Monday, January 17 at 11:12 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. 8-0. This blog may end up fairly short because a) there are only five games to talk about (four NFL and the BCS National Championship game) b) I don't have too much to complain about and c) I don't want you to get sick of "8-0." Today is a great day, though. We're coming off a second straight undefeated NFL Playoff weekend. College and NBA basketball are/have been on TV all day(while I put basketball Predictalator/s through final tests). Mentions in ESPN Podcasts by Bill Simmons (around 47:30) and Chad Millman (entire interview). And, the industry is in the news with fascinating debate sparked by Billy Walters' Interview on 60 Minutes. (I'll discuss Behind the Bets podcast below, but will probably never get into Walters specifically because I don't know him and I don't think the 60 Minutes piece tells enough of the full story. Instead, we will continue to discuss general industry concepts and theories that directly pertain to conversations started after Walters' segment.)
Speaking of the Behind the Bets podcast, there are a couple points that I wanted to elaborate on/clear up. First of all, thanks to all that have listened (or just clicked on the link above and are listening to it as they read this). I am intrigued with and appreciate of the effort that Chad and ESPN are going through to shine a light on the intricacies of the industry, rather than to pretend to ignore it despite the significant percentage of the site's visitors that rely on their content to aid them in wagering decisions. As for some of the topics discussed in the podcast, as someone who covers the industry where everyone is trying unearth the "Holy Grail" number or system, I totally understand the line of questioning regarding the specific metrics that I find important. It is true, that there are several numbers that I often quote: (all strength-of- schedule-adjusted) yards-per-pass for and against, yards-per-rush for and against and turnover margin.
However, it doesn't really work that way. I use those numbers to supplement and better explain the results from the Predictalator. In truth, any direct competition is about matchups - using ones' strengths to exploit opponents' weaknesses. Everything is relative. Having done this professionally for seven years, I can look at a lot of specific numbers and get a good idea of our likely pick and confidence, but it is impossible for me to efficiently and manually compute everything that would be needed to make an accurate pick on a game without allowing technology like the Predictalator to account for all of the interactions of the players, coaches, fans/homefield advantage, weather, officials, ballpark effects, etc.
One of the best ways to exemplify the advantages of simulation is to comprehend results in situations where extremes meet (good vs. good, bad vs. bad, bad vs. good). The technique that I often use in simulation to compare interacting variables and come up with the probabilities of occurrences (against which we draw random numbers to determine if they occur in the specific situation) is commonly known as "log5 normalization." It's a nonsensical term in mathematics as it relates to anything else, but as coined and applied by Bill James, log5 normalization has strong applications in any comparison of interacting probabilities. The premise is fairly simple - in baseball (which we will use as the easiest sport to explain because of the propensity for one-on-one interactions - which is also the reason why baseball and its fans are so numbers-savvy), if a .320 hitter batted against a pitcher who gives up a .320 batting average, he's not going to hit .320; he's going to hit better than that because the hitter is well above average and the pitcher is well below average. Exactly how much better than .320 he hits is what log5 normalization determines. When good hitters face good pitchers of exactly the same degree (for what it's worth, for the nerds, "degree" is not linear, but it does fit to a distribution that has a mean and variance) they will hit exactly league average (-ish, this is independent of all other variables like HBP, BB, K, defense, ballpark, etc.). The same can be said of bad pitchers facing bad hitters of equal degree of "badness." (I'm not hiding the formula. It's pretty straight-forward and found in this presentation. I just find people stop reading when they see formulas.)
Anyway, that's just one interaction between two variables. And if the result of that hypothetical interacation is 0.262 and, at this step in the simulation, we are looking at whether or not a hit occurs, we randomly find a value between 0 and 1. If it's less than or equal to 0.262 - like 0.1 - the result is a hit. If the random number is greater than 0.262, like 0.5, it's not. That's how one decision is made for just one step of one play within the scope of a game. That's also why we can get different results every time we simulate a game. Each step can bring on a new set of circumstances that change the probabilities of events. Each play in every sport has tens (if not hundreds or even thousands) of steps that rely on interacting variables, some directly compared through log5 normalization, and others modifying the probabilities along the way. In the scope of a simulating one NFL game one time, we draw approximately 20,000 random numbers to handle all of the interactions...
Then we do that 49,999 more times and track all of the info we need (score, team and player stats) as the games simulate. At the end of the day, it's that output that really matters to you (and to me). It's how we're judged and how we all benefit from the information. So, while that's probably more technical than I needed to make it, I wanted to clarify that there is no magical "Holy Grail" stat or metric or quality that can be exploited to immediately win a high percentage of games. That being said, I feel we have are employing, simulation, which is the most intricate, all-encompassing and thus, accurate system for projecting future outcomes. Of course, others could do something similar and even use my explanation and presentation above to start down that path. And, there are always ways we attempt to improve the inputs and interactions in the simulation engine, some stick, others don't (but we learn just as much when they don't). But, for now (and the foreseeable future), I know that with our background, expertise and immense experience, the Predictalator provides the strongest model available for doing what we do. 8-0.
