NFL ALPHAS 213-214
BACK IN THE BLACK SUPERBOWL XLVIII LAST YEAR s Blackout Bowl victory by the Baltimore Ravens saw our Super Bowl forecasting model recover its rightful place in the winner s circle. We have now successfully identified the undervalued contender in nine of the last 1 Super Bowls, and our new streak is at an inspiring one-in-a-row. THIS YEAR s Super Bowl pits the NFL s No. 1 scoring offense in the Denver Broncos, against the NFL s No. 1 scoring defense in the Seattle Seahawks. Both teams entered the season with high expectations and both teams delivered. The Broncos and Seahawks represented two of the top three odds-on favorites to reach the big dance, and both finished with an NFL-best 13-3 record. The only thing tougher to forecast than the winner of the matchup is the weather at MetLife Stadium. For sports fans, commercial fans and meteorologists alike, Super Bowl Sunday is rife with speculation. As always, we hope the analysis that follows fuels even more. At, our day job is to deliver risk-adjusted returns in excess of a benchmark; our night job is also to deliver risk-adjusted returns in excess of a benchmark. Yet, somewhere in between, we produce a quantitative survey of each NFL season from an investor s perspective. At the heart of our analysis is a measure of each team s regular season excess performance. This excess performance, or NFL Alpha, is defined as the return-on-investment (ROI) that could have been earned by placing a systematic wager on an NFL team to win each of its games outright in the regular season. We have found that these regular season NFL Alphas possess strong implications for both subsequent season and post-season success. Boom and Bust In a rare encore performance, Andrew Luck and the Indianapolis Colts were perched atop the NFL Alpha P&L (Table 1) for a second consecutive year. A casual bettor would have earned a return of 51.8% with a wager for the Colts to win each of their games outright. Despite having an inferior win-loss record to five other teams, at home wins against the Broncos and the Seahawks, along with away wins against the 49ers and the Chiefs, all as heavy underdogs, made the Colts an unparalleled investment (Chart 1). Bookmakers had a tough time pricing the Colts throughout the season due to an array of injuries and the personnel changes that resulted from being fleeced by the Browns in the 1 Trent Richardson trade. Somehow the Colts repeated an 11-5 record, beat the S&P 5 (32.4%) and made an inept Cleveland Browns organization appear mildly coherent, all in the same year. While the Colts were busy winning, the Houston Texans were cobbling together an impressive 14 consecutive losses to become this year s worst investment in the NFL. Clocking in with an abysmal -82.9% Alpha, they had the lowest Alpha of any two-win team since the NFL went to a 16-game schedule in 1978. After a successful 12-4 campaign in 212, expectations were deservedly sky-high for the Texans heading into the 213 season. They started the season with a pair of unconvincing wins against two teams that they were expected to bulldoze.
Table 1 ALPHAS FOR ALL 32 NFL TEAMS A L P H A 2 1 3-2 1 4 Alpha Favorite/ Team 213-214 212-213 Change Record Underdog Indianapolis Colts 51.8% 59.4% -7.6% 11-5 9-6-1 New York Jets 39.4% -32.8% 72.2% 8-8 5-11 Carolina Panthers 32.% -2.3% 34.3% 12-4 11-5 Arizona Cardinals 31.% 7.% 24.% 1-6 6-1 San Diego Chargers 27.7% -11.2% 38.9% 9-7 6-8-2 New England Patriots 22.7%.7% 21.9% 12-4 11-5 Philadelphia Eagles 19.4% -48.2% 67.7% 1-6 12-4 Cincinnati Bengals 18.2% 21.3% -3.1% 11-5 11-5 Seattle Seahawks 13.7% 34.% -2.2% 13-3 15-1 Miami Dolphins 12.6% -13.8% 26.4% 8-8 7-9 San Francisco 49ers 12.4% 23.1% -1.7% 12-4 14-2 Kansas City Chiefs 12.3% -54.7% 67.1% 11-5 12-4 St. Louis Rams 1.% 19.4% -9.4% 7-9 4-12 Denver Broncos 4.6% 18.5% -13.9% 13-3 16- Jacksonville Jaguars 4.2% -68.5% 72.7% 4-12 -16 New Orleans Saints 2.8% -19.4% 22.2% 11-5 13-3 Chicago Bears 1.7% -3.7% 5.4% 8-8 7-9 Pittsburgh Steelers -.3% -2.8% 2.5% 8-8 8-8 New York Giants -1.2% 3.3% -4.5% 7-9 7-9 Baltimore Ravens -3.1% 2.2% -5.3% 8-8 8-8 Buffalo Bills -8.2% -3.7% 22.5% 6-1 2-14 Green Bay Packers -12.1% 5.1% -17.2% 8-7-1 11-5 Dallas Cowboys -13.% -5.4% -7.5% 8-8 8-8 Tennessee Titans -19.1% -2.1% -17.% 7-9 7-9 Minnesota Vikings -2.8% -49.4% -7.3% 5-1-1 4-12 Oakland Raiders -25.9% -51.1% 25.2% 4-12 2-14 Detroit Lions -26.% -52.3% 3.2% 7-9 13-3 Cleveland Browns -37.4% -36.3% -1.% 4-12 5-11 Tampa Bay Buccaneers -41.4% -5.3% -36.1% 4-12 4-12 Atlanta Falcons -55.% 41.2% -96.2% 4-12 5-11 Washington Redskins -64.2% 44.9% -19.1% 3-13 4-11 Houston Texans -82.9% 6.% 88.9% 2-14 7-9 Source: Analytic Investors, nfl.com 2
