1. #1
    Jay Edgar
    Jay Edgar's Avatar Become A Pro!
    Join Date: 03-08-06
    Posts: 1,576

    Jay Edgar TRUE money lines (based on the last 9 years of NBA play)

    I sense there is opportunity with NBA money lines. If so, I'd like to be there.

    For my money ML betting provides a superior in-game experience -- unless you enjoy the constant terror of a back door cover or a bad beat from guys who are playing to the real and not the ATS scoreboard. I like having a common purpose with the NBA team I'm backing, just like in MLB.

    But in approaching the question of money lines I've been frustrated by lack of good data about the money lines that the best bookmakers hung on past games.

    Good historical MLs are much harder to find than historical point spreads. And when data is available it is usally garbage like -145/+125 rather than the decent MLs available at places like Pinny.

    So I decided to come at it from the other side.

    I spent a little time figuring out the TRUE money lines for the NBA since 1997.

    For example, what's the probability that a 3.5 point favorite will win the game?

    Turns out that since Haloween 1997, including playoffs, there have been 629 games with a closing pointspread of 3.5 (I used the records at Covers, which seem pretty standard.) The favorite won 390 of those 629 games straight up, a .620 winning percentage. The break even money line for a 62% proposition is -163. Mix in the results for games that closed at 3.0 points and 4.0 points to broaden the sample base and account for some pointspread fluctuation, and the teams closing at "-3.5" won 1170 of 1904 games straight up for a 61.4% percentage and an equivalent money line of -159.

    Here are the true money lines as determined by all results and official closing lines in the last 9 seasons (1997-98 through 2005-06.)

    (Usual caveat: I'm not a numbers guy, but I like to play around with them. There are probably faster, better, and more statistically sound ways to approach these questions.)

    TRUE MONEY LINES FOR NBA GAMES -- October 1997 to June 2006

    1.5 point favorites
    won 694 of 1333 SU
    52.1%
    TRUE ML -108

    2.0 point favorites
    won 792 of 1436 SU
    55.2%
    TRUE ML -123

    2.5 point favorites
    won 984 of 1741 SU
    56.5%
    TRUE ML -130

    3.0 point favorites
    won 1065 of 1802 SU
    59.1%
    TRUE ML -145

    3.5 point favorites
    won 1170 of 1904 SU
    61.4%
    TRUE ML -159

    4.0 point favorites
    won 1225 of 1951 SU
    62.8%
    TRUE ML -174

    4.5 point favorites
    won 1257 of 1945 SU
    64.6%
    TRUE ML -188

    5.0 point favorites
    won 1188 of 1826 SU
    65.1%
    TRUE ML -193

    5.5 point favorites
    won 1207 of 1771 SU
    68.2%
    TRUE ML -214

    6.0 point favorites
    won 1169 of 1693 SU
    69.0%
    TRUE ML -223

    6.5 point favorites
    won 1190 of 1667 SU
    71.4%
    TRUE ML -249

    7.0 point favorites
    won 1058 of 1455 SU
    72.7%
    TRUE ML -266

    7.5 point favorites
    won 1014 of 1346 SU
    75.3%
    TRUE ML -305

    8.0 point favorites
    won 908 of 1192 SU
    76.2%
    TRUE ML -320

    8.5 point favorites
    won 908 of 1158 SU
    78.4%
    TRUE ML -363

    9.0 point favorites
    won 828 of 1037 SU
    79.8%
    TRUE ML -396

    9.5 point favorites
    won 782 of 947 SU
    82.6%
    TRUE ML -474

    10.0 point favorites
    won 663 of 790 SU
    83.9%
    TRUE ML -522

    10.5 point favorites
    won 592 of 688 SU
    86.0%
    TRUE ML -617

    11.0 point favorites
    won 486 of 557 SU
    87.3%
    TRUE ML -665

    11.5 point favorites
    won 429 of 493 SU
    87.0%
    TRUE ML -670

    12.0 point favorites
    won 351 of 401 SU
    87.5%
    TRUE ML -702

    12.5 point favorites
    won 416 of 460 SU
    90.4%
    TRUE ML -945

    13.0 and 13.5 point favorites
    won 407 of 440 SU
    92.5%
    TRUE ML -1233

    14.0 point and up favorites
    won 456 of 485 SU
    94.0%
    TRUE ML -1572

    At the very least, this should provide a useful index for some folks. You can also use it in reverse -- for example, if you give a team a 35% chance of winning but want to play them ATS, how many points should you have in hand before you begin to feel comfortable? (Answer: greater than 5.0)

    If I can find more October time, I hope to break out the last 3 seasons and also peek at home-road splits to see what sub-truths might lurk there.

