Methodology: Taking a look at teams' home and away defense and offense efficiencies (based off of dean oliver's possession calculation) and coming up with a projection based on prior efficiencies.
For example:
Boston at Cleveland.
Take average of Boston away dEFF and Cleveland home oEFF for Cleveland score (taking into account average of Boston away possessions and Cleveland home possessions).
Same thing for Boston score.
Caveats:
It's still early on in the seasons, so sample size is incredibly small
If a player gets hurt, the system does not take into account lost productivity on both the offensive and defensive end.
There is nothing in the model yet that deals with current trends (I'd like to try and bake this in once the season progresses).
With that in mind, I'll list the games where my estimate is very far off from Vegas' line, in order to see how some of these odd picks play out. I'll put the threshold of an estimate at a difference of +/-10 from vegas.
I will not be putting money down on most of these bets for now, since the season is young and more data is needed. This is for informational purposes only, and possibly board entertainment
Odds were taken from wagertracker.com, and data was culled from basketballreference.com
11/14
OKC, SAS - 184.5
Projection - OKC - 96, SAS - 103 - Total - 199: OVER 14.5
Model's thinking -Spurs haven't been playing good defense either on the road or at home, but they've played very well offensively at home. Those numbers are really pulling the forecast up, regardless of OKC's below average road offense and above average road defense.
DET, WAS - 186
Projection: DET - 98, WAS -100 Total - 198: OVER 12
This one boils down to Detroit's below average defense on the road so far. It seems like they play MUCH better defense at home. The away games against Orlando and Toronto are pulling this one up.
A nice start, as both games manage to go over. I'll take it with a grain of salt, since the model missed most of the other games, and we're still collecting data.
11/15
LAC, OKC - 179
Prediction: LAC - 94, OKC - 97 - Total - 191: OVER 12
The biggest thing is that OKC plays a very slow pace at home, with about 86.5 possessions a game. However, they've also played some slow teams as home to start. The Clips are the opposite, where they play a bit faster on the road, but they've only played fast teams. I actually wouldn't be surprised if this one played under and chalked it up to low sample size. Clips and Thunder also play pretty solid defense in the situation presented.
Not sure what the lack of Eric Gordon does for this game, but of the 3 games he's been out, they've put in two stinkers on offense, but two stinkers on defense as well.
another victory, as OKC and LAC combine for 194 points. This spread was raised all the way up to 184.5 I believe, but it still wasn't enough.
Currently standing at 3-0
Tomorrow's action doesn't present any games that stick out, but I'll post once I run tonight's data. The Mavs game would be the only one that might stick out since they're the only team that's playing tomorrow that would have an overnight update. Right now, they're only looking at 8 over the O/U, but their performance tonight might bump it up (or down).
No plays today, although the system is picking the Over for all the games - However, it's not crossing my threshold for a difference of 10. I might tease all 3 games and take the over if I'm looking for a quick betting fix.
If anyone's interested, the Blazer's / Hawk's is now in play with the total line sitting at 187.
POR - 97, ATL - 101 - TOTAL - 198: OVER 11
Atlanta has played lights out offensively at home, while Portland has been shutting down teams on the Road. Granted, those teams they've been shutting down on their road trip have not been offensive juggernauts. This one comes down to the pace of the game, and if you believe the home team has some say over that statistic, then this one should slide past the total line. However, in the prior meeting, Portland was able to effectively slow this game down, but that game was in Oregon, and the Hawks still put up an effective offensive efficiency.
Another win, but it feels like a cheap one, as the total gets a healthy bump from the OT. The Blazers were able to force the pace on the Hawks last night, and made them play their slowest game of the season. Both teams played pretty well offensively.
Onto today, where some of these plays don't past the smell test, but I'll post them anyway.
11/17 GSW, CLE - TOTAL 215
GSW - 99, CLE - 105 - TOTAL:204 - UNDER:11
Cleveland loves to slow things down at home, and does a pretty good job of defending the ball as well. The Warriors are the Warriors, so we know they'll only play 48 minutes on the offensive side of the court. This one will come down to the pace of the game, and I'll side with the home team again.
