Sweet mercy, whoever is running the show over at BetOnline has been drinking on the job this afternoon. These openers have been atrocious. Here's another one:
SD/CHC o8 +105
SD/CHC o8 +105
-
Comment -
Matthew - great threat. Are these lines for 5/1 or 4/30?Comment -
Yikes, you also won't find the DET game at the same value either- I noticed that one dropped from u9 -105 to a much more reasonable u8.5 -115, following Bookmaker's opener.
I'm betting once BM sets the CHC line you'll see a massive jump in that game as well. It's like amateur hour out there.
I loves me some amateur hour.Comment -
BOL is the first book I have ever seen that hangs Wrigley Field totals the day before. But do they keep those up overnight or do they take them down when they converge to Pinny/CRIS "real" openers?Comment -
From what I've seen they keep them up, then adjust immediately to the CRIS openers, just like all their other lines.Comment -
Update through 4/30:
UNDER WAGERS
Line movement accordance: 31-12-3
Avg movement: -6.35c
OVER WAGERS
Line movement accordance: 13-5-1
Avg movement: +9.37cComment -
Adding, for 5/1:
BAL/SEA u7.5 -115Comment -
Does Heritage do the same? I saw they posted an overnight last night as well, but I don't have an account there.Comment -
And we're off and rolling for Thursday, 5/2's smaller card:
TB/KC u8.5 -105Comment -
Better get in on this one fast- it is waaaaaay too high. For 5/2:
BAL/LAA u9 -110Comment -
Hey Matthew, Ques - What book are you using for the early lines. I predominately use Bet365, Pinnacle and some other random European books which don't seem to post next day lines until the present day is out of the way.Comment -
I envy your list of options. I use BetOnline. From what I can tell they open first. Twenty cent totals, but for many games, still better value than playing ten cent overnight lines at Pinny.Comment -
Cheers Matt ! Managed to signup and deposit so I am up n running ! Thx for your help.Comment -
My first tail, just for SBR points, on the BAL/LAA game tomorrow.
I worry that Blanton is a gascan and that any huge outlier might be an error rather than a true result.
Regardless of the result I appreciate your posts and wish your model the best.Comment -
Thanks mayne; honored that you see value in my plays- even if only for points. Left coast games just started, so time for an update.
Through 5/1:
UNDER WAGERS
Line movement accordance: 33-12-3
Avg movement: -7.19c
OVER WAGERS
Line movement accordance: 14-5-1
Avg movement: +12.30c
No need to adjust your screen- you are reading those numbers correctly. The last two days' worth of line movement have been very good. Here is a table charting the "Unders." Remember, we want to see large negative numbers in the far right column, indicating strong line movement downward (measured in cents):
And here are the over plays. Here we want to see large positive numbers in the far right column, indicating upward movement:Date Home Away Wager_OU WageredLine WageredOdds PinnyClosingLine PinnyClosingOdds LineMovement 20130430 SEA BAL U 7.5 -120 7 115 5 20130501 NYY HOU U 9.5 -105 8.5 116 -29 20130501 DET MIN U 9 -105 8.5 -107 -24 20130501 SEA BAL U 7.5 -115 7.5 -111 4
Date Home Away Wager_OU WageredLine WageredOdds PinnyClosingLine PinnyClosingOdds LineMovement 20130430 KC TB O 7.5 105 7.5 104 1 20130430 DET MIN O 7.5 -110 8 -113 21 20130430 TOR BOS O 7.5 -125 8.5 -125 36 20130430 TEX CWS O 7.5 -110 8 -125 33 20130501 CHC SD O 8 105 9 -111 68
Obviously these results are not typical in the size of the line movement. But you gotta take the good with the bad.Comment -
Any chance you can show us your line movement results but only of your overnight/day of the game bets (that is, your pinny line bets)?Comment -
I am going to preempt this particular post by repeating that I am NOT one to focus on my hard win/loss record.
I simply don't care. I have a complete disregard for money, and an all consuming faith in the nature of markets and the law of large numbers. Again: I do not care about individual game results. At all.
However, I want to take the time to focus on one game in particular today, that just finished up- the HOU/NYY game. I got in early on this game, spotting an opening line of u9.5 -105. My statistical model indicated that this line was much too high. Sure enough, the line fell to u8.5 +116 - a downward line movement equivalent to 29 cents (and was actually much lower than that for most of the day before closing). Anyone who claims that beating the closing line is not meaningful is wrong. And here's a perfect example of why:
The game ended at 5-4 Yanks. Betting into the closer on the under on this one, you would have lost. I bet it at a better price, and won. That's a 2 unit swing in bankroll on ONE play alone. So far, I've posted about 70 plays, and am on track for ~600 or so for the year. So do the math for how many units that kind of thing evens out to. 10 units? 20, maybe? Whatever the number is, things like this are a key factor between long term losing and long term winning. That's why BTCL means everything.