    (And I'd also like to figure out how Ganchrow gets his data posted inside those snazzy tables.)

    More to come.
    Last edited by Jay Edgar; 10-13-06 at 11:08 PM.

  2. #2
    Dark Horse
    Deus Ex Machina
    Dark Horse's Avatar Become A Pro!
    Join Date: 12-14-05
    Posts: 13,764

    Good stuff. Thanks for all the work.

    Here's to a great season!

  3. #3
    rm18
    Update your status
    rm18's Avatar Become A Pro!
    Join Date: 09-20-05
    Posts: 22,291
    Betpoints: 207814

    It seemed like undedogs winning SU was down last year, definitely the 8 point + ones. ALso I think just on observation, not the numbers that the road team is way more likely to win by 1 point, if you are thinking about the ML or +- 1.5 points

  4. #4
    Jay Edgar
    Jay Edgar's Avatar Become A Pro!
    Join Date: 03-08-06
    Posts: 1,576

    Thanks for the comments. Here's part II:

    Same ground rules and methodology, but looking only at the most recent results -- from the past three seasons (03-04, 04-05, 05-06). There are some slight things going on, but nothing extraordinary. The more recent results show a slight drop in the rate of “upsets” when the favorite is in the 1.5-2.5 range. Conversely a there's been a minor uptick in the upset rate when the favorite is in the 7-8 range. But no wide swings anywhere. (Best test of that: the ML “pairs” created here generally fall well within the normal offered vig range.)

    1.5 point favorites
    9-year true ML -108
    3-year true ML -116

    2 point favorites
    9-year true ML -123
    3-year true ML -130

    2.5 point favorites
    9-year true ML -130
    3-year true ML -137

    3 point favorites
    9-year true ML -145
    3-year true ML -147

    3.5 point favorites
    9-year true ML -159
    3-year true ML -152

    4 point favorites
    9-year true ML -174
    3-year true ML -164

    4.5 point favorites
    9-year true ML -188
    3-year true ML -189

    5 point favorites
    9-year true ML -193
    3-year true ML -197

    5.5 point favorites
    9-year true ML -214
    3-year true ML -226

    6 point favorites
    9-year true ML -223
    3-year true ML -228

    6.5 point favorites
    9-year true ML -249
    3-year true ML -245

    7 point favorites
    9-year true ML -266
    3-year true ML -243

    7.5 point favorites
    9-year true ML -305
    3-year true ML -257

    8 point favorites
    9-year true ML -320
    3-year true ML -275

    8.5 point favorites
    9-year true ML -363
    3-year true ML -370

    9 point favorites
    9-year true ML -396
    3-year true ML -444

    9.5 point favorites
    9-year true ML -474
    3-year true ML -518

    10 point favorites
    9-year true ML -522
    3-year true ML -527

    10.5 point favorites
    9-year true ML -617
    3-year true ML -554

    11 point favorites
    9-year true ML -665
    3-year true ML -763

    11.5 point favorites
    9-year true ML -670
    3-year true ML -675

    12.0 point favorites
    9-year true ML -702
    3-year true ML -820

    12.5 point favorites
    9-year true ML -945
    3-year true ML -931

    13 and 13.5 point favorites
    9-year true ML -1233
    3-year true ML -1267

    14+ point favorites
    9-year true ML -1572
    3-year true ML -3900

    On the 3-year results, the sample size is still pretty good most of the way up the line here. It peaks at about 750 with the "4.5" sample and it does not drop under 200 until we get up to "11.5" sample.

    Now to look at how home favorites and road favorites compare in getting the straight up win.
    Last edited by Jay Edgar; 10-14-06 at 09:30 PM.

  5. #5
    moses millsap
    moses millsap's Avatar Become A Pro!
    Join Date: 08-25-05
    Posts: 8,289
    Betpoints: 1260

    Thanks for the post JE. I don't mind betting ATS, since I'm usually getting points, but insightful post and research on your end. Cheers.

