Weekend Football Review (9/27/10)
Monday, September 27 at 3:08 PM ET
I will always recap the weekend in football each Monday afternoon. It is very important to us to be transparent and to honest about our picks.
The Football Numbers (after College Football Week 4 and NFL Week 3):
- ATS Locks of the Week: 2-0 (Stanford covered by 18.5, Cincinnati Bengals covered by 10)
- All-Time NFL ATS Locks of the Week: 52-11-2
- Year-to-Date Daily Top ATS Plays: 14-4
- Paul's Picks ATS Week: 6-1
- YTD Paul's Picks ATS: 77.8%
- YTD Picks 60%+: 66.3%
- YTD ATS All Games: 56.7%
- YTD O/U All Games: 54.8%
- YTD SU: 84.3%
Notes on Money Management
Before we get into the college best and worst, here are some notes on money management that came out of customer support questions this week:
Question 1: My interpretation of your data is that you give the average scores and win frequency from your simulations in your SU picks, and apply win frequency and score distributions in your ATS picks? Also, I read in your blog that the PM does a better job at predicting NFL outcomes, and based on current results, better at game spread picks vs. totals. Am I incorrect? Or are these standard swings?
Answer: To come up with our picks, we literally simulate every game (50,000 times). From that data, we can tell how often a team wins (scores more points than the opponent), but we can also tell how often a team covers the line the same way (favorite wins by more than line). That's how we come up with the percentages. When you evaluate the games in the Customizable Predictalator, it is actually playing the game at that time. I'm not sure how these compare with historical push rates, but I think our approach is the best way to to do it.
What I referenced in the blog is that our ATS numbers have actually been historically better in the NFL than in college. The reason for this is that we have more data on NFL teams and players. However, so does everyone else. I think our approach is the best possible, but there are still a lot of variables each off-season that can cause error (look at the turnover in Seattle for instance - it's almost as much art as science to figure a team like that out when there are so many moving parts). So early in the year, when the public has not figured out college teams yet, we do really well against the market.
In the NFL, we do our best from about a quarter of the year on. The public spends the first few weeks figuring teams out. Unfortunately, sometimes, so do we. The general public makes up its mind by about Week 2 or 3 in the year. That's when our performance has traditionally taken off because the public does not adjust every week like we do.
Of course, comparing the top 16 picks in college against our 16 NFL picks should always favor college because we should have more confidence in those picks. On the aggregate though, we should get better over time and do better overall in the NFL than in college.
It is also true to state that our ATS numbers are generally better than our O/U numbers. In the NFL, we rarely have a strong opinion on the total.
Question 2: What do you recommend that bettors do to maximize the prediction machine? Should I only play the picks that are over 60%. I was thinking that by betting them all, my odds would be better than by picking only a few of them.
There is a fine line to walk between diversification and optimization. Playing every playable game can minimize your risk because whether you win or lose for the week depends on more than just one or a few outcomes going right, but it will also minimize your accuracy because it brings picks with lesser and lesser confidence into the mix.
Everyone approaches risk different - some people love to play a lot of games, others love to put big money on one or two games - and it's easy to point to records in certain situations after the fact, but I can tell you how we would recommend utilizing our information.
Most importantly, use the Play Value Calculator, especially when it comes to the size of bets related to confidence and other daily plays. PVC "normal bet" values should change as your bankroll changes, meaning that you can wager more when you have more and less when you less in a normal setting. How that value is altered for each game depends on how likely we think that is to cover and how much of your bankroll is tied up elsewhere. Daily plays allows the calculator to account for the overall percentage of your bankroll in play at any time and adjust for each play accordingly. Right now, the cap is at 16 games - for a reason. Every week is different, but we have yet to see value in playing more than 16 games at any given time.
Next, when possible/if necessary, follow the guidelines of the PVC in the Customizable Predictalator. This will allow you to have the most relevant information. Every game that is 60%+ on our picks page may not 60%+ at current lines.
Keeping those ideas in mind, Paul's Picks - our top confidence picks with extra analysis, notes of interest and any words of warning - are a great place to start. This is where we have had the most success as well - as expected because we focus on our top three Saturday and Sunday picks by confidence ATS. Those who purchase on a week-to-week basis or who prefer to diversify/enjoy having action on more than a few games, can easily play more games, but the risk/reward is different every week. Picks above 60% confidence offer a good approach as well. This seems to be a fair level of assumed risk vs. value for most.
