By Harri Kairavuo | eastsideportsanalysis.com
From a betting perspective 2014 was very bad for me and those who followed my predictions. The year got off to a good start with some very profitable Spring Training picks. The beginning of the regular season was even better and I was very satisfied with how thinks where going and with the interest my service had got. I wasn't though fooled by the good start and knew, that things could change fast. And so they did. In the end, it turned out to be the first losing season for me and meant that there was going to be a lot of work to do in the off-season. Everything bad comes with something good and in this case the good was The Handicapper. The purpose with The Handicapper was to create a software, which could help me and other MLB-handicappers in their daily work.
We’re now two months into the off-season, which has been a pretty wild one what comes to trades and free agent signings. I started the preparation for the 2015 MLB Season a few days after the 2014 Regular Season ended. The first thing on my to-do-list was to try to find out what went wrong this season and why the selections were not profitable. I made some changes to my approach for the 2014 season (read about them here), which obviously did not work out as planned. The biggest change was that I started to use simulation as a part of my handicapping routine and when things started to go wrong, and continued like that for the rest of the season, I was pretty sure that the reason was to be found in the simulation algorithm.
My first thought was, that the problem was in how my model used park factors when simulating games. After some research and testing I could exclude this idea. What I instead found out, was that the way the model handled relief pitchers was problematic. The previous off-season I had put a lot of effort in creating a bullpen-motor, which calculates the probability for each relief pitcher to pitch in a game. What it basically does is ranking relief pitchers according to ability, performance, innings pitched and pitches thrown last three days. The bullpen-motor worked as it was supposed but the problem was in the algorithm that chooses which relief pitcher is pitching in each specific situation of a simulated game. The algorithm did not take into consideration what the score of the game was, it just selected a pitcher according to a static scheme. This was a problem, because it did not reflect well enough how things are handled in real life. As a result, the variation of the result for the simulated games was too low, which was reflected in the result of run line- and totals bets.
The problem with the handling of relief pitching (as described above) is now solved. The algorithm now evaluates the game and decides whether it is in a “close”- (run difference under or equal to 3 runs) or “decided” (run difference over 3 runs) mode. When the simulated game is in a “close” mode the algorithm prefers high-end relief pitchers and when the game is in a “decided” mode it prefers the low-end relief pitchers. This is more in line with how things work in a real life MLB-game and I firmly believe, that these corrections significantly improves my chances to succeed as well as makes The Handicapper a lot more accurate in the simulations in 2015.
What to expect of the 2015 MLB Season?
In the 2014 MLB Season Preview I wrote: “Since the season 2006 the average runs per game (RPG) has declined from 9.7 RPG to 8.3 RPG in 2013. Based on my own projections I believe that this trend will continue this season. The average RPG won’t probably fall dramatically, from an already pretty low level in 2013, but I still expect that the RPG will be lower in 2014. This is a very important thing to consider when handicapping games, if the league average RPG is a parameter in the model.” As I predicted RPG in 2014 was lower than in 2013 and actually at the lowest mark in 34 years. I don't think there is much room for RPG to further decline and I believe, that they should be at about the same level in 2015 as they were in 2014.
According to many MLB writers and analysts the trend amongst MLB front offices seems to be a more bullpen-oriented approach to pitching than before. What this could mean is that a bigger part of total innings pitched could be given to relief pitchers. The 2014 Kansas City Royals is of course the ultimate example of successful bullpen usage, and with a fresh example in mind, some teams might try to use the same pattern in 2015.
As I predicted, there were more Runline-picks in 2014 than the year before. Even if the outcome was bad and the picks didn't return as I would have hoped, I still believe that betting on Runlines should be the main focus for MLB-bettors. Runline-bets are a bit more interesting from a handicapper’s point of view because they are a bit more difficult to handicap than Moneyline-bets, and therefore it should be, at least theoretically, easier to find value from them. Runline-bets are more difficult to evaluate because in addition to the measurement of the teams relative strength to each other, the run-environment (ballpark size, weather etc.) of the game should also be measured. Depending on what kind of run environment the game is played in 1 run or 1.5 runs could be worth more/less.
The goal for the season 2015 is to achieve a 104 % rate of return on all published picks. I am probably not going to publish as many picks as in previous seasons, but I hope The Handicapper will fill the hole and provide some value to other MLB-handicappers.