  6. #6
    Ganchrow
    Nolite te bastardes carborundorum.
    Ganchrow's Avatar Become A Pro!
    Join Date: 08-28-05
    Posts: 5,011
    Betpoints: 1088

    Great work, Jay.

    I used your smoothing methodology on Covers NBA regular season data going back to to the 1990-91 season and broke down the results by home and road favorites just as you suggested.

    All Faves
    Spread N Fave
    Win %
    Equiv
    Line
    1,987 52.19% -109.2
    2 2,134 55.58% -125.1
    2,636 56.71% -131.0
    3 2,755 59.46% -146.6
    2,944 61.45% -159.4
    4 2,911 63.35% -172.8
    2,950 63.83% -176.5
    5 2,772 64.75% -183.7
    2,801 67.01% -203.1
    6 2,607 69.04% -223.0
    2,593 71.65% -252.8
    7 2,244 73.13% -272.1
    2,184 75.55% -309.0
    8 1,922 75.65% -310.7
    1,944 78.45% -364.0
    9 1,721 79.20% -380.7
    1,684 81.89% -452.1
    10 1,421 83.04% -489.6
    10½ 1,322 84.95% -564.3
    11 1,038 86.61% -646.8
    11½ 953 87.72% -714.5
    12 798 88.97% -806.8
    12½ 808 89.73% -873.5
    13 659 91.20% -1036.2
    13½ 615 91.06% -1018.2
    14 479 93.53% -1445.2
    14½ 463 94.17% -1614.8
    Home Faves
    Spread N Fave
    Win %
    Equiv
    Line
    1,068 52.53% -110.7
    2 1,154 54.94% -121.9
    1,459 55.65% -125.5
    3 1,604 57.92% -137.6
    1,799 59.09% -144.4
    4 1,855 61.83% -162.0
    1,935 62.48% -166.5
    5 1,860 63.98% -177.6
    1,905 65.56% -190.4
    6 1,827 68.09% -213.4
    1,850 71.30% -248.4
    7 1,654 72.55% -264.3
    1,634 75.28% -304.5
    8 1,505 75.55% -309.0
    1,575 78.98% -375.8
    9 1,426 79.38% -385.0
    1,412 82.15% -460.3
    10 1,192 83.39% -502.0
    10½ 1,148 85.28% -579.3
    11 914 86.76% -655.4
    11½ 866 87.76% -717.0
    12 742 89.35% -839.2
    12½ 749 90.39% -940.3
    13 615 92.20% -1181.3
    13½ 573 91.97% -1145.7
    14 456 93.86% -1528.6
    14½ 445 94.16% -1611.5
    Road Faves
    Spread N Fave
    Win %
    Equiv
    Line
    919 51.80% -107.4
    2 980 56.33% -129.0
    1,177 58.03% -138.3
    3 1,151 61.60% -160.4
    1,145 65.15% -187.0
    4 1,056 66.00% -194.2
    1,015 66.40% -197.7
    5 912 66.34% -197.1
    896 70.09% -234.3
    6 780 71.28% -248.2
    743 72.54% -264.2
    7 590 74.75% -296.0
    550 76.36% -323.1
    8 417 76.02% -317.0
    369 76.15% -319.3
    9 295 78.31% -360.9
    272 80.51% -413.2
    10 229 81.22% -432.6
    10½ 174 82.76% -480.0
    11 124 85.48% -588.9
    11½ 87 87.36% -690.9
    12 56 83.93% -522.2
    12½ 59 81.36% -436.4
    13 44 77.27% -340.0
    13½ 42 78.57% -366.7
    14 23 86.96% -666.7
    14½ 18 94.44% -1700.0

  7. #7
    Dark Horse
    Deus Ex Machina
    Dark Horse's Avatar Become A Pro!
    Join Date: 12-14-05
    Posts: 13,764

    Good stuff.

    Copied, printed, and filed.

  8. #8
    Jay Edgar
    Jay Edgar's Avatar Become A Pro!
    Join Date: 03-08-06
    Posts: 1,576

    Thanks for training your talent on it, Ganchrow. That's invaluable stuff. I think there's a legitimate news item in that home/road breakdown. I'd be interested in other thoughts. I would put it like this:

    In the NBA, small home favorites (2-7 points) do not win their games outright as frequently as small road favorites giving the same number of points.