% ALPHA % ALPHA CUMULATIVE ALPHAS 12 1 8 6 4 2-2 -4-6 -8 Chart 1 Indianapolis Colts vs. Houston Texans -1 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 6 5 4 3 2 1-1 -2-3 -4-5 -6-7 -8-9 -1 1 An NFL Alpha is predicated on the assumption that an investor in a team receives the fair payoff on an investment to win outright. In reality, bookmakers never pay fair odds, but our model assumes that odds are always fair. The fair payoff is calculated as (1/p-1), where p=probability of victory. For example, if a team has a 25% probability of victory, then the fair payoff is 3 1 (invest $1, win $3). The team s fair probability of victory is derived from a model, whereby we calculate the probability of victory as a function of the point spread. Indianapolis Colts Houston Texans GAMES Chart 2 Denver Broncos vs. Jacksonville Jaguars Denver Broncos Jacksonville Jaguars -11 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 GAMES Since the Texans were heavy favorites in these games, the Alpha generated from these wins was miniscule and therefore did little to offset the 14 losses that came once the wheels fell off (Chart 1). It is certainly rare for a two-win team to have been a heavy favorite in both of its wins, which was the primary catalyst for their worst since 1978 accomplishment. Manning Expectations To shed further light on the engineering behind the alphas, it is useful to examine the seasons of two seemingly disparate NFL teams, the Broncos (13-3, 4.6% Alpha) and the Jaguars (4-12, 4.2% Alpha). While their win/loss records would indicate one to be clearly a better investment than the other, their performance relative to betting market expectations (NFL Alpha) tells a different story (Chart 2). Bookmakers had the Broncos favored in each of their 16 games, predicting an average margin of victory of nearly 1 points. On the other hand, the Jaguars played the role of underdog in each of their 16 games, with an average predicted margin of loss of nearly 1 points. Thus, for the average game, the Broncos were 1-to-4 favorites while the Jaguars were 4-to-1 underdogs to win. In other words, it would take close to four Bronco wins to earn the same return as one Jaguar win. In October, these two teams met in what was expected to be the most lopsided game in NFL history. 3
LOW VOLATILITY BY GAME WEEK 6 Chart 3 Low vs High Risk Return 2 Chart 4 Low vs High Risk Compound Return % RETURN 4 2-2 -4 % RETURN 15 1 5-5 FAVORITE LONGSHOT -6-1 -8 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 17-15 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 17 GAMES GAMES LOW HIGH As 26.5-point underdogs, the Jaguars were a 5-to-1 longshot to win. Not surprisingly, if the Jaguars had defied expectations and won the game, they would have ended the season with a 389% Alpha, nearly three times as high as the best alpha ever recorded of 132% by the 1992 Indianapolis Colts. Slow and Steady Wins the Race For roughly the last 1 years, we have authored research pertaining to the low volatility anomaly. Namely, this anomaly describes the propensity for low systematic risk securities to outperform those with high systematic risk. In equity markets, systematic risk can be quantified by a security s beta. In gambling markets, systematic risk can be analogous to the pointspread of a game, as defined by bookmakers. As discussed in a prior NFL Alpha paper, longshot or high systematic risk bets tend to be overpriced and make worse investments than their favorite or low systematic risk counterparts. This year, we examined a potential betting strategy to exploit this anomaly. Before each of the 17 game weeks this year, we separated the week s matchups into three bins based on point-spread/systematic risk: low, medium, and