Better thinking through technology

Current state of text-based projects



I don't post very often on the text-based machine-learning systems, and since that's been the focus of my recent efforts that largely means I don't post much at all. One of the big challenges with any sort of text-based machine-learning system is reading and comprehension. One of the reasons that system like IBM's Watson is considered a big deal is that it can scan a massive pile of texts and pull together all the ideas into a meaningful response.

[Read More]

FBS, Week 10, 2016



I was shocked to see this morning that almost all the line movement was in the direction of the modeled picks. That's a rarity. We shall see what that actually amounts to.

[Read More]

FBS, Week 9 , 2016



One thing worth noting this week is that I'm trying to creep back toward the more +value positions on the spreads. At the same time, I'm also trying to hug common spreads (betting around the -14 by taking the +14.5).

[Read More]

FBS, Week 8, 2016



Pretty much doing the same thing as last weekend.

[Read More]

FBS, Week 7, 2016



FTR, I'm trying to get back to what was working well a couple years ago. I think we can safely say the NFL-style QB model is an abject failure in FBS play.

[Read More]

FBS, Week 5, 2016



If you've been following the last year, you're aware there's been a pretty massive split in performance between the NFL and FBS models, with college games doing badly while the NFL games are killing it.

[Read More]

FBS, Week 4, 2016



Okay, we're now at the stage in the season where data should be getting priced in well enough to expect something from break-even to profitable on the day's bets. It's not quite adventure time, but it's getting close.

[Read More]

TNF, Week 3, 2016, HOU@NE



Okay, this is obviously a game that's on the wrong side of some of the protocols I use. In other words, this is a mostly a "play for fun" kinda game. Also, I suspect no one is going to sway the Belichick believers anyhow.

[Read More]

NFL Saturday games, Week 2, 2016



Not really in love with some of these higher lines, and that makes almost all of the alternate high lines a real stretch. Hopefully one of the juicier +300 type lines lands.


[Read More]

FBS, Week 3, 2016



Week 3 is, if trends hold up from previous seasons, the week where the bot starts finding its mark. Typically as teams start playing opponents whose quality of play is relative and new information starts getting priced into the model, the hit rate improves. (Given last week, it'll be hard for the hit rate to not improve week-over-week.)

[Read More]