    All other things being equal, and with ATS and ML lines in a normal relationship to each other, the ML is more likely to yield the superior value if you like the road team (favorite or dog) in a low-spread game and the pointspread is more likely to yield the superior value if you like the home team (favorite or dog) in a low-spread game.

    At least that’s what it looks like to me.
    Last edited by Jay Edgar; 10-15-06 at 12:35 PM.

  9. #9
    Ganchrow
    Nolite te bastardes carborundorum.
    Ganchrow's Avatar Become A Pro!
    Join Date: 08-28-05
    Posts: 5,011
    Betpoints: 1088

    Quote Originally Posted by Jay Edgar (emphasis added)
    All other things being equal, and with ATS and ML lines in a normal relationship to each other, the ML is more likely to yield the superior value if you like the road team (favorite or dog) in a low-spread game and the pointspread is more likely to yield the superior value if you like the home team (favorite or dog) in a low-spread game.
    With this comment you're making the implicit assumption that the presumed "normal relationship" that exists between point spreads and money lines is not a function of any other factors (for example, offensive/defensive performance, coaching style, playoff proximity, or, as most relevant to this discussion. home/road status).

    We do have a priori knowledge that all else equal home teams exhibit an advantage over their road team counterparts. But while we do know this to be true from the historical record, there is no unambiguous rationale for this phenomenon. I propose that it may be this same unspecified factor that also influences the relatively diminished likelihood of a given lower spread home favorite winning the game straight up versus a road fave with the same spread.

    This is somewhat intuitively appealing as it's consistent with what seems anecdotally accurate; namely, that a home favorite with a given likelihood of winning a particular game SU is more likely to win by a relative blowout than a road team with the same likelihood of winning the game straight up. This is obviously a rather simple hypothesis to test and can be accomplished simply by comparing the sample deviation of MOVs for home teams versus that for road teams. Unfortunately, I don't have the analysis I've done on this in the past handy, but if memory serves me correctly what you do indeed find is that the sample dev for fave MOVs is indeed generally greater than for road teams.

    So anyway, this hypothesis, if true, does provide an explanation for the variation in fave win % for given point spreads across home/road status. The next step, of course, would be comparing the sample numbers with actual posted money lines. However, given the fact that linesmakers as a general rules have access to the same (or better) raw data as that which can be found on Covers or Don Best, I'd suspect (and indeed from experience, know) that by doing this work all one would find is that books already translate MLs to spreads differently for home teams than they do for road teams. That doesn't mean that given actual supply and demand considerations there isn't value to be found in judiciously choosing on a per situation basis between MLs and pointspreads; but, without looking at actual historical ML data, I don't think it accurate to infer in the general case that do or do not MLs imply excess value strictly across home/road classifications.

    Anyway, I'll try (but no promises) to do something with the raw ML data at some point later in the week.

  10. #10
    Ganchrow
    Nolite te bastardes carborundorum.
    Ganchrow's Avatar Become A Pro!
    Join Date: 08-28-05
    Posts: 5,011
    Betpoints: 1088

    Let me just say in conclusion that while this is definitely a very good start, linesmakers are simply too smart to ignore this particular phenomenon. I'd suggest that you'd find it more profitable to investigate ML/point spread mispricings in relation to match up variables more complex than home faves versus road faves. (The real challenge lies in identifying these variables.)

    Without giving away too much on a public message board, I can say you that if you're clever about it, you will find unambiguous value by proceeding in this direction.

  11. #11
    Jay Edgar
    Jay Edgar's Avatar Become A Pro!
    Join Date: 03-08-06
    Posts: 1,576

    Ganchrow, thanks for the solid comments and analysis. I hope you find the time to drill down further and will share freely.

    I also hope that the way we write (“self-assured” would be a gentle way to describe both our styles) won’t stop others from joining in with their thoughts.