high. We then computed the hypothetical return to each bin for a set of fixed wagers on its constituents to win their matchups outright. In Chart 3, also known as a compulsive gambler s post-mortem, we see the return of the low and high systematic risk bins by game week. Almost definitionally, the high risk, longshot bin has much more return fluctuation than that of the low risk, favorite bin. What may surprise some is that the average return of the longshot bin is -14.8%, while the average return of the favorite bin is 7.3%. When compounded throughout the season, the return to the high systematic risk bin is an alarming -99.4%, versus 169.4% for the low systematic risk bin (Chart 4). While a 5-to-1 payout may get headlines and fuel reveries, it may not serve as a prudent investment. Another Anomaly What happens when you combine a player s nobody believes in us mentality with a stadium filled by his inebriated fans dressed in adult Halloween costumes? You get the home underdog effect, and you may also get some alpha for your troubles. The home underdog bias/ anomaly is no stranger to academic literature; by definition, the anomaly suggests that home underdogs tend to be systematically undervalued in NFL wagering markets. A recent study here at Analytic Investors has found this effect to be magnified for slight home underdogs defined here as games in which the home team is expected to lose by a field goal or less. 4
SLIGHT HOME UNDERDOGS Chart 5 1978-212 Seasons Chart 6 213 Season % WAGE SUCCESS RATE 6 5 4 3 2 1 STRAIGHT UP AGAINST THE SPREAD % WAGE SUCCESS RATE 6 5 4 3 2 1 STRAIGHT UP AGAINST THE SPREAD ACTUAL EXPECTED Between 1978 and 212, and for the 213 season alone, we have computed the record, both against the spread and straight up, for slight home underdogs (Charts 5 and 6). In all wagering scenarios, the actual betting success rate was at least 4% more than what bookmakers expected. This year, we found the home underdog bias to be even more pronounced than usual. While the cause of this anomaly is surrounded by its fair share of speculation, it is reasonable to suppose that the benefits of playing at home are more meaningful when the teams are nearly evenly matched. So the lesson is, don t get rid of those Halloween costumes just yet, because they are a gift that keeps on giving. Low Alphas Need Not Apply An interesting by-product of the NFL Alpha hierarchy is its implications on relative coaching success. Despite the fact that the Jacksonville Jaguars (4.2%) finished with four wins, their two-win team effort in the 212-213 season kept expectations lower than Richard Sherman s approval rating in San Francisco. Put simply, the wagering markets were not expecting much, and did not get much from first-year head coach Gus Bradley s roster of junior varsity footballers. This fact is reflected in the Jaguars Alpha relative to another four-win team, the Tampa Bay Buccaneers (-41.4%). After a seven-win campaign in the 212-213 season, the Buccaneers were expected to be a formidable tackle football team this go-around. After a series of staph infections and an -8 start, head coach Greg Schiano became the last staff infection to be eradicated at the season s end. Along with the Buccaneers, the Minnesota Vikings, Detroit Lions, Cleveland Browns, Washington Redskins, and Houston Texans all parted ways with their coaches, representing six of the eight lowest alpha teams. 5 Ended at the Bottom To continue harping on this year s underachievers, the three lowest Alpha teams also possessed the three worst year-over-year changes in NFL Alpha. Per Table 1, the Falcons, Redskins and Texans saw their Alphas change by a leagueworst -96.2%, -19.1%, and -88.9%, respectively. Historically, this has not been an uncommon phenomenon. We have found that between seasons, there tends to be a negative correlation between the Alphas earned by NFL teams. Thus, this season s underachievers tend to become next season s overachievers, and vice versa. Prior research here at Analytic Investors has also suggested that this relationship is prevalent between each NFL regular season and post-season. It is through this relationship that we forecast our post-season and Super Bowl selections.
