DG probabilities will be up within the next couple hours. We’d like to hold off longer but are hearing a fair amount of grumbling from subscribers. I figured it would be useful to start a dedicated thread on this as it’s an important topic.
For this week, I won’t be surprised if the odds that are already posted by bookmakers are pretty in line with ours as this is a new course, so things can be pretty well-calibrated from last week’s betting markets. But, there are a few players crossing over from the European and Korn Ferry Tours which always makes things more interesting.
Based on what unfolded last week with the euro odds, I’d be in favor of holding back for a day or so. It would be a really interesting experiment that you could try - hopefully the grumblers won’t be too mad over one week
Random question, though, that I think I know the answer to…have you ever thought of any realistic ways to keep the books prying eyes off the site? Essentially ensuring that your subscribers are true end users/bettors. I imagine it would be nearly impossible to identify them versus another average joe, but just wondering if you’ve ever thought on it
The obvious way would to be to have separate membership licences for personal and corporate use. Compliance officers will not be keen on their employees using the sites data without a licence then.
Would mean legal documentation and tracking of users. Maybe the most likely economic return for Datagolf is to serve the industry rather than those using it for betting. I get the sense that betting funding the site is a happy byproduct of the statistics rigour and interest in the game and how it can be analysed.
As for when to post the mode output, I’m happy to see experimentation. Whichever approach you take, my plea would be for a specific time to be disclosed in advance understanding that sometimes shit happens and you won’t be able to make that time always. Evening could be anytime from 2:30pm to 11:30pm depending on the context.
No matter what happens the race to lock in wagers will not end, and eventually books will raise the juice to ~30-99% on the T5/T10/T20 markets (up from ~20%-99%).
The juice on the cut lines are also unreal, I’m seeing spreads like -1200/+480 and -220/+140.
We might as well enjoy the edge as long as it lasts because the juice will reach the point where horse racing starts to look pretty good.
I think it had to do with bookies’ adjustment based on datagolf predictions. I’m guessing that the late release for the European Tour caught them off schedule and they didn’t adjust the odds every 5 minutes while increasing the juice 10-15% on every single golfer on the board on the T5/T10/T20 markets.
I think a fair chunk of the DG edge comes from course fit adjustments. If DG posts early, before bookmakers post even, then the bookmakers just suck the course fit adjustments into their numbers for free, without even taking a bet. By waiting we are more likely to see them post odds which are pretty far off, especially from a course fit perspective.
I posted some of the odds from last week in the ‘anxious about future model improvements’ thread where this topic was starting to come up. Again just before first tee time am not seeing anything to suggest anyone is purely copying datagolf.
What I suspect is happening is that many golf models are feeding into the pricing, so we do see cuts on players datagolf have identified as value - but those cuts will happen regardless of datagolf due to other models out there. Datagolf probably accelerates those cuts being a publicly available model that drives various users bets (users that are being identified by the books as smart money too), but I don’t think is the sole cause.
I was already seeing smart money coming for some players datagolf highlighted before datagolf released their model this week.
In my opinion all that delaying does is give other models a head start at beating the market… and consequently handicaps datagolf users not having an equal start.
I don’t have anything to back it up but I’d suspect the same. As incredible as the DG product is, I personally doubt that it’s users are solely responsible for moving the lines.
I think it depends on the market. Matchup markets are clearly still influenced by a lot of independent people / models, of which we are one; exchanges I know there is still a lot of discrepancies with our model. But the bookmakers’ outright and finish position markets seem to be different. Pinnacle will now open their T5/T10/T20 markets and our model will show zero value, which is only really possible if they are heavily influenced by our numbers. DK it’s pretty much the same. I’m not sure how long you’ve been following our stuff, but a year ago it would not have been unusual for the model to show +20% expected value on 15-20 outrights, and +10% EV on a similar amount of T20s. (Maybe it was more than a year ago – but things have gotten noticeably worse in 2021.)
But, as you point out, books still don’t copy exactly. In certain situations they will always be different (e.g. Koepka at majors, Spieth / Leish / Westwood recently), and I’m sure the responsive books still move their lines if they get some large bets on a relatively unknown player (whether we like that player or not). Even this week there will be usually be a few guys who hold their positive expected value all the way to close, but in a market with 156 players, many of whom have pretty long odds, you would expect to see a lot more large discrepancies (at least at opening) if our model was independent of the price-setting process.
Just focusing on outright odds. I have only recently signed up so don’t have the history, I guess the most relevant current point is whether to post or not -
Even if certain books are factoring in your model surely they will just do so as soon as it is available anyway? So delaying posting would only delay the inevitable whilst giving other models a headstart at the value.
If they are copying there is also an argument releasing as soon as possible is beneficial, knowing there are some outliers in datagolf v the exchange. Not saying too much on a public forum but that could yield a big edge, hell you could even throw a couple distortions in and really milk them!
Also on my point of markets moving before datagolf model i found an extract i pulled which i think was a while after the model was posted and has DK on. 7 of those had 20%+EV and 10 with 10%+EV - but from those I saw money for Werenski, Huh, List & Vegas before the model was posted.
@matt_courchene any thoughts here eight months later? It does seem like the books are waiting to post anything except outrights until after DG posts (except DK which protects itself). For Puerto Rico I would be very curious if you held everything till Tuesday/Wednesday if there would be value in the T10s/T20s. I think one tournament is worth the experiment.