    Just a couple of comments for now:

    1. On what Ganchrow saw as my “implicit assumption”: I guess I should have made a more blatant disclaimer. When I wrote “all other things being equal” I didn’t mean to imply that it was safe to assume that “all other things” will always be equal. I have no doubt that the leading books (the ones that actually make the opening line) are aware of the man-made trend* (see footnote) we have just identified. But what I don’t know is whether they feel the need to adjust for it regularly or – and this is the important part – to what extent the adjustment, if any, survives in the closing line as moved by the public. That’s what matters. And that’s where the actual historical closing MLs become the useful and logical place to look, as Ganchrow points out. But recall how and why I started the thread – because I wasn’t finding much good historical ML data. There’s very good historical pointspread data, and of course complete historical W/L and H/R records, so I was working with what I had. If anyone can uncover a source of good ML data (such as what the 10-cent money line at a place like Pinny opened and closed at over the last few years) that would be a huge leap forward for a certain kind of follow-up inquiry.

    2. But that leads me to the other main thing I want to say. Some of Ganchrow’s comments reflect an interpretation (perhaps reasonable) that my main goal is to discover something that the bookmakers are systemically overlooking and exploit it. (I may have fed this misimpression by how I phrased the end of my last post.) Actually, that’s not my goal at all –- in part because I agree with Ganchrow that it is largely a fool’s mission to hunt for any systemic oversight by the books. (At least I think he is saying that: he sort of hints toward the end that me may be privy to knowledge about such a systemic oversight.)

    Ganchrow says that you need to consider more information than who’s at home and what the line is, and I agree. Like many people, I do this, and I think that the best way to do it is without reference to any line set by anyone else and based solely on the “natural trends”* (see footnote), examples of which are “offensive/defensive performance, coaching style, playoff proximity,” plus recent game results and strength of opposition, fatigue level, special motivations, etc. The main goal here is to sharpen the tools used to make that personal line for individual games, especially the tools that will make a personal pointspread and a personal ML equally accurate as predictors of an upcoming result in an individual game.

    3. Luckily for us, in making a personal line we are not in a pure sharpness competition with the books alone – that would be tough. We are in an (easier) sharpness competition with the line as moved by the public. What keeps us involved is the belief that it is not a fool’s mission to try to make a personal line that is sharper than the closing line, which I believe (and maybe I am naïve) goes where the public takes it. If the books with all their knowledge set a smarter line and the public brings it to a dumber place, or if the books set a dumber line (whether on purpose because of the anticipated public response, or because they simply make a relatively rare error), and we can identify when that happens, we have a decent chance.

    4. On “giving away too much on a public message board.” To each his own, and again maybe I’m naïve, but I’m not much worried that any golden eggs we uncover here will need to be kept under wraps so that the goose may live. As Ganchrow says, the books already know everything. And as for alerting or educating the public, well . . . . at least in football, the public has overbet favorites and moved the line consistently in the wrong direction for 40 years running. The public has shown no ability at all to digest and employ improved handicapping techniques of any complexity. Plus, for better or worse, we are sheltered from the public in here. All due respect, this ain’t exactly the Sports section of the USA Today. This particular thread has about 100 views, which is high for a thread on this board. (That’s one of the things I love about the SBR Forum -- have you seen what kind of garbage posts your average 5000-viewer thread attracts elsewhere?) Educating each other is what this is supposed to be all about, in my view.

    That’s longer than I wanted to post. This is fun. Getting other ideas and comments would be great.

    ======================================== =
    * I’d call the home/road imbalance in how often a 4-point favorite wins SU a “man-made” trend because it depends on where the pointspread closes and (contrary to Ganchrow, perhaps) I would not be comfortable strictly analogizing it to a “natural trend” such as the 61-39% imbalance in whether the home or road team wins SU. Linesetters (books) and line movers (public) have a role in creating a “man-made” trend but only the players have any say over a “natural” trend.
    Last edited by Jay Edgar; 10-16-06 at 12:20 PM.

  12. #12
    bigboydan
    bigboydan's Avatar Become A Pro!
    Join Date: 08-10-05
    Posts: 55,425

    good stuff guys.

    I'm glad to see others are doing there NBA homework right now besides just me

  13. #13
    Ganchrow
    Nolite te bastardes carborundorum.
    Ganchrow's Avatar Become A Pro!
    Join Date: 08-28-05
    Posts: 5,011
    Betpoints: 1088

    Quote Originally Posted by Jay Edgar
    I also hope that the way we write (“self-assured” would be a gentle way to describe both our styles) won’t stop others from joining in with their thoughts.
    Hear, hear. Btw, I'd happily take a -1000 line saying that you're a lawyer.