Table 2 POST SEASON ANALYSIS (5-4-1 RECORD) A L P H A T E A M S HIGHER Alpha LOWER Alpha Favorite Result WIN WC Indianapolis Colts 51.8% WC Philadelphia Eagles 19.4% WC San Diego Chargers 27.7% WC San Francisco 49ers 12.4% Kansas City Chiefs 12.3% Chiefs by 2.5 Colts 45-44 New Orleans Saints 2.8% Eagles by 3 Saints 26-24 4 Cincinnati Bengals 18.2% Bengals by 6 Chargers 27-1 Green Bay Packers -12.1% 49ers by 3 49ers 23-2* N/A DP Indianpolis Colts 51.8% DP Seattle Seahawks 13.7% DP Carolina Panthers 32.% DP San Diego Chargers 27.7% New England Patriots 22.7% Patriots by 7 Patriots 43-22 4 New Orleans Saints 2.8% Seahawks by 9 Seahawks 23-15** 4 San Francisco 49ers 12.4% 49ers by 1 49ers 23-1 4 Denver Broncos 4.6% Broncos by 7.5 Broncos 24-17** CC New England Patriots 22.7% CC Seattle Seahawks 13.7% Denver Broncos 4.6% Broncos by 5 Broncos 26-16 4 San Francisco 49ers 12.4% Seahawks by 4 Seahawks 23-17 WC = WILDCARD / DP = DIVISION / CC = CONFERENCE * The result of this game matched the point-spread exactly, and therefore did not result in a win or a loss for a wager against the spread. ** While in these games lower-alpha teams did/did not lose to the higher-alpha teams, the predictions are correct/incorrect because the lower Alpha teams covered/did not cover their respective point spreads. 6
Super Bowl XLVIII This year s Super Bowl matchup features the lower-alpha Denver Broncos (4.6%) against the higher-alpha Seattle Seahawks (13.7%). As noted previously, we have found evidence that lower-alpha teams in the regular season tend to be undervalued throughout the post- season. Thus, as current 2-point favorites, we think Peyton Manning & Co. will overcome Richard Sherman s arrogance to win this one by at least a field goal. SUPER A L P H A T E A M Prediction BOWL Date HIGHER LOWER Favorite Result Correct Correct XXXVIII 2/1/4 New England Patriots Carolina Panthers Patriots by 7 Patriots 32-29* 4 (67.%) (39.%) XXXIX 2/6/5 New England Patriots Philadelphia Eagles Patriots by 7 Patriots 24-21* 4 (33.5%) (12.6%) XL 2/5/6 Seattle Seahawks Pittsburgh Steelers Steelers by 4 Steelers 21-1 4 (25.%) (11.4%) XLI 2/4/7 Chicago Bears Indianapolis Colts Colts by 6.5 Colts 29-17 4 (17.8%) (14.5%) XLII 2/3/8 New England Patriots New York Giants Patriots by 12.5 Giants 17-14 4 (22.9%) (14.6%) XLIII 2/1/9 Pittsburgh Steelers Arizona Cardinals Steelers by 6.5 Steelers 27-23* 4 (33.6%) (-6.4%) XLIV 2/7/1 Indianapolis Colts New Orleans Saints Colts by 4.5 New Orleans 31-17 4 (37.6%) (12.8%) XLV 2/6/11 Pittsburgh Steelers Green Bay Packers Packers by 3 GreenBay 31-25 4 (28.6%) (1.3%) XLVI 2/5/12 New York Giants New England Patriots Patriots by 3 Giants 21-17 (32.8%) (15.8%) XLVII 2/3/13 San Francisco 49ers Baltimore Ravens 49ers by 4.5 Ravens 34-31 4 (23.1%) (2.2%) XLVIII 2/2/14 Seattle Seahawks Denver Broncos Broncos by 2?? (13.7%) (4.6%) * While in these games lower-alpha teams did lose to the higher-alpha teams, the predictions are correct because the lower-alpha teams covered their respective point spreads. The opinions expressed herein are those of Analytic Investors and are subject to change without notice. This research report is prepared for general circulation and is circulated for general information only. It does not have regard to specific investment objectives, the financial situation and the particular needs of any specific person who may receive this report. 555 West Fifth Street, 5th Floor Los Angeles California 913 T 213.688.315 www.aninvestor.com 7