SBSQ #30: Will liberals turn against sports betting?
So far, the sports betting backlash has a bipartisan flavor. Plus, why I don’t use prediction markets in my models. And are Democrats going to blow the California governor's race?
This is Silver Bulletin Subscriber Questions #30 — the Steph Curry edition! Since we’re already two-thirds of the way through the month (hey, we’ve been busy…), let’s keep it to three questions. We’ll try to get back on schedule with SBSQ #31 in April. As always, I’d encourage paid subscribers to leave questions for next time in the comments below.
In this edition:
Why I don’t use prediction markets in my models
Is the sports betting backlash coming? And will it come from the left?
Will Democrats get shut out of the California gubernatorial race?
Why I don’t use prediction markets in my models
Andrew Bennett asks:
Question for #30 for Curry episode. Are you considering incorporating prediction markets data into your political forecasts models in addition to polling data? If no, why?
Hey Andrew, thanks for the question. I get this kind of question a lot, but with all the interest in prediction markets lately, plus models being very much on my mind these days1, let’s take a fresh look.
The short answer is “no”. We don’t account for prediction market prices in our models, and we don’t plan to.
That reflects no lack of respect for prediction markets. (Indeed, as you probably know, I consult for Polymarket.) Market prices aren’t easy to beat in the well-traveled terrain that Silver Bulletin models cover, namely sports and elections. But I think treating prediction market prices as an input would create more problems than it solves.
In fact, I’d go further: I think it would sort of defeat the purpose of building a statistical model. Our goal, both with our writing and our models, is to provide you with an independent perspective. Then treat you as an adult in terms of how much you might adjust your personal views in response to that perspective. I’d hope that people aren’t treating any model output, or for that matter any of my Hot Takes, as the gospel truth. But I don’t want to essentially bet against myself by pre-diluting my views. Even though I know that “out-of-consensus” views are fairly often wrong.
There are also some technical issues that factoring in prediction market prices would introduce. One is the potential for recursivity. If the markets are influenced by Silver Bulletin forecasts — which I doubt is true on the sports side but is plausibly true for elections2 — then the markets and our forecast no longer provide an independent perspective. Instead, you’d get into weird feedback loops: our forecast of the Georgia Senate race influences Polymarket, which in turn influences the Silver Bulletin model, and so forth.3 It could get messy fast, and it’s hard to build robust statistical assumptions around this.
A more subtle technical issue is that whenever you take two highly correlated inputs — for instance, Silver Bulletin’s statistical forecasts and prediction market prices or Vegas odds — it’s hard to ascertain their relative predictive importance when fitting a model. Some of our models are more “opinionated” than others. (For instance, ELWAY, our NFL model, often produces some divergence from Vegas, whereas our NCAA models tend to produce results that are very close to consensus lines.) But even for ELWAY, the projected point spreads it spits out every week are probably something like .95 correlated with Vegas lines. In cases like these, model specifications can be insatiable and highly sensitive to changes in assumptions.4
A related issue is that prediction markets or Vegas lines are a moving target. Let’s say our Silver Bulletin models are picking up on something sharp. For instance, so far this year in the men’s tournament, COOPER has tended to project slightly higher totals (i.e., over-unders, the combined number of points between both teams) than Vegas. If the Vegas total for a game is 143 ½, we’re often at 145 or something. I happen to like how we calculate our totals.5 It’s possible there’s some edge there. But to the extent there is, the market will eventually notice that its totals are coming in too low and adjust.
But mostly it comes down to what I wrote about above. I don’t know whether you’d say that Silver Bulletin is competing with prediction markets. To some extent, I’m sure we are in the sense that they’re both places you can go to look up a reasonable estimate of the odds. But I’m also sure that some paid subscribers are interested in Silver Bulletin forecasts in part because they’re interested in betting on them. And I think those readers are getting a better product if they get an original opinion from us, derived from a repeatable process, instead of something that bakes in a bunch of conventional wisdom.
Is the sports betting backlash coming? And will it come from the left?
Yung Capital asks:
Could banning Sports gambling gain support from the left? It’s historically been more a conservative position to be anti-gambling. But now with how our government operates, it was legalized because lobbyists. In 2027 will we see any presidential candidate on the left take a strong position against it? I would genuinely be a single issue voter over it. I hate the ads and what it turns fans into.
We’d flagged this question last week as a potential SBSQ candidate, but it’s newly relevant today. That’s because a bipartisan pair of Senators, California’s Adam Schiff and Utah’s John Curtis, just introduced a bill that would restrict prediction markets from offering sports-related contracts.
Alexandria Ocasio-Cortez also tweeted a criticism of Polymarket’s new deal with Major League Baseball on Friday; her tweet has drawn more than 10 million views and I saw it getting a favorable reception both from conservatives and liberals that I follow. Indeed, “hasn’t this betting thing gone a little too far?” is hardly a contrarian sentiment. McKay Coppins’s outstanding piece in the Atlantic earlier this month — I basically got to serve as McKay’s sports betting sherpa (but of course, he mostly ignored my advice!) — went viral in a way that stories rarely do these days. Even Silver Bulletin subscribers like Yung Capital are tired of all the ads.
But let’s back up a second. Here’s a map of states where online sports betting still isn’t legal.6
That’s a pretty interesting pattern, with a decided geographic skew. It looks a little bit like the 1976 election map, where Jimmy Carter basically won everything in the Eastern half of the country and Gerald Ford couldn’t make up enough ground out west.