Hi guys, I'm trying to go back and reconstruct all the plays that the spreadsheet trends would have generated had we been following them from the beginning of the season. Please jump in and add any solid trends to the sheet and I'll try to grind away at it this weekend. It's time consuming, but if it shows anything similar to what the NBA is doing it's definitely worth keeping track of. Only two trends done so far, but $100 bets would be up $1270. Any suggestions are welcome.
-
Comment -
Predators/Blue Jackets under and Predators ML
Here are my picks for today:
However, I think there would be more queries, but it seems like the NHL database has not been updated from yesterday's results:
predators/Blue jackets under:
H and conference!=o:conference and total=5.5 and po:goals=0 and p:H
Predators ML (However, it's really juiced--it may be worth the puckline):
po:goals=0 and p:HF and F and season>=2011 and month=11
po:goals=0 and p:HW and game number<=42 and WP>50 and season>=2009
playoffs=0 and po:goals=0 and p:site=home and season>=2010 and HF and WP>50 and o:WP<50Comment -
^I would use the ML on parlays with other picks rather than risk the PL. Even though its a really high percentage win, a lot of those wins were by one goal.Comment -
^..i wished they did the PL as they do RL calculations in mlb, but one can figure them out by using the margin parameter. Extra work but needed.Comment -
Ok wow. I just plowed thru all the trends and came up with 31 separate trends tonight. All are on the spreadsheet. Looks a little crazy, now as there are lots of duplicate and conflicting plays, but I came up with 9 distinct plays by comparing trends. Emceeaya's Predators play was supported by 4 different trends and there was only one on Columbus, so I gave weight to the the Preds. I don't know how sustainable it will be for me to do this every day, so if anyone wants to help out it would be great :-)
Dumping in more queries is great too (for you escism, or anyone else). The more the better, then maybe we can start filtering for the best ones. Maybe it doesn't hold up once we get all the trends entered in, but just blindly tailing the first 25 or so would have someone up 30 units to this point in the season.Last edited by Cutler'sThumb; 11-29-14, 07:41 PM.Comment -
Alright, my head is starting to hurt from the sheer number of plays available each day if I go thru all the trends in the spreadsheet. There were 31 plays generated just for tonight's games. It yielded 9 distinct plays (that went 5-4), but I had to make choices regarding which trends took priority. More than half the games tonight had contradicting trends on them, and more than half the plays were duplicated by other trends. I think I'm going to start looking at sample size and ROI to filter out trends that I follow for the "plays" sheet. Otherwise it's just going to get overwhelming. If someone has any other suggestions, I'm open. This is an evolving thing so any collaboration from guys who have done this for awhile is welcome.Comment -
Thanks, yes. Problem is 5Dimes wouldn't let me do a parlay of the under and ML of one of the same teams in the under. I just bet them separately.
[TABLE_COL][/TABLE_COL]QUOTE=moshi;23097329]^I would use the ML on parlays with other picks rather than risk the PL. Even though its a really high percentage win, a lot of those wins were by one goal.[/QUOTE]Comment -
Redwings ML
2-3 queries pointing to Redwings today... Don't have them handy...Comment -
You're not kidding--It can really take longer than one would like to sift through the queries to eliminate contradictions and choose the ones that will be your picks. As a matter of routine, I usually will go with the picks supported by the most supportive and unique queries (those that are comprised of terms reflecting distinct variables and sampling both sides). I routinely add day, month, streak, o:streak, site streak, o:site streak, rest, o:rest, WP, o:WP and other variables to more closely match the unique situations of the matches in question that the original queries often do not include. I also go and see about any key injuries (and more recently how good the goalies are) that can affect the outcome. So far I have been dealing with conflicts by simply excluding them as plays. I really like Nash's method of rating the relative strength of a query to determine the strength of the play..I find that the more routine my process becomes, the more quick and efficient it gets... But I agree, it can take a while and I am welcome to any way to make the process more efficient.Last edited by emceeaye; 11-30-14, 11:56 PM.Comment -