The Football Numbers (through the BCS National Championship and NFL Divisional Weekend):
ATS Locks of the Week: 1-0-1 (Green Bay +2.5 won 48-21 over Atlanta; Oregon +3 pushed against Auburn 22-19)
Football Year-to-Date ATS Locks of the Week: 26-12-1 (68%)
All-Time ATS Locks of the Week: 75-23-4 (77%)
YTD Daily Top ATS Plays: 71-36-1 (66%)
Paul's Picks ATS Week: 4-1
YTD Paul's Picks ATS: 95-63 (60%)
YTD ATS All Games: 56%
YTD O/U All Games: 54%
YTD SU (NFL and FBS vs FBS College): 72%
NFL Playoff Probabilities (updated after Wild Card Weekend)
The Predictalator uses current rosters and strength-of-schedule-adjusted team and player stats from the last 48 games to play, one play at a time, the NFL Playoffs bracket 50,000 times before it's actually played. For this analysis, we are tracking how likely a team is to win the Super Bowl. The playoffs are played all the way through individually, with the team that wins each game in that instance advancing.
With the New England Patriots' upset by the New York Jets means that the Pittsburgh Steelers are now our favorites to win the Super Bowl, bringing home the title 33.6% of the time. Just behind the Steelers are the Green Bay Packers at 33.5% to win the Super Bowl. A Packers vs. Steelers Super Bowl looks too close to call at this point (this may change depending on what happens this weekend) and both teams are about equally likely to get to the Super Bowl. Ultimately, the slight Steelers edge comes from the fact that the Chicago Bears are a slightly easier matchup for Pittsburgh than the New York Jets are for the Green Bay Packers. The Jets are currently 17.8% likely to win the Super Bowl, while Chicago is 15.1% likely to win the Super Bowl. Going into the weekend, the Bears were 14.1% favorites to win the Super Bowl, which illustrates both the relative bye that Chicago got against Seattle as well as the strength of the remaining teams (all significantly better than the Bears). Anyway, this means that the Super Bowl futures value is in Pittsburgh 2:1.
Interesting Super Bowl matchups and their relative likelihoods include (numbers subject to change with published picks later in the week): Most Likely - PIT vs. GB is 40.2% likely, Our Preseason Prediction - NYJ vs. GB is 23.6% likely, Biggest Blowout (or at least Highest Projected Line) - PIT vs. CHI is 22.8% likely and the Least Likely Super Bowl - NYJ vs. CHI is 13.4% likely.
NFL Super Bowl Odds (based on 2010-11 NFL Playoffs played 50,000 times)
New York Jets
NBA and college basketball Picks: With our recent successes and exposure, we have received a tremendous amount of interest in predictions beyond football. Our goal is to have basketball picks soon and then a full season of baseball before we get into football again next year. We have been testing the daily NBA and college basketball products for the last two weeks and are very happy with the results (I obviously can't claim any record or set expectations for likely unsustainable performance, but let's just say that our record on 60%+ ATS picks in college and NBA games so far is eerily similar to our performance in the NFL Playoffs so far). Watch for announcements and further information coming very soon.
College Best and Worst:
Best Wins: While we ultimately hit four of our top five bowl picks (all four covering by 17+ points) and 64% ATS of the 60%+ bowl picks, we ended with an ATS Push and an O/U loss in the BCS Championship game... Because the "push" was actually a loss for most, I'll throw the whole game into the "Toughest Losses" category.
Toughest Losses: Technically, it's a push. But, with rapid line movement in the direction of our pick down the stretch, what was a 58.5% play in Oregon +3, become a "no pick" by the time it closed at Oregon -1 at game time. Most people probably got it somewhere in between, where anything less than Oregon +3 was a "weak," but playable play. Obviously, the game could have gone either way. I could complain about several plays, calls, etc., but the truth is, that we did not foresee the defenses playing nearly as strongly as they did. While we can account for just about everything that goes into a game in a normal situation, having 37 days to prepare for the blur and Malzahn offenses, definitely played a role beyond anything we considered. In addition to all the points that were left on the field (though I'm not sure that there were 33+ points left on the field), the defenses played much better than expected. The game totally defied our projection in terms of the O/U. For the first time in a long time, I can definitively acknowledge that something that we did not consider (beyond injuries) created that situation. We didn't sell this pick, but that does not mean that we did not want to get it right (enough negatives in that sentence?). I can assure you that we will do everything that we can to improve our projections in late season bowl games pitting two unique offenses/defenses in the future.
2011 Teams To Watch (my ten most intriguing for next season): Alabama Crimson Tide, Stanford Cardinal, Oregon Ducks, Oklahoma Sooners, Oklahoma State Cowboys, TCU Horned Frogs, Auburn Tigers, Notre Dame Fightin' Irish, Florida Gators and Nebraska Cornhuskers.