Interestingly, the predicted spread on this one is a lot smaller than the 15 / 15.5 that's currently out there. If you wanted to take a risk, a tease of the under and the Warriors could be quite the nail biter. Beware the recent trades, because the model doesn't project the loss of SJax, but personally, I think GSWs minutes are perfectly capable in another's hands.
PHO, HOU - TOTAL: 220.5
PHO - 102, HOU 106 - TOTAL 208: UNDER 12.5
Another nail biter - Both these teams offer high octane offenses, but for this game to hit the over, both teams either need to play above average offense AND below average defense, or the possessions have to hit 100 - both teams have shown they can do either of those things, so an over isn't out of the question.
Made a tweak in order to average possessions between the two teams when they play, so one game has one possession number instead of two - this tweak brings the Cavs game to a difference of -9. Still, the spread is at a difference of -9.5, but there's the caveat of the recent trade and injury that the GSW has faced. I'll tease the spread and the under for a tiny amount if I make a bet today on this one.
1 Win and 1 Loss last night brings the record to 5-1. The Under in that Phoenix / Houston game just made it, while the Cavs and Warriors went well over. Cleveland manged to play its worst defensive game of the year, but also its 2nd best offensive one. These things will happen early in the season.
EDIT: Finished updating my formula - It was the Cavs BEST offensive game of the year AND The WORST defensive game. These things certainly do happen against the Warriors.
11/18 NYK - IND - TOTAL: 214
NYK - 96 - IND -103 - TOTAL - 199 - UNDER: 15
DET - POR - TOTAL: 179
DET - 95 - POR - 99 - TOTAL - 194 - OVER: 15
Two potential games with some big differences.
The model only has 3 away games for the Knicks (beware), but so far, they haven't played efficient offense, but have certainly played bad defense. The Pacers play an average level of offense and above average defense at home. This one could beat the over, as it looks like the Pacers can run a fast game at Home and the Knicks obviously try to do that anywhere.
Detroit is no longer a defensive juggernaut, having put in 3 bad defensive games in the past 3 games. They have become a rather efficient offense though. Portland has been shooting the ball just as well on the road as at home, and plays slightly worse defense at home, although it's still good defense. Look for both teams to be comfortable playing a slow tempo style and put up points despite that lack of possessions.
Record moves up to 6-2 following a 1-1 evening last night. The Knicks actually played above average offense AND defense, but they were still off by a one total point. Again, there were only 3 road games on the Knicks, so we're still developing a picture of them.
The Pistons and Blazers played a very slow game, as predicted. However, the Blazers defense was able to force the Piston to play their worst offensive road game, and second worst offensive output on the season. The Pistons also played some solid D, posting their second best road defensive performance after getting torched 4 or 5 games.
11/19
PHO - NO - TOTAL: 217
PHO - 104, NO - 103, TOTAL - 207: UNDER 10
Tread carefully, as NO has had a lot of disruption lately that the formulas aren't taking fully into account. However, they did just play the CLIPPERS well at home, and played the Hawks about as well as you could expect. They also play very slow at home, while Phoenix plays very fast on the road. I'm not quite convinced with this one, as I expect Phoenix will probably push the pace. Both teams have above average offenses and below average defenses, but that's accounted for in the prediction.
A nice win, as the Hornets forced the suns into playing a slow enough game to land under. Now up to 7-2
Tonight there are two games 11/20
DEN - LAC - TOTAL - 208
DEN - 100 - LAC 96 - TOTAL - 196: UNDER 12
This game could be slower paced than you think, however both teams have above average variation in possession amounts. The Clips have played very poor offense at home, which looks like the culprit that drags the total down for this one. There's not much else to glean from this matchup, and, if you think JR Smith is a key factor to the nuggets, then he hasn't factored much into this calculation.
CHA - MIL - TOTAL: 179
CHA - 92 - MIL 101 - TOTAL - 193: OVER 14
This is basically a flip of the coin in my opinion, because Charlotte is very erratic on offense and Milwaukee can play shutdown defense or give up a ton of points. Regardless, Miwaukee has been playing very effective offense the past 4 games and can put up a big total despite the lack of possessions. We'll see what another game with Steven Jackson brings the Bobcats. This could turn out like either one of my Pistons picks, but if you're going to tease these two, might as well have a combo of over / under.
EDIT: With no Bogut for Milwaukee tonight, I probably wouldn't trust the numbers, but I'll leave it in here anyway.