Lightning/Rangers under
There are 2-3 queries posted in this thread and/or in the database supporting Rangers/Lightning under. I don't have them handy... But also noticed that one of Ross Benjamin's picks today supports it... Here's the translation of his description into a query:
p:HL and game number<=42 and p:division=po:division and WP>50 and o:WP<50 and season>=2010
If someone has the time to post any of those other queries, then others could see the support...Comment -
Had some free time yesterday, was messing around and came up with these:
day=Thursday and month=1 and tA(o
ower play goals)/tA(penalty minutes) < oA(oower play goals)/oA(penalty minutes) and tA(penalty minutes) < oA(penalty minutes) and tA(power play goals) > oA(power play goals) and tA(power play goals)/tA(oenalty minutes) > oA(power play goals)/oA(oenalty minutes) and tA(shots on goal) > oA(shots on goal)
(day = Saturday or day=Sunday) and game number>=42 and tA(o:goals) / tA(penalties) < oA(o:goals) / oA(penalties) and tA(goals) / tA(oenalties) > oA(goals) / oA(oenalties) and tA(shots on goal) > oA(shots on goal) and o:streak <= -2 and o:rest > 1
tA(o:goals)/tA(penalties) < oA(o:goals)/oA(penalties) and tA(goals)/tA(oenalties) > oA(goals)/oA(oenalties) and tA(shots on goal)>oA(shots on goal) and o:streak <= -5 and o:rest > 1
Please feel free to tweak to improve WP and/or ROILast edited by emceeaye; 12-01-14, 04:46 PM.Comment -
Here are the other stronger plays from the database today:
Arizona and Boston:
p:streak=-2 and p:margin<2 and site!=p:site and pp:margin!=-2 and ppp:margin>-6 and o:streak!=2 and o:streak!=1 and month!=5 and 6>site streak>-5 and n:rest<3 and on:rest<3
Montreal:
p:margin + pp:margin = -1 and p:goals > 1 and pp:goals > 0Comment -
Oh yeah... Thanks Cutler!
Bruins ML also supported by one of Ross Benjamin's situations for today... Here's the query for his description:
p:goals<=2 and pp:goals<=2 and opo:goals>=6 and season>=2010
Adding "A" however makes it not look so great in terms of the record but ROI remains good although adding "o:streak<=-2" by itself helps
Thoughts?Comment -
Can anyone figure out how to get this query to not count the same game twice (once for home and once for away)? It looks like a really good under trend, but the numbers might be skewed as it doesn't always double count.
total = 5.5 and 14 < p:shots on goal < 32 and playoffs = 0 and day != Thursday and day != Friday and day != Sunday and rest < 4 and o:rest < 4 and rest + o:rest > 0 and 7 > streak > -4 and (game number < 38 or game number > 51) and ou streak < 5 and p:u margin > -3 and 4.5 > op:u margin > -3 and 43 > op:shots on goal > 19Last edited by Cutler'sThumb; 12-02-14, 09:44 PM.Comment -
The higher rated active trends are up on the spreadsheet. Nash and I will hopefully try to get a unified rating system down. Mine is based on longevity, sample size, and ROI. I think Nash has some formulas he applies that show him how stable a trend is even if there is a smaller sample size.Comment -
Regression to the mean
Hi! I want to make query that covers the concept of regression to the mean. I want to find a way to find teams that have done good lately because of luck, not because they're that good. Im not to good at making querys (im new to sports betting, this is actually my first post on the forum), so the only thing ive come up with is this one: WP < 50 and streak >= 3 that does pretty good. But I wonder if someone has some better paramters to single out teams thats overrated because they have done better than normal lately because of luck (maybe something with number of powerplays or scoring percentage or something like that). Does anyone have som suggestions?Comment -
usually when an outcome is at least about 1.5 times or 2 times the standard deviation, you can consider it an outlier. However it becomes difficult to really determine the degree to which a result is an outlier, because in sports the mean changes all the time, and there are many variables that affect that mean as a team goes through the course of a season. Some teams stay consistent across a number of variables throughout a season, while others fluctuate more dramatically....it's not an easy question to answer in a cut and dry way.