NFL Best and Worst:
Best Wins: 8-0...Go Packers! For the second week in a row, our "Lock of the Week" was a 60%+ opinion on the Green Bay Packers in which we expected them to win outright despite being road underdogs in the NFC Playoffs. As alluded to last week, it's great to be able to root for my team. I always root for our picks first and foremost. It's just better when I naturally root for that team anyways. While the 21-16 and 48-21 wins in Philadelphia and Atlanta respectively highlight our 8-0 ATS playoffs to-date, it seems the books and public are catching on. I would love to still be able to root for my team to win and cover in the next two games, but that may not happen. We'll see what the numbers say when the picks are posted on Wednesday, but this looks like a situation that could get out of hand with respect to the overreaction to the success of the Packers (likely to artificially inflate their line if they make the Super Bowl). I'm glad we had at least two weeks to take advantage of the undervalued Packers - more like the overvalued home teams. The final four teams in the AFC were all elite teams pretty much all season. In the NFC, there were no elite teams for the whole year. However, as constructed at the end of the regular season, the team most capable of being dominant/elite was Green Bay. Next on that list was the New York Giants and then Philadelphia and Chicago and New Orleans. I would have ended the list of those caliber teams in the NFC before I would have mentioned Atlanta (or Seattle or any other NFC team). The Falcons were solid, yet never spectacular and the entire NFC South benefited from a fairly easy schedule. On the other hand, the Packers dealt with so many injuries that it was difficult for the public to understand just how strong they could be. With an elite quarterback, an ultra-aggressive and productive linebacking corp, at least two Pro Bowl caliber cornerbacks, the deepest and mo8-st talented group of receivers in the league, a huge, run-stopping defensive line and an emerging offensive line, Green Bay has excellent talent and finally fought through injuries to come together by the end of the year. Here's hoping the Packers go on to win the Super Bowl - as long as I can root for them along the way (To some extent, we are in a great position because our preseason Super Bowl projection, Green Bay vs. New York Jets, is still alive. We had the Jets winning 21-17 at that time. While I always want the ATS picks to win, that situation could present many interesting and positive situations for our projections.)
Quick hitters: Whoa Jets. Yes, the New York Jets were our preseason Super Bowl favorite, but, even though they still won as many games as projected, they did not really look like a legitimate Super Bowl team until they manhandled the Patriots on Sunday. Consider me and the Predictalator fans... And remember, as we noted often, despite being our Super Bowl "favorites" entering the playoffs, we never had a strong enough opinion on the New England Patriots to warrant action on them relative to the lines. The "field" always won more than three out of four Super Bowls we simulated... Matt Hasselbeck and company found a way to make it interesting for everyone in this industry, but the Bears were the definitively superior team in Chicago on Sunday. Seattle left the best homefield advantage and a cushy matchup against the New Orleans Saints to take on a very good team in front of a very raucous crowd at a very famous stadium. How did anyone let that line go DOWN from one week to the next? Don't ever let one game dictate your next move like it did for those who played New Orleans -10.5 two weeks ago and then Seattle +10 this week. Unfortunately (for the public) the majority of public players were on those sides. Fortunately (for us and you hopefully), we took advantage of both lines... While the Seahawks and Patriots came way too close to seemingly improbable backdoor wins/covers, the Pittsburgh Steelers looked like a great bet to push/lose ATS at -3/-3.5 when they were tied with Baltimore at 24 and driving late in the game. Then, fourth/fifth string wide receiver Antonio Brown caught a ball on his head David Tyree-sytle (or Braylon Edwards-style) to set up a go-ahead touchdown and salvage a Pittsburgh win at any line. That was definitely lucky, but 8-0 does not happen without a lot of luck... Our published confidence made the likelihood of us going 8-0 ATS about 4.5 times more likely than if we had just thrown darts/randomly selected picks, but that does not mean that it was that likely to happen. In fact, for the most part, our confidence in a pick has coincided with how comfortable we ended up being watching it - meaning that Green Bay +2.5 at 61.5%+ to cover looked imminently more correct and less in doubt than Pittsburgh -3 at 54.4% to cover... To many, this is a what have you done for me lately industry where entities are only perceived as good as their next pick. That's not necessarily the appropriate way to evaluate any resource (the right analysis will be the one that is most successful in the long-run). That being said, while it is great to go 8-0 ATS and somewhat comforting to know that 8-3 ATS is our worst case scenario in the playoffs, we definitely want to win every game. So, come next Monday, here is hoping we are talking about 10-0 ATS and looking forward to hitting my eighth straight Super Bowl pick ATS...
Toughest Losses: For the second week in a row, we missed our top total pick when Seattle @ Chicago went Over (41). In the last five minutes of Sunday's games, there were 39 collective points scored. It definitely could have been worse had Seattle gone for two to Push and/or if the Patriots found a way to tie the game and then have the first ever nine point OT win with the new playoff rules.
As usual, if you have any of your own suggestions about how to improve the site, please do not hesitate to contact us at any time. We respond to every support contact as quickly as we can (usually within a few hours) and are very amenable to suggestions. I firmly believe that open communication with our customers and user feedback is the best way for us to grow and provide the types of products that will maximize the experience for all. Thank you in advance for your suggestions, comments and questions.