2012 Adjustments
I always research my methods during the summer, but I dig especially deep after a losing year and I’ve certainly proven that one down year is not an indication of future problems - and I expect to bounce back again in 2012-2013 given that I've never had consecutive losing years (Football and Basketball handicapping combined) in 25 years as a professional handicapper. Losing years are going to happen randomly regardless of how good my long term record is, and some of best years have come after a rare losing year.
The research of my methods this summer has unveiled some interesting findings that should help return me to my normal winning ways this next year and I’m going to make a few changes this year to enhance my likelihood of success.
I’ll break down my handicapping sport by sport with some details on my research and some added features for the upcoming year. One feature that I will be adding is the release of some college football Best Bets earlier in the week to beat the line moves that have been occurring before my release in recent years. Line moves have been occurring earlier in the week than they used to and that has affected my ability to get down on games at the better lines since I have always released my College Football Best Bets on Thursday. Starting this season there will be some games that I will release earlier in the week to try to beat those early line moves.
COLLEGE FOOTBALL
College Football has been my best sport over the years in terms of winning percentage (although Basketball is more profitable because there are more games), and I’m +209.9 Stars or profit on my College Football Best Bets over the last 13 years (at -110 odds), but last year I suffered my most unlucky season ever and ended the season at -1.3 Stars. I once again won on my futures bets, which I release to subscribers before the season starts, as I was +12.9 Stars on those plays (I’ve been profitable all 3 years since I started releasing my college football futures) but I was just 42-41-1 for the season (98-102-2 on a Star Basis for -14.2 Stars at -110 odds) on my Best Bets.
However, my College Football handicapping was actually as good as it’s always been and hitting 51% on my Best Bets is actually good considering that my Best Bets were a ridiculously unlucky -19 in fumbles lost margin. Fumbles are 90% random in college football, so being -19 in fumbles is simply bad luck and there isn’t much I can do about it. Teams I had Best Bets on were actually positive in fumbles when I didn’t bet on them, which makes the -19 in fumbles even more random.
My 2011 College Best Bets covered the spread by an average of 1.8 points per game (normally good for 55% winners) despite being -19 in fumbles, which is another indication that my handicapping was good. I had 8 Best Bets where I was on the right side with my teams dominating from the line of scrimmage and losing because of random events (UCF, Ohio, and Utah State in week 4, Illinois and Syracuse in week 5, Toledo in week 7, Hawaii in week 9, and Iowa +14 in the bowls). I had my only lucky win of the season and a lucky push in week 7 with Navy covering despite being badly out-gained and Oregon pushing a game they probably should not have. I also had 17 toss-up games that could have gone either way and I was just 6-11 on those games.
So, I was only 42-41-1 on my College Best Bets in 2011 but I had more 7 more unlucky losses than lucky wins and I would have been 49-35 for 58% winners if I won the 8 games that I lost that I had the right side on and lost the two games that I won and pushed that were undeserved (and would have had an even better record if I split the toss-up games instead of being 6-11 on those).
It was a tough year record wise due to the negative variance of fumbles and close games, but my handicapping was very good, as it has always been in college football. I’ve mentioned many times that I’ve also had years when I had more lucky wins than unlucky losses and my record was better than it should have been in those years (like the 74% I hit in 2005, when I should have been at 66%). Unfortunately, 2011 was a season with the most negative variance I have ever had, which is why my record was 51% rather than the 58% it should have been. My College Football handicapping was still at a 56% or higher level so don’t make the mistake of losing faith in my methods, which have worked for me for years. My College Football Best Bets are 887-690-37 (56.2%) since 2004 and 56% lifetime. My handicapping was actually as good, or better, than normal in the 2011 season and I expect to win in 2012 since it’s highly unlikely I’ll experience such negative variance in fumbles and close games.