But certainly regression to the mean is a phenomenon that affects performance in sports..but which is difficult to determine due to the interaction of non-static real identifiable variables and luck that, together, can influence an outcome.Last edited by emceeaye; 12-03-14, 03:39 PM.Comment -
Blackhawks ML
Queries NHL 88 and NHL 89 and NHL 92 in spreadsheet:
HF and tA(margin, N=2)>=3 and month>=10 and WP>=48 and season>=2012
HF and po:goals<=1 and ppo:goals<=1 and oA(period scores[0]) - oA(oeriod scores[0]) >=.25
p:margin>=2 and pp:margin>=2 and op
vertime=1 and oppvertime=1 and op:W and opp:W
Comment -
OK, I think I know what you're talking about and the angle you'd like to take in coming up with queries, namely capturing those games that follow a significant deviation from a team's expected performance (like your example WP < 50 and streak >= 3). There is likely an interaction between streak + a team's WP in predicting outcome such that the more above 0 their streak is and the lower their WP is, the higher the likelihood that the other team will win, especially if the other team has has a winning/better record...
WP < 48 and streak >= 3 and o:WP>55 and ADLast edited by emceeaye; 12-04-14, 10:22 AM.Comment -
Mostly dogs today for me after tinkering with spreadsheet trends to more closely match today's situations:
Hurricanes ml
Senators ml
Predators ml
Lightning/sabres over 5.5
Bruins ml
Devils ml
Islanders/Senators under 5.5
Maple leafs/Devils under 5.5
GL todayComment -
Lightning ML
Here are two queries with super strong support for Lightning ML today:
H and month=12 and o:rest > 0 and tA(assists) >= oA(assists) and tA(power play goals/oenalties) >= oA(power play goals/oenalties) and tA(shots on goal) >= oA(shots on goal) + 3 and p:shots on goal >= 26 and pp:shots on goal >= 25 and ppp:shots on goal >= 25 and tA(o:goals) <= 3.23 and season >= 2010
p:HW and po:goals=0 and WP>50 and o:WP<50 and season>=2009 and game number<42
But since its so juiced we can use the following queries (offshoots of each other) with crappy/not high sample size records but high ROIs to offset it perhaps:
total=5 and p:AD and line>145 and line<190 and month=12
total=5 and p:AD and line>145 and line<190
total=5 and p:AD and line>145 and line<190 and day=Saturday
total=5 and p:ADL and line>145 and line<190 and day=Saturday
...maybe try Lightning ML and Sabres (1.5+)
by the way, same situation as above with Kings and Flyers...Kings heavily favored in some queries while some queries have Flyers favored with high ROIs and crappy-ish records.Last edited by emceeaye; 12-06-14, 04:52 PM.Comment -
I rode the Lightning into the gutter with you tonight emceeaye. Not too often trends point us to that much juice, so it stings a little to get burned on it. At least I had the under to soften the blow a little.Comment -
It really was a bummer, Cutler... I also had the under thankfully. I've been experimenting more with picking the juiced team if more than 2 queries support it and then the other side PL if more than 2 queries support it, to see if we can win both sides and in what situations that strategy works...and if it doesn't work, at least the blow is softened.Comment -
One of the more solid trends is on the Oilers tonight (#48):
tS(W, N=3) = 3.0 and oS(W, N=3) = 0.0 and day!=Saturday and P:margin>-2 and rest - o:rest!=-1
250 play sample with a 28% ROI. 20-30 plays/year going back to 2006 and it's been super consistent. Can't ask for much more than that. Only 1-4 so far this year tho, so hopefully a good spot.Comment -
Good call... Problem is that there were a few other strong queries pointing in the opposite direction, albeit with a significantly smaller sample size than the one you posted... I just layed off of it. I guess after 11 straight losses, they were finally due for a win
Either way, good job.Last edited by emceeaye; 12-08-14, 12:46 PM.Comment -
Yeah, this is really the basic challenge when we have this many good queries floating around. So many conflicting trends can make you just throw your hands up. I'm in the process of re-booting the back test I did last week just to see where someone would be if they followed the unique plays from the best trends, (according to my eye anyway: long term, large sample size, good ROI...not sure if that's the best way, but at least seems to make sense).Comment -
Hockey plays are up. I'm really liking the two total plays tonight (Rangers U5.5, Devils O5). Multiple solid trends are on them, and totals have not been too good recently, so I think a correction is due.Comment -
I like Rangers u5.5 as well, the most of all the nhl plays today in fact. Despite the Hurricanes/Devils over suggested by one of the queries, one of the Devils leading scorers, Adam Cammalerri, is not in lineup today, which is a variable that could make it go lower than usual...Comment