Last edited by djiddish98; 11-20-09 at 01:11 PM.
Reason: Update to bucks game
Might as well add another one - I've been fiddling around with some four factor tables to see if they can shine any additional light
WAS - OKC - TOTAL: 196
WAS -87 - OKC - 98 - TOTAL - 185: UNDER 11
Washington has not played good offense on the road. Look for the thunder's ability to turn the ball over to disrupt the Wiz. The Wiz HAVE played pretty solid defense on the road, and this should disrupt the Thunder's shooting. I don't expect either team to really push the pace in this one, so the under seems attainable.
Ended up going 2-1 last night, but I think I'm going to take a break till early December with these picks to allow for more data to develop and to further hone the model. That OKC - WAS game was a classic case of two teams playing above their averages. It was the fastest pace either team had played, and both teams played awful defense / amazing offense. It certainly can happen when you're looking at averages.
I didn't like the way my formula was accounting for recent performance, home / away performance, and total performance. It was a little arbitrary with nothing to back it up. I've tried to randomize it a little bit. I'll start tracking this new system (which is largely based on the old methodology) in here. Here are some quick hits to test the new system.
I feel like last night was bad for the math systems. The suns played one of the top 5 or 6 most efficient offensive games on the year. Fortunately, it wasn't enough to go over, which ended up in a push. However, they did blow Detroit out of the water. OKC played on of their worst offensive games on the year as well (if you adjust for opponent, it was their worst). Or perhaps the Lakers played their best raw defensive game on that year. Either way, it was another spread loss, but the under hit fine.
Combined with the Hornets hit, the new system was 2-2-1, which brings the composite system grade up to 11-5-1.
Revised the formulas some more - here are the updates.
I'll include all the picks made today in the record
11/23 update
MIL / SAS - MIL +9 - 70% chance
MIN / LAC - LAC -7 - 62% chance
MIL / SAS - Under 192 - 83% chance! (Ha, we'll see about that)
I've been fiddling around a lot with how to incorporate recent games (within the formula and how many games to include). Right now, the latest system is looking at each teams last 3 games based on location (home / away) when factoring in recent performance.
Clearly, these %s don't mean anything that the system spits out - 0-5 last night, rounding things out to a 11-10-1. I'm going to keep tweaking and continue posting, but clearly this model is still in the developmental stages.
I've tried several different models, with not much success - the past week has been brutal for predictions based on past trends (see last night's Denver / Minn game), but I'd like to track my latest model to see how things shake out.
Away Team Home Team Spread Total AwayScore HomeScore %Away Cover %HomeCover %Under %Over
Philadelphia 76ers Dallas Mavericks 10.5 200 104.099 105.598 0.676 0.324 0.379 0.606
Indiana Pacers Golden State Warriors 2.5 228 110.746 121.035 0.374 0.626 0.45 0.537
Chicago Bulls Milwaukee Bucks 3 196 93.754 104.193 0.387 0.6 0.501 0.499
Memphis Grizzlies Utah Jazz 10 208 103.306 117.396 0.415 0.57 0.312 0.688
If interested, to make sense of all this data:
First two columns - team names
3rd column - spread at the time of the simulation
4th column - total at the time of the simulation
5th - projected away score (first team in list)
6th - projected home score (second team in list)
7th - % of time away team covers
8th - % of time home team covers
9th - % of time under
10th - % of time over
So for tonight, the model likes Philly to cover 67.6% of the time - i'd count that as a pick.
to summarize, here are the system picks (anything over 60%).
PHI: +10.5 - 68%
PHI / DAL: Over 200 - 61%
GSW: -2.5 - 63%
MIL: -3 - 60%
MEM / UTH: Over 208 - 69%
Again, knowing my luck lately, I'll bet all of these and lose. This is more of a tracking thread for the time being.
Last edited by djiddish98; 11-30-09 at 08:19 AM.