    Quote Originally Posted by Jay Edgar
    1. On what Ganchrow saw as my “implicit assumption”: I guess I should have made a more blatant disclaimer. When I wrote “all other things being equal” I didn’t mean to imply that it was safe to assume that “all other things” will always be equal.
    Nor did I intend to imply that this was a notion that you somehow overlooked in your analysis. I really just wanted to make this factor explicit.

    Quote Originally Posted by Jay Edgar
    I have no doubt that the leading books (the ones that actually make the opening line) are aware of the man-made trend* (see footnote) we have just identified. But what I don’t know is whether they feel the need to adjust for it regularly or – and this is the important part – to what extent the adjustment, if any, survives in the closing line as moved by the public. That’s what matters.
    Extremely well-put (although see my response to your footnote in my next post). I'm thinking I'd now take the the lawyer-line up to -1400.

    Quote Originally Posted by Jay Edgar
    And that’s where the actual historical closing MLs become the useful and logical place to look, as Ganchrow points out. But recall how and why I started the thread – because I wasn’t finding much good historical ML data.
    I'll grant that I certainly perverted the original intention of this thread.

    Quote Originally Posted by Jay Edgar
    There’s very good historical pointspread data, and of course complete historical W/L and H/R records, so I was working with what I had. If anyone can uncover a source of good ML data (such as what the 10-cent money line at a place like Pinny opened and closed at over the last few years) that would be a huge leap forward for a certain kind of follow-up inquiry.
    The Don Best odds archives do have historical ML data from Pinnacle, CRIS, and maybe two or three other top-shelf books. I would not, however, vouch for either their cleanliness or general accuracy.

    Quote Originally Posted by Jay Edgar
    But that leads me to the other main thing I want to say. Some of Ganchrow’s comments imply that the main goal is to discover something that the bookmakers are systemically overlooking and exploit it. ... Actually, that’s not my goal at all
    That's just what I personally happen to find most interesting.


    Quote Originally Posted by Jay Edgar
    I agree with Ganchrow that it is largely a fool’s mission to hunt for any systemic oversight by the books. (At least I think he is saying that: he sort of hints toward the end that me may be privy to knowledge about such a systemic oversight.)
    To be perfectly blunt, what I'm saying is that while systematic oversights certainly do exist the analysis necessary to uncover them is going to be a bit more complex than that which we've done above. But please do bear in mind that the reason why I responded to this thread with any more than just raw tabular data is because if any part of your goal were indeed to uncover relative value then IMHO you're very much heading out along the right path.

    Quote Originally Posted by Jay Edgar
    Ganchrow says that you need to consider more information than who’s at home and what the line is, and I agree. Like many people, I do this, and I think that the best way to do it is without reference to any line set by anyone else and based solely on the “natural trends”* (see footnote), examples of which are “offensive/defensive performance, coaching style, playoff proximity,” plus recent game results and strength of opposition, fatigue level, special motivations, etc.
    Invaluable to be sure, but the point I'm trying to get across is that if a bettor were so inclined, applying this same analysis to identifying relative mispricings between related assets might not be the worst usage of his time (that is, of course, if he weren't too busy overbilling a client.)

    Anyway I'm going to need to respond to the rest later ...

  14. #14
    Jay Edgar
    Jay Edgar's Avatar Become A Pro!
    Join Date: 03-08-06
    Posts: 1,576

    Quote Originally Posted by Ganchrow
    Hear, hear. Btw, I'd happily take a -1000 line saying that you're a lawyer.
    I could take the 5th, but there wouldn't be much point in it.
    Pay the man.
    Last edited by Jay Edgar; 10-16-06 at 01:15 PM.

  15. #15
    RickySteve
    SBR is a criminal organization
    RickySteve's Avatar Become A Pro!
    Join Date: 01-31-06
    Posts: 3,415
    Betpoints: 187

    Quote Originally Posted by Jay Edgar
    What keeps us involved is the belief that it is not a fool’s mission to try to make a personal line that is sharper than the closing line, which I believe (and maybe I am naïve) goes where the public takes it. If the books with all their knowledge set a smarter line and the public brings it to a dumber place
    By that brilliant logic, fading every line move is the path to easy riches. How do you count all of the money you're piling up?

    at least in football, the public has overbet favorites and moved the line consistently in the wrong direction for 40 years running.
    From 2002-2006 in the NFL and NCAA FB, only once in those ten seasons(NFL '03) has the closing line had a higher mean square error than the opening line. Please post your data from the previous 35 years which support your claim, including the source for opening/closing lines beginning in 1967. Or spare us your idle conjecture.

