The system relies on older lines to derive new lines. Sometimes that mean that the system will spit out obvious nonsense, which is a lot of why we’re using human judgement. The system requires common sense sanity checks if it is going to work. The best way to have those is to know the pitchers and teams, and make sure that any inputs that are out of whack stand out.
901 Diamondbacks/Reds Over 8.5-106
This seems great, we agree on the moneyline and have a big edge.
PASSING: 903 Mets -132
Lucchesi is a reasonable pitcher, but we’re going off of one start and somehow he’s being treated like DeGrom, who is not pitching because he asked for an extra day of rest. I’m confused how this .334 rating got there, but I can’t take it seriously, and we’ll consider this game’s Aikido to be fully offline. No opinion, no wagers.
905/906 Marlins/Giants Under 8.5-105
Good edge, good balancing.
911 Mariners +155
Seems like a good number, hard for this to be too bad.
913 Angels/Astros Over 8.5-118
These are two very good offenses and the pitchers are rated as rather bad as well. When the difference is two full runs, it’s worth looking up pitchers, and no, rating either of these pitchers worse than replacement doesn’t make much sense, and given the moneyline the error has to be roughly evenly split here. I certainly wouldn’t go e.g. Over 10 in this spot on reflection even if that was the market line. But I’m still willing to go over 8.5.
915 Pirates +112
Pinnacle is offline but 5Dimes has the game, and it seems fine to assume this is about where the game is going to open and consider it a pick. If you’re doing tracking yourself, you can choose to wait for the Pinnacle opener, and we’d take +106 or better, pass if it got closer than that. In practice we’d also consider what other books had at the time on the margin (e.g. we’d always take +110 and pass on +102, but in between we’d care if Pinnacle had a substantial lean).
Final note is that I will be tracking the results in this spreadsheet, but won’t be saying what the current tracking result is that often. As I write this, I’ve only scored things through 4/19, and often this will be a few days behind to avoid spoiling Mets games for me - I could keep it current if I had reason to do so, but if anything I think not tracking too quickly adds perspective.
Note: Please do not get overexcited by early results. There will be big swings, there always are. But yes, as of Tuesday we were off to a simply fantastic start.
I like the idea of playing the odds given by the bookmakers against themselves. If you find inconsistencies and bet on them you should come out ahead if the inconsistencies are not based on something material. It also seems like much less work then trying to actually model everything yourself.
One thing I've been doing recently is arbing in play NBA moneylines and it works out fairly well however the limits on in game bets are fairly low on most bookmakers unless there is a timeout or something.
One other thing I've been thinking about is just pulling all in game data for NBA or MLB and then trying to bet on the swings. Let's say for example for NBA games that are within two points with 5 minutes left one basket results in an x% increase in that team winning but the bookies are moving the odds y% where y is much bigger than x. I think you could find an edge that way and again it doesn't require you to know much about the teams because the strength of the teams is already baked in. You are just trying to bet the relative change in odds after a given play.