Thursday, October 23, 2014

The Long Con of @RJinVegas: #TERNDZ (Part Two)


RJ Bell is the founder and CEO of Pregame.com, a sports betting website that includes forums, odds, contests, and picks (both of the free and for purchase varieties). It is one of the most successful sites of its kind, in no small part due to the popularity of Bell, who has accumulated over 80K Twitter followers and established himself as the mainstream media's go-to "expert" for stories involving gambling on sports.

An American success story if there every was one. That's only one side of it, though. In some corners of the internet (including this blog), there is a feeling that Bell has not found his success in a legitimate way. This is the 13th post in a series detailing the reasoning behind that particular feeling. If you'd like to start at the beginning, it's here.

Last week, some folks disagreed with each other on the internet. It started with an exchange between our protagonist and ESPN MLB writer/analyst Keith Law:

The majority of the discussion ended up being about the second half of this tweet, but I'll take a minute to touch on the first part. The Week 2 NYJ/GB line fluctuated between GB -7 and GB -8 at CRIS throughout the week. The line did indeed close at 7 most places, but plenty of bettors won with Jets +7.5/+8 (they ended up losing by 7). Bell knows this himself, as no shortage of his touts lost on Packers -7.5/-8.

Of course, with #TERNDZ, there's no room for nuance, so Bell just uses the "consensus" closing line. Which is fine, but for practical purposes it doesn't even make sense.

Forget the cherry picking, multiple arbitrary endpoints, and lack of predictiveness (which we'll get to). Let's consider a practical application. Say it's Tuesday, and Bell tweets out a #TERND for the following Sunday's game. When are you supposed to bet it?

In general, one could argue that you should bet something like this ASAP, before the market matures and has factored in all the relevant #TERNDZ. On the other hand, these stats rely on the closing line, and Bell himself says:
So if you act too early in the week, maybe that's bad because the public hasn't yet had time to react to the #TERND? I'm really not sure.

Trying to figure out how to use these historical, "situational" ATS records to actually bet makes my head hurt. The whole point is to take advantage of the public overreacting, but if this information was actually worthwhile, wouldn't the #shraps already be doing this? Wait, don't answer that, RJ's got it:


Oh okay. So we're not supposed to place wagers based on this info at all. Why's that?


Bell can provide example after example of "proof" of how going against the grain gives you an edge, but the funny thing is that it's all cherry picked. When pressed for anything beyond that, he either provides info that only serves to make your laugh:


Or moves the goalposts so far that Seabass kicking on the fucking moon couldn't reach them:


This is exhausting. To summarize, with (bullet!) points to which I don't believe anyone would object:
  • Bell's #TERNDZ use arbitrary criteria which are data-mined from the ~infinite potential combinations (teams favored by 3+ the week after losing to a team with a winged mascot, etc. etc.)
  • Bell's #TERNDZ use very arbitrary endpoints which are designed to allow them to be as eye-popping a possible (cutting off the data at W14 four years ago, etc. etc.)
  • Bell uses these two strategies to come up with #TERNDZ which will sound logical to the average fan, thus granting the #TERNDZ more weight/credibility
  • These #TERNDZ are not predictive of NFL ATS results going forward
To be clear: the stats/tweets that emerge from this methodology are, by definition, COMPLETELY WORTHLESS. If they came about naturally, rather than being data-mined, that'd be one thing. Or if they were predictive, that would also change the equation. 

But if neither of those thing are true (and they aren't!), we're left with one question: what's the fucking point? Data-mind historical trends that don't have any predictive value are a celebration of the wonders of randomness. Nothing more, nothing less. At least in this situation, this:

...is simply false.

At this point, Bell would argue that his approach is to let the "customers decide what they value". In fact, here he goes again:

Here we have a guy who runs a tout site acting like he's completely mystified by the idea that any would look at his tweets about ATS NFL #TERNDZ and infer that that information could be used for betting.

It seems we have two groups here. One group understands these stats are not predictive, considers them to be worthless, and retweets/replies disingenuously. These folks could aptly be described as the "pathological haters".

The other group infers that these stats have some predictive value, and/or are recommendations for what to bet on; RJ himself just explained he doesn't believe either of these things to be true. This group is responsible for RJ's sincere retweets/replies, of which there are many. These folks could aptly be described as "potential marks".

Any response, Randall?

Welp.

Previous Long Con post: "Repackaging" Info
Previous #TERNDZ post: #TERNDZ

2 comments:

  1. The bizarre thing is that he's made the exact opposite argument many times in justifying why he considers DEN to have been favored over NE last year, even though NE closed the fave at every sharp book. He claims DEN should be considered the fave because "most money was bet with DEN as the fave." Well most money here was bet on NYJ at +7.5 or better, and every single Pregame tout who released the game did so at GB -7.5 or worse. But he obviously could care less about consistency when all he cares about is selling shitty picks.

    ReplyDelete
    Replies
    1. Yeah, what was weird about that one was that the way he does it, the Broncos did close as the favorite. Of course, the way he does it is stupid, and Denver closed at pk or -1 at Pinny/CRIS/LVH/CG. But if you're going to be a clown and just look at Vegas shops and weight them all equally, at the 10 I checked it was NE favored only at Cantor, pk at three of them, and Broncos favored at the other six (-2 at TI lol).

      Which is what makes this tweet so funny. RJ is obviously a huge fan of changing his argument in the middle of the conversation, but this time he didn't even need to, and it went directly against other arguments he makes all the time.

      Delete