College Football Math Model
There have been concerns in recent years, although not by me, that the odds makers have caught up to my College Football math model, but that is certainly not the case. In fact, my College Football math model has been nearly as good the last 5 years as it was during my very profitable years from 2004 through 2006. The games that my College model gave a 56% or higher chance of covering are 443-327-13 ATS (57.5%) since the 2007 season, including 72-52-1 ATS (58.1%) last season (the math model plays are now 623-440-19 ATS (58.6%) since I started using my current model in 2005). I did discover that I needed more of a differential between my math model prediction and the line in the first 3 weeks that the model applies (i.e. weeks 5 through 7), which makes sense since there is more variance in the math prediction with fewer games in the model, and that change should raise the winning percentage of the games that apply.
College Football Technical Analysis
I noticed a slip in the effectiveness of my technical analysis (i.e. situations, fundamental indicators, and team trends), as the technical analysis has been 55.2% lifetime but just 52.1% the last 5 years. The stronger technical games have still been good in recent years (57.8%) but the technical plays not in that group will no longer be played unless the math model is significantly on that side, and that change should improve the win percentage of my Best Bets going forward.
Combined College Football Analysis
I also have a spread sheet that combines my math model with my technical analysis and recent years have shown that I need to weigh the math model heavier and the technical analysis lighter than I have in the past. More reliance on the math model should increase my win percentage, although I’ll probably have slightly fewer Best Bets.
NFL
As most of you know the NFL football analysis has gone from my best sport early in my career to my most troubling sports in recent years as my old math model became less effective as the odds makers and public became more aware of the contrary nature of the NFL, which was built into my original model. As I mentioned last year, I altered my math model prior to the 2011 season and the new math model was a very good 81-52-3 ATS (60.9%) when the difference between my prediction and the line was 1 ½ points or more. (the NFL math predictions are released each week with the Best Bets). The problem is that I didn’t realize that the threshold for significant results was going to be so much smaller than in the previous model, which needed a 5 point or more difference from the line to start producing really good results. So, I didn’t release as many games as Best Bets based solely on the math as I should have. The NFL technical analysis was also very good in 2011, going 82-53-5 ATS (60.7%), and the combination of the new math model and a good year for the technical analysis led to a very good 144-102-6 ATS (58.5%) record picking every NFL side (with 15 no opinions) in 2011 (the win percentage picking every game in the NFL is 57% the last two years).
Despite the very good overall year that my new math model and technical analysis had, my NFL Best Bets in 2011 were a poor 18-20-1 ATS for the season while my Strong Opinions were 28-10. Part of the reason for the bad record on my NFL Best Bets was being an unlucky 4-10 on games that could have gone either way (rather than being 50% on the toss-up games). I also didn’t know that my new math model would work as well as it did and a lot of those games with line value became Strong Opinions rather than Best Bets. It’s a shame that the games I considered as possible Best Bets were 46-30-1 (that's the record of the Best Bets and Strong Opinions combined) while the Best Bets were just 18-20-1, but my methods obviously worked well overall (59% ATS picking every NFL game), which is a good sign for the 2012 season.
NFL Math Model
As I mentioned above, my new math model performed very well last season (81-52-3 ATS (60.9%) when the difference between my prediction and the line was 1 ½ points or more), but I didn’t trust the model as much as I should have since it was my first year since tweaking it. The old model needed a 5 points or more differential between the prediction and the line to be significant, but the new model was more precise and needed only 1 ½ points or more difference to produce good results. This year I will trust the math model more and will release more Best Bets based solely on the math.
NFL Technical Analysis
2011 was a very good year for my fundamental and match up indicators (34-14-1 ATS) and a bit of a down year for my situational analysis (61-54-3 ATS) and overall the technical analysis was 82-53-5 ATS (60.7%) in 2011 (when either the fundamental or situational apply but don’t go against each other). That great record is part of the reason why my overall record picking every game was so good in 2011, but the long term record of the NFL technical analysis is a more modest 54.4% and will be weighted accordingly.