Remember that the confidence is what it is. If every one of our picks each week came in covering exactly 60%, at the end of the season, we would expect to be exactly 60% accurate on those games. There would be no way to know which game to play and which not to. You could play them all and have the same expectations as playing one. In that scenario though, the more games played, the closer the actual running accuracy would be to 60%. After your first game, you would be 100% or 0%. After 30 games or more, you should pretty close to 60% accurate.
In a world where not every pick is created equally, marginal value matters. If there are 3-4 solid plays on one day or in one timeslot and then a big gap before the next pick, you've found your stopping point. If there are ten picks that are all about the same confidence, you can play them all or just play one and you would have the same expectation, just less or more risk depending on the number of games played.
College Best and Worst:
Best Wins: It was another solid week for the Predictalator, particularly for Paul's Picks and for Over/Unders with high confidence. Our Lock of the Week, Stanford -4.5 @ Notre Dame winning by 23 had to be our most gratifying win. I wonder what this line would have looked like 20 years ago (with the same caliber of teams facing each other)? It probably would have favored Notre Dame even more. Back then, Notre Dame playing against a West Coast team with little history that had yet to come East may have warranted a field goal line favoring the Irish. Even today, when almost every game is on TV, the sharps can affect the lines in so many ways and data on every team is readily available, it was one of the most exploitable lines we had seen to-date. The Harbaughs seem to know what they are doing. Stanford is a tremendous team, easily ranking in the top ten in the country. Now everyone knows it. The Predictalator knew that a few weeks ago... Temple +15 @ Penn State was great as well. We don't really love Temple, but we just could not see how enough points would be scored in this game for the Nittany Lions to cover. What we predicted as a 25-14 Penn State win became a 22-13 win. ATS, SU and O/U in one on some of our strongest overall opinions. I'll take it... Oregon is pretty good too. Alabama, Stanford and Oregon have been the most impressive teams so far this season. With the Cardinal and Ducks as well as Arizona, Cal, USC and even a UCLA team that just destroyed the Big 12's best, the gap between the Pac-10 and the Big Ten behind the SEC as the second-best college conference is very narrow. No other conference currently comes close to those three (the Big 12 should be more worried about the Mountain West passing it than being in the discussion with the Pac-10, Big Ten and SEC)... Auburn and NC State were two teams that the Predictalator was very high on that produced great wins for them and us... The line was just two big for West Virginia. An 8.5 point line in a defensive battle is like a 13 -point line in a normal college game... Nevada is good - better than Oregon State. Boise State has to go to Reno in November... SMU's ATS win makes us 23-3-2 all-time in Thursday and Friday night ATS and O/U college plays. I don't have a great reason for that... Four of the six totals that were 60%+ to cover did - and most did with ease.
Toughest Losses: We only missed one Paul's Pick, but we'll get to that in a second. Arkansas State @ Troy provided the toughest loss(es) on the week in college football. With the side (Troy -11) and total (OVER 63.5) both about 60% to cover, that was a big game for us. Troy won 35-28 to miss both picks by a score (any score to get the over). A scoreless third quarter that included a Troy interception return deep into Arkansas State territory that was called back on a penalty and a goal-line stand by the Red Wolves, who stuffed the Trojans twice (on third and fourth downs) from the one-yard line... Boise State, the lone Paul's Pick to miss ATS in either sport this week seemed to almost take it easy on Oregon State. The Broncos dominated, out- gaining the Beavers by 232 yards overall, 4.4 yards-per-pass and 2.5 yards-per-rush. Unfortunately, the usually disciplined Broncos also gave up an impressive punt return for a touchdown to James Rodgers, committed eight penalties and turned the ball over once (to Oregon States' none). And still, Boise State had an opportunity to rout Oregon State by more than 13, but decided to down the ball at the end of the game instead. Nobility would be easier to applaud with a cover. Doesn't Chris Petersen understand the Jim Harbaugh rules on covering to make the alumni happy?... The outcome of the Fresno State - Mississippi game may never make sense to me... How is it that Air Force can put up big points on everyone but Wyoming (that includes a 10-0 game last year)? The Falcons are better than that... Growing up, I'm not sure that I ever envisioned myself being most interested in watching an FIU/Maryland game on ESPNU, while most Big Ten and ACC powers played in the noon ET games on a college football Saturday - but I was glued. Maryland figured out FIU's defense much better than Rutgers or Texas A&M were able to. That was a tough side and total loss from the early games...