Reason: messing with formatting
Some sense of normalcy last night - The model goes 4-1 and misses that MIL pick by a point. I'll keep my old record intact, which bumps picks on the year up to 15-11-1
Away Team Home Team Spread Total Away Score Home Score % Away Cover % Home Cover % Under % Over
Boston Celtics Charlotte Bobcats -3.5 179 86.022 100.595 0.204 0.796 0.39 0.61
Golden State Warriors Denver Nuggets 13 234 116.177 122.692 0.571 0.414 0.439 0.561
New Orleans Hornets Los Angeles Lakers 13.5 206 96.644 110.086 0.515 0.485 0.482 0.494
Phoenix Suns New York Knickerbockers -7 225 119.74 103.703 0.65 0.341 0.538 0.442
Miami Heat Portland Trail Blazers 7 186 97.787 92.43 0.714 0.271 0.417 0.583
Washington Wizards Toronto Raptors 7 212 102.708 115.347 0.404 0.581 0.422 0.566
The model really likes Charlotte to cover in this one - they've been playing very strong defensively at home all season and now have some semblances of an offense (perhaps the Steven Jackson trade is the result of this?). I'm sure they'll get even more points towards tip off, but Boston (while variable) has been mediocre for most of the year away. It also likes the over in this one a little bit - I'll add it to my picks, but I'm scared to bet it.
The Suns / Knick pick seems pretty self explanatory.
The Miami / Blazers games is interesting - Miami has played pretty well on the road, and the blazers are really only an average team at home. Their last 3 games at home have not featured much defense. The model actually likes Miami to win - I don't bet money lines, but this one could be a decent move for the daring.
Last edited by djiddish98; 12-01-09 at 06:53 AM.
Reason: formatting
Model definitely missed on the Celtics and the Knicks. That's what happens when you go with the statistical trends and ignore any outside factors that might influence play. However, the Heat game and the Over hit, so last night was 2-2. Total is 17-13-1
Away Team Home Team Spread Total Away Score Home Score % Away Cover % Home Cover % Under % Over
Toronto Raptors Atlanta Hawks 9.5 212 95.839 121.206 0.316 0.684 0.459 0.541
Detroit Pistons Chicago Bulls 3 195 87.013 83.63 0.594 0.393 0.812 0.178
Phoenix Suns Cleveland Cavaliers 8 213 103.118 108.218 0.56 0.426 0.532 0.454
Houston Rockets Los Angeles Clippers 3 195 93.337 85.607 0.786 0.191 0.807 0.172
Memphis Grizzlies Minnesota Timberwolves 3 195 100.915 102.869 0.497 0.471 0.361 0.624
Dallas Mavericks New Jersey Nets -7.5 189 97.381 83.085 0.616 0.384 0.659 0.341
Philadelphia 76ers Oklahoma City Thunder 6 198 102.933 109.199 0.486 0.499 0.327 0.665
New York Knickerbockers Orlando Magic 13.5 210 104.813 108.769 0.642 0.358 0.477 0.515
Indiana Pacers Sacramento Kings 2.5 217 106.596 122.956 0.167 0.833 0.294 0.695
Milwaukee Bucks Washington Wizards 3 196 95.541 100.013 0.479 0.504 0.521 0.463
Lots of plays today (too many in my opinion).
The raptors have been playing like crap lately away from home. I wouldn't let Atlanta's last stinker worry anyone just yet.
I would probably throw out the Bulls game - the "recent" data that I'm using is 18 days old due to their long road trip (where they stopped playing defense). Perhaps they'll be inspired at home.
The clips have shown signs of defense lately at home, and the Rockets have been flashing their own D on the road. They should still have little trouble with the clips.
Lastly, I'll address the Indiana / Sac game - Indiana has not played that well on the road lately, posting below average offense and defense. Sac town has been blowing by teams at home the past 3 games and shouldn't have trouble with the Pacers coming to town.
Last edited by djiddish98; 12-02-09 at 06:14 AM.