  16. #16
    Jay Edgar
    Jay Edgar's Avatar Become A Pro!
    Join Date: 03-08-06
    Posts: 1,576

    Thanks for the close reading and the comments, Ricky Steve. Looks like something in my tone hit a nerve. Sorry about that. To me your issue seems a minor point within the entire series of posts, but any feedback is good feedback.

    (Just remember, I'm the guy trying to get the best objective info to set a sharp line and take advantage of a single-game discrepancy no matter who is responsible for it and when it shows up in the betting cycle. Ganchrow's your guy on systemic flaws in somebody's approach. )

    And if you read the offending comment in context, it's clear that the purpose of that throwaway line about "40 years" in a very long post had everything to do with trying to coax Ganchrow out of his shyness about sharing any insights for fear that the public would get wind and sap them of value. As for support for the proposition that the public is historically slightly more wrong than right, I had in mind stuff like:

    (1) Memory of reading over and over again in the 1980s that the line move picked winners in the 48-49% range. That stuff's not online that I can find (sorry), but it was an article of faith in those years. Wish I had back issues of Gambling Times or old interviews with people like Bob Martin at my fingertips, but I don't.

    (2) A bunch of studies like this one from 2003

    http://www.econ.yale.edu/seminars/ap...itt-040304.pdf

    and earlier ones cited in there. (Favorites cover 48%+ from 1981-2002 even though bettors prefer them. All Wagerline bets hit at 49% in 2001.)

    (3) A general sense that, at least as measured by number of bets as tracked by Sports Insights and others, the public continues to back the wrong horse at least slightly, if not substantially, more often than not.

    But you are surely right that there are more wise guys and syndicates now and that times have changed radically from Bob Martin's day to Simon Noble's. Noble's little history lesson here is surely accurate.

    http://www.docsports.com/pinnacle-pulse-10.html

    I'm also sure that your statement on post-2002 data is solid. Can you "spare us" a link to it? That's the ideal -- the more info the merrier, with no calls for anyone to stifle themselves.

    (P.S. - as I'm sure is obvious, vigorish explains why nobody could get rich flatly fading a 48-49% handicapper.)
    Last edited by Jay Edgar; 10-16-06 at 11:57 PM.

  17. #17
    RickySteve
    SBR is a criminal organization
    RickySteve's Avatar Become A Pro!
    Join Date: 01-31-06
    Posts: 3,415
    Betpoints: 187

    Quote Originally Posted by Jay Edgar
    (1) Memory of reading over and over again in the 1980s that the line move picked winners in the 48-49% range. That stuff's not online that I can find (sorry), but it was an article of faith in those years. Wish I had back issues of Gambling Times or old interviews with people like Bob Martin at my fingertips, but I don't.
    Vegas in the 70's : contemporary bookmaking :: Mayberry flea market : EBay

    Quote Originally Posted by Jay Edgar
    (2) A bunch of studies like this one from 2003

    http://www.econ.yale.edu/seminars/ap...itt-040304.pdf

    and earlier ones cited in there. (Favorites cover 48%+ from 1981-2002 even though bettors prefer them. All Wagerline bets hit at 49% in 2001.)
    Steven Levitt generally does interesting work, but this paper is deeply and embarrassingly flawed.

    Quote Originally Posted by Jay Edgar
    (3) A general sense that, at least as measured by number of bets as tracked by Sports Insights and others, the public continues to back the wrong horse at least slightly, if not substantially, more often than not.
    If the line were driven by votes instead of dollars, and the data provided by sportsbooks was anywhere close to accurate, you might have a valid point.

    Quote Originally Posted by Jay Edgar
    I'm also sure that your statement on post-2002 data is solid. Can you "spare us" a link to it? That's the ideal -- the more info the merrier, with no calls for anyone to stifle themselves.
    http://tbeck.freeshell.org/

    Quote Originally Posted by Jay Edgar
    (P.S. - as I'm sure is obvious, vigorish explains why nobody could get rich flatly fading a 48-49% handicapper.)
    Why not? Between reduced juice and basic line shopping, 1-2% is more than enough of an edge.