Combined NFL Analysis
I was too conservative in the NFL and having only 39 Best Bets was a mistake given that my overall record on Best Bets and Strong Opinions (46-30-1 ATS) was very good and my NFL Free analysis was also very good. Part of the reason for being so conservative in the NFL was my recent lack of success and an uncertainty as to how well the new math model would perform (I didn’t realize that a smaller differential from the line was needed until later in the year). I vow to release more NFL Best Bets in 2012, which should improve my chances for a profitable season by limiting the affect of variance.
I had a good grasp on the NFL overall in 2011 given that I was 58.5% ATS picking every NFL side and it's pretty random that my Best Bets didn’t win (I was also 134-105-3 ATS picking every NFL game in 2010). I've never had a season in which my methods have worked so well and my analysis overall has been so good yet I've lost on my Best Bets. However, I'm excited about the performance of my new math model and I'm looking forward to being profitable in the NFL in 2012.
BASKETBALL
My Basketball Best Bets were just 206-192-5 (51.8%) for -20.8 Stars and the problem was the NBA. I was a profitable 158-136-4 on College Best Bets but only 48-56-1 on my NBA Best Bets, as the NBA situations did not work at all – most likely due to the strange scheduling of the shortened season.
College Basketball Methods Study
While my College Basketball handicapping was profitable in 2011-12, I was not satisfied with my 53.7% win percentage (53.2% on a Star Basis) so I went back over my notes for the last 3 years and added up the record of every game that applied to a significant situation and every game that my ratings showed significant line value to see if I can improve my performance.
For every game in which a situation applies I assign a rating from 1 to 5 and I generally play games with a 4 or 5 situational rating as long as my ratings suggest that the line is at least fair. I will play games with situational ratings of 2 or 3 if my ratings suggest that I have line value on the same side in addition to the positive situation. What I found is that I’m not giving enough credit to my situational analysis in College Basketball and that a game with a good situation shouldn’t be passed if my ratings perceive the line value to be just a bit negative. In other words, I’m passing on a lot of winning games that have a situational rating of 2 or 3 because my ratings suggest negative line value of ½ a point or a point. My ratings do show overall value, but my ratings are also not 100% what the true line should be, so a difference from the line of 1 point or less is not strong enough for me to pass on a good situation that favors the other side.
Last season, for instance, the college games that had a 2 or higher situational rating were 414-340-14 ATS, a very profitable 54.9% and +40 units at -110 odds, but my record on College Best Bets with a 2 or higher situational rating were 138-114-4 ATS (54.8%), which means that the significant situational games that I didn’t play were 276-226-10 ATS (55.0%). That’s a lot of profit that I passed on for various reasons. Reason #1 is that I didn’t have enough faith in games that I applied a situational rating of 2 or 3 and the other is because I was passing on too many games when my ratings even slightly favored the other side. My ratings do have merit, as games with a significant rating difference (i.e. when my ratings prediction was 2 points or more from the line) were a decent 53.4% ATS and Best Bets that had a ratings difference of 3 points or more that also had a positive situational rating were a very profitable 45-18-1 ATS last season.
Over the 3 years of my study the games with a situational rating of 2 or higher were a very profitable 1079-878-41 (55.1%). If I just made every one of those games a 2-Star Best Bet then I would have been +226.4 Stars over those 3 years. Those are not back-fitted results, as 3 seasons ago I began assigning situational ratings to every game before the games each day were played. I just didn’t get around tabulating the record of the situational ratings until this summer because I had been +65.0 Stars in Basketball Best Bets the two years prior to last season (so I didn’t think I needed to fix anything) and I spent most of my time last summer working on my new NFL math model. I just assumed that the games with situational ratings of 2 or 3 were going to be worse than the games with situational ratings of 4 or 5 and that was not the case. The games rated at 5 are actually better at 58.1%, but the games with situational ratings of 2, 3, or 4 are not significantly different from one another and, in fact, the 2 rated games were actually slightly better.