Reason: formatting
I forgot that I have a default spread / total of 3 and 195 if no spread is present. Some lines broke late. I've also been trying a different data source. Here are some updates. The Pistons / Bulls are still expected to go under even at 183 as well as the rockets. The actual Grizz total is more accurate than my fill in
Away Team Home Team Spread Total Away Score Home Score % Away Cover % Home Cover % Under % Over
Toronto Raptors Atlanta Hawks 9 211 94.069 123.087 0.289 0.696 0.445 0.547
Detroit Pistons Chicago Bulls 7.5 183 82.592 82.721 0.619 0.381 0.773 0.217
Phoenix Suns Cleveland Cavaliers 7.5 213 102.521 111.361 0.518 0.482 0.522 0.469
Houston Rockets Los Angeles Clippers -2 193 94.359 86.671 0.634 0.344 0.752 0.233
Memphis Grizzlies Minnesota Timberwolves -3 202 103.02 99.907 0.47 0.508 0.515 0.469
Dallas Mavericks New Jersey Nets -7.5 190 95.616 81.423 0.632 0.368 0.694 0.306
Philadelphia 76ers Oklahoma City Thunder 6 198 98.674 109.012 0.406 0.576 0.358 0.629
New York Knickerbockers Orlando Magic 13 208 100.764 112.277 0.515 0.474 0.421 0.567
Indiana Pacers Sacramento Kings 1.5 216 102.95 125.271 0.107 0.893 0.303 0.697
Milwaukee Bucks Washington Wizards 3 198 91.425 100.148 0.43 0.564 0.595 0.392
Last edited by djiddish98; 12-02-09 at 01:41 PM.
Reason: formatting
ATL Spread - W
Piston Spread - W (at 7.5)
Rockets Spread - W
Rockets Clips Under - W
Mavs Spread - W
Mavs Under - L
Kings Spread - W
Kings Over - Push
I'll add the Knicks from the first post as well - which was an L.
Forgot to Highlight the 76ers - OKC over, which was a W.
Turns into 7-2-1 for the night. Teams basically played within their averages, which always makes the model shine. These days don't typically seem like the norm in the NBA though. Total record is sitting at 24-15-2.
Away Team Home Team Spread Total Away Score Home Score % Away Cover % Home Cover % Under % Over
Miami Heat Denver Nuggets 8.5 209 109.262 108.534 0.635 0.365 0.387 0.594
Houston Rockets Golden State Warriors -2 222 118.203 107.498 0.606 0.385 0.454 0.537
Boston Celtics San Antonio Spurs 1.5 187.5 105.851 104.852 0.522 0.478 0.293 0.707
Still tweaking with things. System likes Miami, probably due to the fact that the timberwolves game is weighing the Nuggets down a little. To Miami's credit, they've been playing well on the road recently. When I run the games more than once, the %s can change depending on a couple factors. This MIA / DEN game is close enough to over where I'll bold that one as well.
Houston has also played well on the road lately and just edges over the 60% mark. Golden State has had some strong home showings as well as some duds mixed in. Interesting to see how this plays out.
Finally, the Celtics and Spurs are projected to go over. This is undoubtedly helped by the Celts thrashing over the Bobcats the other day, who previously had the best home defense. The Celts offense is rated higher than normal because of it. The Spurs defense has been playing very strong at home as of late (excluding their game against the Bucks). The 70% chance of over the system projects is definitely lofty, and I wouldn't be surprised to see this one fall under, but we'll play it anyway.
I did ten runs of the system to work out run to run differences - highlighted plays are > 60%. I'll keep the highlighted plays above in the record as well just for the fun of it.
Last edited by djiddish98; 12-03-09 at 04:58 PM.
Reason: formatting
The Celts and Spurs game looks like a defensive minded one from the get go - I was hoping for the score to pickup in the second half, but the end of the 4th was a clunker for points. Not surprised in the least on this one, as I expected the data to inflate the score. A tough L though. It should be noted that both teams played their worst offense and best defense with respect to location on the year.
The Heat game also let me down, but the over hit by a point. I'm looking at the Heat's past data - they don't have too many away games to bank on, and their first two defensive games were probably outliers. The weighting should account for this by now, but it might explain why the system picked them. Denver has been a bit hot and cold offensively at home, but last night was their 3rd best offensive outing. Their defense is surprisingly strong at home as well, save one game.
The Rockets pushed, so Thursday's record is 1-2-1 with a total of 25-17-3. Will post the plays once my new data set comes up.