  18. #18
    Jay Edgar
    Jay Edgar's Avatar Become A Pro!
    Join Date: 03-08-06
    Posts: 1,576

    Quote Originally Posted by RickySteve
    RickySteve, that's a cool football resource, with a great bibliography.

    Thanks for providing it to those like me who hadn’t seen it. Getting a little of your heat (whether reasonably dispensed or not) is a small price to pay.

    Got to say, though, for the current point of controversy (which has probably gone on too long as it is), it doesn’t look like a gamechanger to me.

    It tracks how line movement does ATS only from 2002 (thus heavily weighing the anamolous 2005 season). And, even then, the way I read it, the “opening line” actually picks the pointspread winner against the final line in 1467 of 2903, or 50.53%, in NFL and NCAA since 2002.

    Which means that the line movement is hitting winners at 49.47% in the games that this chart tracks -- fairly consistent with what I was saying.

    Maybe I’m reading it wrong, but I don’t think so.

    Here’s a direct link to one of the 10 pages (5 NFL and 5 NCAA) that I looked at to get the 1467-1436 record for "Line Opening" vs the pointspread, so folks might see what we’re talking about.

    http://tbeck.freeshell.org/fb/results2004.txt

  19. #19
    nosuzieno
    nosuzieno's Avatar Become A Pro!
    Join Date: 09-18-06
    Posts: 593

    somewhat amused by use of the term "the public." Most accurately, I believe that includes everyone except the books. In the alternative, it is used to describe "the majority." Either way, of course "the public" bets wrong more often than not, they don't build new casinos on the losses...

  20. #20
    RickySteve
    SBR is a criminal organization
    RickySteve's Avatar Become A Pro!
    Join Date: 01-31-06
    Posts: 3,415
    Betpoints: 187

    Quote Originally Posted by Jay Edgar
    RickySteve, that's a cool football resource, with a great bibliography.

    Thanks for providing it to those like me who hadn’t seen it. Getting a little of your heat (whether reasonably dispensed or not) is a small price to pay.
    I encourage you to make a donation to help support the site.

    Quote Originally Posted by Jay Edgar
    It tracks how line movement does ATS only from 2002 (thus heavily weighing the anamolous 2005 season). And, even then, the way I read it, the “opening line” actually picks the pointspread winner against the final line in 1467 of 2903, or 50.53%, in NFL and NCAA since 2002.
    How do the mean square errors of those samples compare?

    As MSE -> 0, ATS% -> 100. If you're consistently generating probability distributions which are less accurate than the market, you will lose long-term. Good fortune may help turn those small errors into wins in the near term, but eventually you'll be broke, especially using a proportional staking method.

  21. #21
    RickySteve
    SBR is a criminal organization
    RickySteve's Avatar Become A Pro!
    Join Date: 01-31-06
    Posts: 3,415
    Betpoints: 187

    Quote Originally Posted by nosuzieno
    somewhat amused by use of the term "the public." Most accurately, I believe that includes everyone except the books. In the alternative, it is used to describe "the majority." Either way, of course "the public" bets wrong more often than not, they don't build new casinos on the losses...
    Right. When you open a book offering +100/+100 to the idiotic masses, please let me know.

  22. #22
    nosuzieno
    nosuzieno's Avatar Become A Pro!
    Join Date: 09-18-06
    Posts: 593

    "If the line were driven by votes instead of dollars, and the data provided by sportsbooks was anywhere close to accurate, you might have a valid point."

    As indicated by stats above, "the public" hits at around 48% so 100/100 would probably pay the book anyway...remarkable as that seems based on 50/50 probability.

    But I was really just supporting your point made above-that many line movement studies inaccurately presume more people than not like a particular side, rather than more money itself. That is the danger for me in using the generic term "public" without defining it first and it can also encourage me to presume no money comes in early to move the line favorably for a larger bet and/or middle close to game time. When I see a line jump, especially late, I have to speculate whether the idiot masses rather than a few smart, large waging players are on a particular side...

    at any rate, best of luck. I have enjoyed following this post and appreciate the work done by all...

Top