[And if you're still not convinced, take a look at the SD/CHC game today- I made the sharp play on an overnight line at o8 +105, while it closed at o9 -111. If you wanted the over at close- you lost. Where I took it: a push. There's another unit swing in bankroll.]Comment -
Sorry, I don't keep those numbers, only the line and odds that I got in at.
What I can say is that in my training and validation dataset, I noticed an interesting phenomenon. My model had a winning rate against closing lines (in the 53% range), though was not as successful as it was against the openers (for obvious reasons). However, when looking at JUST games for which the line looked "good" at open but "off" at close, I was pretty much coin flipping.
On the other hand, for games where the line looked "bad" at open and "bad" at close (in other words, I would have bet the line at open OR at close, based on my confidence intervals) I was doing very well- in line with how it performed against openers.
Take that how you will. What I interpret it to mean is that my model may have been missing key real time information that was present for certain games- in the real world, the line was moving to accommodate this, but my model was blind to it. (For example, injury reports, pitchers working through injuries, etc).
Because of this, I'm working with sort of a Bayesian view: my model prediction represents my prior, and the market movement represents the data. The more the data has influenced the line (as seen by line movement), the less I believe my model.Comment -
I've wondered how sensitive your model (or any baseball model really) was to various "real world" factors. In particular I'm curious about if/how you get lineups for the overnights and how sensitive the model is to a a star (e.g. Prince Fielder) being given the day off.
Cool. And the converse too, one hopes, when the market moves in your favour.Comment -
The lineups I use are estimated pretty much as accurately as possible for overnights. There will be occasional substitutions and injury announcements between then and game time that I'm blind to (unless I re-run my analysis with the updated info, which is a lot harder to do- I have been essentially eyeballing in cases like these so far). But those will be relatively rare, and generally will have small impact. Even Fielder being out does not have a very large impact on the full run total- maybe five cents or so.
I've thought about ways to improve this further, but it is definitely a diminishing return on investment. Preseason baseball though- that is where I think an accurate lineup will give you an enormous edge. I plan to go after that angle some day in the far off future, when I've already made it and am relaxing in Florida with nothing better to do with my time.
Another play for today, 5/2:
BOS/TOR o9 +102Comment -
Stared at this one for a while. Decided to grab it now.
5/2:
DET/HOU u9 +112Comment -
Not even going to wait for the reduced juice here. Firing off my 5/3 wagers now:
MIN/CLE u8.5 -120,
BAL/LAA u8.5 -110,
TB/COL o9.5 -115,
STL/MIL o7.5 +110,
SEA/TOR o7.5 +110,
CHW/KC u8.5 -120
For reference, Miller park's roof is closed tonight, and I am assuming the same will be true tomorrow.Comment -
Got 9.5 @ bet365 now
thanks for everything man, like to read Your posts. A fresh perspective is always cool.
Comment -
Today's 5/2 he wrote 5/3Comment -
Weather caused a push in yesterday's KC game, and looks like it might cancel that under bet for today's game as well.
A pitching scratch also made the SEA/TOR wager a no action, but I just hit the over again for that game.
For 5/3:
SEA/TOR o7.5 -115Comment -
BOS/TEX o9 +105 fpr today as well.Comment -
And last but not least, took CHW/KC u7.5 +100,
for 5/3 (I'm bummed the original u8.5 bet got cancelled with Axelrod now pitching, but I still like the under here).Comment -
Is your model based on individual batter/pitcher matchups throughout the game? Do you adjust for stadium, wind and heat/humidity levels too? Do you adjust pitching matchups as pitch counts get deeper and adjust for at bat appearances too?
Also, why aren't you grading your bets vs the no vig pinnacle closer? I usually grade off of pinnacle and other market closers depending on sport. I've found different sports to have more predictive closers at books other than pinnacle. I don't know whether this is because the recent agent busts stymied volume being bet, or it's just variance or whether the other books are just the market leaders in some areas now.
Just a friendly piece of advice as well don't post more than a handful of plays. If your model turns out to have predictive value and can continuously beat the market prices for MLB totals don't degrade its edge by posting about it. There's more value in joining/being employed by syndicates or just betting yourself. Going tout is just lol though.Comment -
Got TB/COL over 9 -110 @bet365 again
Thanks for all Your hard work man ! The thread is a delight, I study every post
Comment
Are these today's games? I see TAM playing KAN? I don't see some others. Lines are different. CWS playing TEX? Am I on a different day?
