1, but you might be giving me a fine tuning.

According to the above probabilities, no, it would not be profitable.

See http://www.sbrforum.com/Betting+Tool...alculator.aspx

**Obioha** · 04-23-08, 08:45 PM

Ok, got it. Now, do you have any playoff tweaks?

**Ganchrow** · 04-23-08, 08:49 PM

Originally posted by Obioha

Ok, got it. Now, do you have any playoff tweaks?

Unfortunately due to the relative lack of data I do not.

**Obioha** · 04-23-08, 09:07 PM

Sir,

Have you done any work with NHL ties and their relationships to totals? Assuming that a game with a total of 5 is more likely to tie than a game with a total of 7. If that is the case, is 4-1 a good proposition on a regulation tie in NHL game with total of 5, 6, and 7?

**Ganchrow** · 04-23-08, 09:16 PM

Originally posted by Obioha

Sir,

Have you done any work with NHL ties and their relationships to totals? Assuming that a game with a total of 5 is more likely to tie than a game with a total of 7. If that is the case, is 4-1 a good proposition on a regulation tie in NHL game with total of 5, 6, and 7?

No, I have not posted anything on this.

To do this you'd need to analyze regulation time-only total scores and then perform much the same analysis as above.

**1capper2008** · 11-02-08, 09:48 PM

Hi,

I'm new to this forum, and I have a question about your half point calculator. Here is my example (I was looking to wager on the over on an NHL game):

Over 5.5 goals -110

Over 5.0 goals -165

My question is, "Using the half point calculator, is it a better value to wager on the Over 5.0 goals than it is to wager on the Over 5.5 goals?"

Thanks in advance!
1capper2008

**Ganchrow** · 11-03-08, 02:10 AM

Originally posted by 1capper2008

Hi,

I'm new to this forum, and I have a question about your half point calculator. Here is my example (I was looking to wager on the over on an NHL game):

Over 5.5 goals -110

Over 5.0 goals -165

My question is, "Using the half point calculator, is it a better value to wager on the Over 5.0 goals than it is to wager on the Over 5.5 goals?"

Thanks in advance!
1capper2008

Over 5 -165 is much better,

See the detailed example at the bottom of the Half-Point Calculator page.

If you have any questions about that please let me know.

**evo34** · 11-10-08, 04:01 AM

The half-point calculator seems to assign the same probabilities of score outcomes regardless of the line. That is, if you put in a total of 5, 5.5, or 6, the odds of landing on 5 are still showing as 25.6%. I see the same thing for ATS sides in other sports. Obviously, the odds of the final margin in an NFL game landing on 7 if the spread is 7 are much higher than if the spread is 2. But the calculator shows the same probability in both cases. Is your data simply the average of all past games, or does it take into account the line of the game in question, and only look at historical games with a similar line? Just making sure I understand how it works. Thanks.

**Ganchrow** · 11-10-08, 07:08 PM

Yes, Evo, that is correct.

The half-point calculator is a suitable approximation tool when traversing spreads/totals with probabilities "close" to 50/50.

**evo34** · 11-13-08, 07:05 PM

Either I don't understand or I don't agree. If an NFL game has a spread of -10, and I want to estimate the odds of the outcome landing on 10, I certainly do not want to look at what pct. of *all* games have historically landed on 10. That number will be much to low.

So I guess I am saying that I don't see how your half-point calculator can be anywhere near accurate enough unless you are looking only at point distributions of "similar" past games. [I am inferring from your pevious answer that you are not doing this].

**Ganchrow** · 11-14-08, 11:54 AM

Your inference is incorrect.

Read the methodology briefly outlined in the first post of this thread.

Different sports and spread/total combos use different radii.

**evo34** · 11-15-08, 06:39 AM

If this is true, then why would the pct. chance of a 7-point margin of victory show as the exact same number, regardless of the game spread entered into the NFL calculator? It should differ as you change the spread.

**Ganchrow** · 11-16-08, 03:00 AM

Originally posted by evo34

If this is true, then why would the pct. chance of a 7-point margin of victory show as the exact same number, regardless of the game spread entered into the NFL calculator? It should differ as you change the spread.

I'm not sure I understand your points. Have you carefully read the HPC Docs?

If so could you perhaps provide a more precise illustration of your issue?

**BatemanPatrickl** · 03-29-09, 08:42 PM

Damn Ganch. Will Hunting calls and wants his brain back

Could you take my GMAT for me?

**PapaGeorge** · 03-30-09, 11:26 PM

ok, this under over calculator..either im doing somthing wrong or i cant read....im doing last nights nhl results and im not understanding this calculator, someone help me plz

**Ganchrow** · 03-31-09, 12:37 AM

Originally posted by PapaGeorge

ok, this under over calculator..either im doing somthing wrong or i cant read....im doing last nights nhl results and im not understanding this calculator, someone help me plz

I'm not sure I exactly understand your issue. Have you carefully read the HPC Docs?

If so could you perhaps provide a more precise illustration of your problem?

**PapaGeorge** · 03-31-09, 12:14 PM

Devils -1½ +260 -110 Over 5½ +115
Rangers +1½ -320 -110 Under 5½ -135

thats last nights line, what does you calculator offer for this info?

**playinragz** · 04-19-09, 12:46 PM

forget math just call the game in as under every game from here on out !!!!

**lekki** · 10-20-09, 04:30 AM

Hi,
Based on my research about 11% of NHL games ands with total 6 goals after regular time. So it only makes sense to bet over 6 and under 6.5 only if payout it more then 96%.

Regards

**Unitage** · 10-22-09, 04:10 AM

Can anybody tell me the probability (percent) of games that go under 4.5?

**GELATINOUS CUBE** · 10-24-09, 07:19 PM

Got an ot frequency link? I did it caveman style. Pens, a lot of the old school teams were up there, while the expansion teams were low. On a global scale though nhl is hi, with finland, austria, um.. Swiss, germany. Some times. I found some other interesting trends. Anyone follow this?

**GELATINOUS CUBE** · 10-24-09, 07:20 PM

Unitage haven't seen that total in two years (nhl).