Most intriguing games of Week 5: Texas A&M @ Oklahoma State (Thursday), Miami (FL) @ Clemson, Temple @ Army (really), Texas vs. Oklahoma (not as much as usual though), Wisconsin @ Michigan State, Virginia Tech @ NC State, Florida @ Alabama, Stanford @ Oregon and Penn State @ Iowa
NFL Best and Worst:
Best Wins: The Paul's Picks went undefeated so I have plenty more to note on the positive side than negative this week. Our strongest opinions of the season - six picks with 60%+ confidence - led to our strongest week. We hit four of those picks, covering by an average of 12.4 points per pick. That's the way it should typically be from now on. Every week is different, but it's good to see some "green" - in the picks and in the wallet. Even though Cincinnati -3 was our Lock of the Week and covered easily to extend our Locks of the Week to 3-0 in the NFL this year, 6-1 for football in 2010 and 55-12-2 all-time, Pittsburgh -2.5 may have been our best win of the weekend. That game was the one game not included on the NFL Sunday Ticket Game Mix channel - which was great because I never really had to watch it after the Steelers took a commanding lead. The betting public dramatically underestimated the role that playing the Browns and Panthers had in Tampa Bay's start. It also underestimated Charlie Batch. That was fun... Remember those Panthers? They helped us cover our Lock by ten points. The Bengals offense had a pretty uninspiring day, with 305 yards on 75 playes (just over four yards-per-play) and the team still won by 13 points. If Jimmy Clausen remains the quarterback in Carolina, it could be a very long season for the Panthers... Was (former) 49ers offensive coordinator Jimmy Raye fired before of after the team's completely unnecessary last-second touchdown against the Chiefs? I need to know to whom I address the thank you note. Either way, thank you to Mike Singletary for channeling his inner-Jimbo Fisher on the late cover... The score at the time (and at the end of the game) might have been surprising, but it was fairly obvious that St. Louis - Washington was going over in the first six minutes of that game... Michael Vick is having a pretty awesome season. Not only were the Eagles -3 a great play in Jacksonville, Vick and DeSean Jackson were two of our top fantasy plays... Even when we didn't have them winning outright more often than not, I like to see the underdogs we are picking win outright. Thank you Seattle, New York Jets and Atlanta (also known as every underdog we were picking ATS)... Tennesee @ NYG and Oakland @ Arizona are my new favorite kinds of games. We didn't have a strong enough ATS opinion on either team to play, but we had the over in both. Rooting for points is fun. Those teams helped us out by scoring them.
Toughest Losses: We did not get any of Paul's Picks wrong ATS, but it was another absolutely brutal week for the free pick. Since we get to choose the free pick each week, be sure that we will no longer choose a game involving the Detroit Lions. Last week, the Lions had a late, backdoor cover against Philadelphia to cost us the Eagles -5. This week, Shaun Hill threw two interceptions in the last three minutes of the game to cost us Lions +11. They may not ever be one of our top three picks, but we understand the importance of getting our free picks right. To-date, we have had some incredibly poor luck with those. Hopefully we can give out another 60% pick this week that actually covers... The worst possible scenario happened when the Patriots beat the Bills by eight. Our (soon to be released) Survivor pool optimizer had us picking the Ravens against the Browns. Our picks had the Patriots covering -14. If the Patriots lost outright, great - half the remaining survivors (is that redundant?) in our pools would be gone. Otherwise, we needed a cover. Future All-Pro tandem Ryan Fitzpatrick to Steve Johnson connected on a 37-yard TD with about four minutes left, but then couldn't hold onto the football for more than one play the rest of the game to give us a tough blow and make us 0-3 with the Bill ATS so far. Not cool guys... We knew Dallas was good. At home, Houston just seemed a little better. It was not a strong ATS pick, but everyone asked me about that game last week. Right now, Dallas is more likely to make the playoffs than Houston. Dallas is projected to win 9.1 games, which would rank second behind the Bears among Wild Card contenders after likely division winners Philadelphia, Atlanta, Green Bay and a too-close-to-call-doesn't-really-matter-anyways NFC West team. Houston is projected to win 9.6 games, which would rank fourth in the Wild Card behind the Ravens, Patriots and Bengals after division winners the Jets, Colts, Steelers and Chargers.
Most intriguing games of Week 4: BAL @ PIT, SEA @ STL, CHI @ NYG, NE @ MIA and WAS @ PHI.
Got a question on sports betting, bankroll management, coaching strategy, fantasy football or anything else? Contact Paul