My College Basketball ratings have been pretty good, but they simply haven’t been as strong as the College situations have been and most of the profit from the ratings comes when a situation is favoring the same side. As I mentioned earlier, those games were 45-18-1 ATS last season and they have been 123-85-5 (59.1%) the last 3 years. A game that has a ratings differential of 3 points or more from the line that didn’t have a positive situational rating were a more modest 53.2% and the games with a ratings differential of 2 or 2 ½ points from the line with no positive situation in support were just 52.1%.
The biggest mistake I’ve made the past 3 years in College Basketball is allowing a small negative ratings differential to keep me from playing games that apply to a significant situation. I’ve had good success on my College Basketball Best Bets the last 3 years (403-335-11 for 54.6%), but I’ve left a lot of profit behind by under-weighting the situational analysis and over-weighting my ratings. I also have done a poor job of assigning Star values to my College Basketball Best Bets, as my 2-Star Best Bets have the exact same long term win percentage as my 3-Star and 4-Star Best Bets in College Basketball. I now know that the games with a situational rating of 4 are no better than those rated at 2 or 3 and the only games that have really proven to be higher percentage plays are the games that have a positive situational rating along with a difference of 3 points or more in my ratings from the line. There are grey areas in between that will be worthy of a higher rating, but most of my higher rated plays will be games that apply to a significant situation AND have significant line value.
If I had just played every single game with a situational rating of 2 or higher I would have been 1079-878-41 (55.1%) the last 3 seasons for +226.4 Stars even if I had made every game only a 2-Star Best Bet, so I’ve been way to conservative and that will change this season. I do realize that it may be hard to kick my habit of being too conservative, but I will also include every single game with a situational rating of 2 or higher in my daily email, regardless of whether it is a Best Bet or not, so those of you that just want to play those can do so – even if I’m not.
NBA Methods Study
My NBA handicapping has been just as good as my College Basketball handicapping over the years, as I am 53.7% on a Star Basis on my NBA Best Bets the last 13 seasons for +123.7 Stars of profit at -110 odds. However, last year was my worst year percentage wise (46.2%) and my second worst year in 25 years in profit/loss at -32.4 Stars. The NBA situational analysis was just 47.0% and the games with a significant difference between my ratings and the line were only 49.1% last year (the first year my ratings have been under 50%). I’m hoping the problem was the strange scheduling in a shortened regular season, as that is a plausible reason why my situations didn’t work. I’m not sure I can blame a down year in my ratings to the different scheduling patterns but it certainly didn’t help.
Over the 3 years since I’ve been assigning situational ratings to each game the NBA situations are 503-468-19 ATS, but they were a horrendous 151-170-6 ATS last season, which means that the situational analysis was a profitable 352-298-13 ATS (54.2%) in the seasons with a normal 82 game schedule. My NBA ratings were down last year but they’ve been a solid 53.5% over the years on significant differences from the line and I’ll chalk up last year’s down year for the ratings to variance.
With the NBA going back to a normal 82 game schedule, I expect my situational analysis to start working again and the long term record of my NBA ratings has been profitable too. So, I’m expecting to bounce back with a profitable NBA season, as I’ve never lost in the NBA in consecutive seasons in 25 years.
2012-13 Expectations
Last year was a down year at 51.3% on all Best Bets in football and basketball combined (266-253-7), but I’ve never had consecutive unprofitable results in 25 years as a professional handicapper and I’ve averaged +52 Stars of profit per year (+585 Stars the last 13 years for an average of +45.0 Stars). My long term results over thousands of games is certainly more indicative of my future results than last year’s 51% record and the research I did on my methods should help me better navigate a return to profitability this coming year by being more focused on what handicapping methods have worked for me in recent years.
Image Source: nsrecreation.wordpress.com
No comments:
Post a Comment