Hi - love the site, it’s essentially got me really into golf! A few ideas I’ve had whilst using it re interaction:
Notes functionality - it would be really useful to ‘tag’ player performances or indeed have an overall notes section for each player; things like injuries or other news to provide some context for a performance e.g. new putter.
Course filters on player pages - would it be possible to add tours beyond PGA/DP at the bottom? Occasionally get some crossover with other events on other tours which seem to be captured by the course history tool but not here? Secondly, some broader filters would be interesting e.g. types of course (links, parkland etc), types of green or even ‘courses similar to x’ via the Course Fit tool
Bookmaker scraping - would it be possible to add the Betfair Exchange? (From a selfish point of view as that’s the primary one I use!)
‘My Bets’ area - enjoying using this already; once I record a bet I can’t then edit it, would be useful to do so; also, in line with the Exchange comment above, if those prices can’t be scraped a commission option would be good
And then some further thoughts re the predictive model:
Weightings - have you tried something exponential/decay related here? So EXP (-t/T) with t = days ago and T = constant term, so as a weighting on SG round-by-round or tourny-by-tourny, with a view to optimising the T? (Modelling far from my strong point btw but wondered if this might be a ‘clean’ way as something I’ve looked at before)
Course history / course fit interaction - perhaps more an observation than something to fundamentally change in the model; a couple of hypothetical examples:
i) Player with very good course fit but has missed the cut in 8 outings there
ii) Player with very good course fit and has top 10’d in 8 outings there
In case of i) does the history override fit?
In case of ii) is this a double count for purposes of prediction?
(I’m stronger on (i) than (ii) btw!)
Separate course history thought - do you consider previous performances at a course more/less repeatable based on the SG categories they did well/badly in?
Round 1-4 performance - are there players who systematically come out of the traps well or finish strongly (relative to pressure)?
Please excuse me if the answers to these can be found on the site - have tried to read as much of it as I can! Thanks
I use the exchange as well and it’s better if you scrape them yourself or use BetAngel. The delay from scraping and how the price fluctuates rapidly would make the Betfair exchange odds kind of pointless.
With that being said fuck Betfair, their commission is massive and they need to have more liquidity in smaller golf markets and their EW 1/5 T10 market sucks + no missed cut market on regular season events. Better than Smarkets though.
It’s mainly for pre-tournament where prices are static enough - I tend to start by looking at players DG deems value but often get bigger prices on the Exchange that won’t appear so have to find those manually. 2% commission not a total disgrace, hopefully one day win enough to enter Premium charge territory
Thanks for all the suggestions! Very thoughtful. I’ll respond to the few that are more in my domain, and I’ll let Will respond to some of the others if he wants.
I am working on integrating the Challenge Tour and DP Tour course keys, so those should be available on the site soon (I assume you noticed we were using Galgorm Castle’s data from a few of the Challenge Tour events it hosted? That was a one off). We could do broader groupings of courses – something to think about.
Possible I guess, we don’t have an account anymore so we need to get by that hurdle. We are in Canada.
I guess it would depend on the course and how large the fit adjustments are. In general course history adjustments are smaller than course fit, but a specific case will depend on the course. We look at a player’s course history after taking out course fit effects (i.e. performance relative to baseline + course fit), so in theory we shouldn’t be double counting, but I am skeptical of that sometimes.
We don’t account for this but I’m sure it matters a bit.
Again, not something we account for systematically but I wouldn’t be surprised if there are some meaningful differences across players. Definitely in round 4 it seems most people have strong opinions of who plays well when in contention. It also would be easy for us to check if over-performing in Round 1 for example is predictive of future Round 1 over/under-performance… should check that. It would be interesting.
Sounds good - yes Galgorm looked like it was getting picked up. Not sure how you read the courses in but happy to put together some sort of manual csv if that helps!
Understood - certainly more in nice-have territory that one anyway. Pinnacle usually a pretty good steer for the same purpose.
Thanks for the link, will try and wrap my head around it!
Sure - maybe the double count isn’t such a bad thing overall. Had another thought linked to course fit, or at least course similarity (it would essentially a big project!) - as much as tournaments played on different courses reveal what skill set ‘does well’ there, it feels like comparing actual physical characteristics i.e. dimensions/par#, surface (fairways/greens/rough), location/size of hazards etc would be more robust? Probably end up with some monster list that makes it hard to compare and would want different weightings on certain things too I imagine. Think though it would help where we don’t know much about a course e.g. can see which holes are like those already on the tour. I guess it’s just how much that gives you compared to observed performances.
Yeah I wondered if a hot putter at certain course was more repeatable than it would be otherwise.
Cool - yeah feels like a nice extra layer on the FRL pricing.
Saw an interview with McIlroy after his 1st round at the BMW the other day btw - how common is it players will play on the course before a tournament? Rory said he played there Sat-Wed every day after missing the cut at Southwind, felt a decent ‘experience’ bump! Also wondered what the indications were from the first couple of rounds this weekend what sort of course fit Wilmington is looking like? Looked like accuracy was tallying up well with SG OTT after R1 but distance back into things R2 (if that’s the right sort of interpretation)!
(Display has messed up numbering there but each point taken in turn!)
A thought linked to the round-by-round performances - does anyone know where you could find the ‘In Running Low’ price for each player (from any bookmaker really) in a given event? Say two guys finish on -7 and T20 having both started the event at 50/1, I prefer the guy that actually gave himself a chance of winning and went to like 5/1 IR than the one who played well in R4 when the pressure was off and was never really shorter than his starting price at any point. I feel like there could be players out there fitting the latter profile who look better on SG than they actually are; as a consequence, they could be much better top 10/20 propositions than for the outright win? I suppose it comes out in the wash by dint of the fact they don’t win tournaments or get close to it but it could provide some added context?
A very quick request here - would it be possible to add Date of Birth to the ‘Player List & IDs’ within the API please? Find it quite useful to have for context and would save me doing something manual!
Possibility to merge season results/alltime/grand total in the bet tool-page, and/or in general sort results by category (win/place/h2h/props) in an easy way.
…and also showing total number of bets and units at the bottom on the archive page:
I’ve mentioned this before, but the possibility to sort simulated groups by odds or name in the prop tool. Helps when simulating big groups. Preferable a “sort by…” next to “add country to group”
Haha it is… Will put it live a couple hours ago. I’m not sure about it… I thought maybe the load time should be longer so it doesn’t accidentally pop up too much. It’s also a step towards doing this for players, which will be very useful (on matchup tool and other places).
I did notice it by chance which could indicate that the load time is too low, but was more surprised than anything. No problem after that when browsing around.
Will definitely try to add some more functionality to the bet archive page soon. I realized I couldn’t even get my total bets easily when responding to that other thread, which is not great. Also would be good to break things down by tour / bet type.
