21st Century NBA Basketball Prediction Program - Stuck In Neutral

  I had high hopes for my basketball prediction program heading into the 2017-2018 NBA season. My program had better than the magic 52.4% percentage in each of the last 5 seasons, 4 in retrograde analysis and the 2016-217 season which was the first one in which I took donations for advance predictions and put my money where my mouth was by placing bets on the gambling site Bovada. The only sour taste was that my Bovada account was suspended (you can read about it here) during a hot streak which made my season bets $4.06 in the red. This sting was offset by $70 in donations for advance picks and anther $126 in a Bovada welcome bonus so all told I ended the season over $190 on the plus side.

  My gambling problem from last year wasn’t my program – it was psychological. I started off with $10 bets but then switched to $5 bets and then varied my bets from $2.50 to $20 dollars depending on how I felt about the computer picks. In retrospect this seemed panicky and I resolved to place bets designed to win $5 on a successful outcome on each computer prediction and attempt to remain detached from the bets. In large part I kept to this resolution and only broke it a handful of times by adding three of my own predictions and leaving three computer bets unacted upon (the reasons of which I will get to later).

  The way I had this season planned was that my program would break off to a good start which would allow me to advertise my success online and pick up lots of donations to fund even more advertising which would lead to even more donations which would allow me to increase my bets from $5 to $7.50 with the extra capital I would be accumulating. Unfortunately my program started off by going 1-3 and only got close to .500 three 3 weeks into the season at 19-20 which didn’t give me anything to advertise about since in the gambling world an even record is losing money since you have to have 11 wins for every 10 losses to make up for the 10% penalty on losses (otherwise known as the vig or juice) which is why the break even winning percentage is 52.4% and not 50%.

  From there the pattern was maddenly consistent. My picks went from 19-20 to 22-29 on November 27th to 31-31 on December 3rd to 36-44 on December 13th and bottomed out at 54-65 on January 3rd with losses of $86.25. At this point the program started doing better and on February 27th the record stood at 87-84 with losses of $25.25. Since then the program and I have treaded water and heading into April 3rd the record stands at 110-110 with $52 in losses and only a week to go in the regular season at which point the computer doesn’t pick playoff series and I am on my own.

  I haven’t started a deep dive into the individual games that made up this year’s medicore result but a cursory glance showed that my record picking teams playing the second half of back to back games has been far worse than last year. I think the reason for this is paradoxically that the NBA schedule has been altered this season to reduce the number of back to back games which likely makes each set of back to back games more of a burden than something a team gets used to. I will research whether teams are faring worse on back to back games than in years past and try to adjust my formula for it. The problem is now that the scheduling has been so drastically changed I can’t just change a formula and test it on prior years.

  Another and far more serious reason for this year’s decline is that teams are resting players far more often than in years past. I have never made an adjustment for injured players because there is generally a short-term bounce in results that the formula has been able to take advantage while the betting line over-reacts to the injury as role players tend to have a few good games in increased roles before showing why in fact they are role players. Resting players is a bigger issue for me. I lost several games where Marc Gasol of the Grizzlies and Joel Embiid of the 76ers were rested on the front end of back to back games after I made the pick. There may have been some ‘insider trading’ in these games although the main culprit is my making the bets earlier in the day instead of trying to wait until the last minute to determine late scratches. Teams have been sitting uninjured players out for ‘rest’ in ever increasing numbers this year or at least it seems to me. I even stopped picking Warrior games since they stopped playing their All-Stars to get them ready for the playoffs that coach Steve Kerr insisted they would be available for if they had started last week. There was a flip side to the Warriors situation since they were still undervalued as all-stars Kevin Durant, Klay Thompson, and Draymond Green returned and I lifted my ban in time for a few winning bets.

  Having lost $50 in the season is no big deal - the playoffs haven’t started yet and I have always done pretty good in the playoffs and have every expectation of getting back in black by the time a champion is crowned. A bigger concern is that since I didn’t have a good year I didn’t get any donations for my picks in advance of when I post them on my blog after the last game of the night finishes. I started the season blogging my picks and results and am driven to finish what I start even though I haven’t been able to turn my program into a revenue stream as I hoped. Once this season is over I will be done with my predictions blog and the true question is going to be whether I want to continue entering in schedules, lines, and results and placing bets. That is what I have the summer to think about.