Away Team Home Team Average of Spread Average of Total Average of % Away Cover Average of % Home Cover Average of % Under Average of % Over Average of Away Wins Average of Home Wins
12/4/2009
Boston Celtics / Oklahoma City Thunder -4 192.5 41.84% 56.66% 41.82% 58.18% 46.84% 52.16%
Miami Heat / Los Angeles Lakers 11.5 199 54.62% 45.38% 55.72% 42.66% 31.34% 66.94%
New York Knickerbockers / Atlanta Hawks 11 216.5 44.56% 54.06% 46.74% 53.26% 31.96% 66.94%
Chicago Bulls / Cleveland Cavaliers 12.5 190 43.90% 56.10% 54.46% 44.46% 26.88% 71.80%
Milwaukee Bucks / Detroit Pistons 3.5 186.5 49.00% 51.00% 65.20% 34.80% 42.28% 55.88%
Dallas Mavericks / Memphis Grizzlies -5.5 206 47.50% 52.50% 48.90% 50.04% 53.68% 44.92%
Charlotte Bobcats / New Jersey Nets -3.5 181 52.00% 48.00% 69.94% 28.90% 56.74% 41.66%
Minnesota Timberwolves / New Orleans Hornets 9 200.5 38.02% 60.64%65.74% 34.26% 26.74% 72.24%
Indiana Pacers / Utah Jazz 8 210 27.54% 70.20% 46.36% 52.24% 16.88% 81.86%
Toronto Raptors / Washington Wizards 3.5 215.5 43.74% 56.26% 47.18% 52.82% 38.36% 60.16%
I've added 2 more columns, which count up how many times the home team wins straight up, as well as the away team - I've highlighted games with a greater than 65% chance of home or away winning. There's a .86 correlation between the spread and the home win % for this batch of stats.
The model really likes Utah at home right now. The Pacers have yet to play a respectable game on the road, and Utah has been lighting it up at home the past 3 games. Even the 8 points the pacers get aren't enough. The Utah money line is also looking like the safest bet of the night (we'll see how that plays out).
Another game that sticks out is the Under on the Bobcats / Nets game - The Cats aren't as close to as good on the road as they are at home. And the Nets stink period (but they do have a projected 40% chance of winning tonight!) A warning - the Nets played their best offensive game of the year against the Mavs and also posted their worst defensive effort at home. But if you want to go outside the statistics, I imagine the Bobcats are pretty angry about the thrashing that Boston gave them, and will respond will lock down defense against a team that (despite being win-less) no one will take lightly, in fear of popping the Nets' cherry.
I had a big post for today's games, but my internet crapped out while posting. Last night was 2-2-1, for a total of 27-19-3.
Away Home Spread Total % Away Cover % Home Cover % Under % Over Away Wins Home Wins
Houston Rockets Portland Trail Blazers 5 190.5 60.04% 38.74% 51.04% 48.96% 52.80% 45.78%
Indiana Pacers Los Angeles Clippers 2.5 200.5 49.28% 50.72% 72.74% 27.26% 44.08% 54.68%
Toronto Raptors Chicago Bulls 6 205 45.24% 53.22% 72.76% 26.08% 36.28% 62.42%
Philadelphia 76ers Charlotte Bobcats 6 188 42.06% 56.38% 58.42% 40.24% 33.96% 64.84%
Atlanta Hawks Dallas Mavericks 4.5 204.5 48.92% 51.08% 59.80% 40.20% 42.16% 56.12%
Orlando Magic Golden State Warriors -9 221 51.04% 47.86% 55.30% 43.90% 60.08% 38.84%
Utah Jazz Minnesota Timberwolves -7.5 196.5 48.44% 51.56% 69.66% 30.34% 62.70% 35.30%
Sacramento Kings Phoenix Suns 11.5 225.5 42.46% 57.54% 56.94% 43.06% 24.60% 74.04%
Denver Nuggets San Antonio Spurs 4 206.5 37.92% 60.62% 50.78% 49.22% 32.66% 65.92%
Picks are
Rockets +5
Spurs -4
Pacers / Clipps - Under
Raptors / Bulls - Under
Jazz / Timberwolves - Under
Moneylines (I won't include these in my record)
Spurs, Suns - The Spurs might have the best ML to Win ratio on the night.
I feel like my over / unders have been very...predictive. i'll have to look at my possession calculations and see if something's up there. However, last night, the Nets forced the bobcats to play the most possessions they've had all year on the road, which juiced the score up. Same holds true with the Pistons playing their fastest game at home. These things certainly do happen.
DJiddish Bizarro Bets
Reply With Quote
