+- Beyond the Bets Sports Betting Forum (http://www.beyondthebets.com/forum)
+-- Forum: General (/forum-general)
+--- Forum: Sports Betting (/forum-sports-betting)
+--- Thread: The @MLBPriceMachine - 2013 Edition (/thread-the-mlbpricemachine-2013-edition)
The @MLBPriceMachine - 2013 Edition - mount187 - 04-09-2012 01:33 PM
I talked with BTB about posting this here, and wanted to let everyone know that the brain child of myself and @lindetrain has gone live at http://www.mlbpricemachine.com. The site is free of charge, so please don't kill me as being a "tout". We feel like it should turn out to be a pretty valuable resource for people here, and will also be looking for some feedback from you guys to make things work better. The site itself is in a somewhat beta stage, but the numbers we're working with are coming out as we intended.
When looking at the lines we post there, keep in mind that there is no juice applied, so when you see Team X -125, that makes Team Y a direct inverse of that (+125).
The way we've been using the model is to take any line overnight that provides $.10 of value in our favor or more (5Dimes offers $.05 overnight lines, so you can get real value early). So far it's given us mixed results. We're beating line moves all over the place (Opening Day III we made 8 plays with total line moves of $.99 in our favor), but that hasn't born a lot of fruit to this point. I believe we're roughly 5-13'ish, but have played almost all dogs; still a loss, but not as bad as if we were chasing big faves early. Like all bettors, we took some bad breaks with late runs / blown saves, but those should even out.
I'm sure more questions will come up, so feel free to ask and I'll try to answer them as best I can. Feel free to ask here, email us at mlbpricemachine@gmail.com, or on Twitter @MLBPriceMachine. Again, suggestions / bugs fixes are welcome, so don't be shy. This should be my only post on the site, so fear not, I won't be treading on BTB's forum as a way to "tout" my own work.
Good luck!
RE: The @MLBPriceMachine is ready. - SkyBlues42 - 04-09-2012 01:42 PM
Why are the Red Sox 100% to win against the Blue Jays?
Forget that. I take it, it's because there's no starter listed for the Red Sox?
RE: The @MLBPriceMachine is ready. - Beyond the Bets - 04-09-2012 01:44 PM
If you guys get a chance, check this out and provide Mount with some feedback. I think it has the potential to be an extremely useful tool.
RE: The @MLBPriceMachine is ready. - mount187 - 04-09-2012 01:46 PM
@SkyBlues42:
I made a blurb about it on the "Home" page of the site; since we're working with projected numbers right now (we'll be integrating season stats in incrementally), we don't have projections for some players. Doubront was projected as a reliever for us, so if we put him in as a starter, the BOS bullpen would get evaluated for ~8.0 IP, which we don't want. If you notice on the Schedule page, BOS's starter is listed as "Pitcher", and on the detail page, the line is "Infinity". I should have a fix for this in tonight that says no line is available.
[Edit] That is correct.
I'll be updating the site with fixes / enhancements as quickly as I can. Same old, "Real job, wife, kid, blah blah ..." story, but I'll do my best.
RE: The @MLBPriceMachine is ready. - TMoney72 - 04-09-2012 02:00 PM
Just checked it out for a couple of minutes. Pretty interesting idea. When I have a little more time I will dig into it a little further.
RE: The @MLBPriceMachine is ready. - ChiefZilla - 04-09-2012 02:01 PM
That's great, mount. I like it.
I particularly like that the site is aimed at finding value and beating the closing number as opposed to W/L. If you make certain that beating the closing number is the primary objective, I believe the W/L will take care of themselves.
RE: The @MLBPriceMachine is ready. - mount187 - 04-09-2012 02:07 PM
(04-09-2012 02:01 PM)ChiefZilla Wrote: That's great, mount. I like it.
I particularly like that the site is aimed at finding value and beating the closing number as opposed to W/L. If you make certain that beating the closing number is the primary objective, I believe the W/L will take care of themselves.
Everything I've read / been told by people "in the know" says exactly that. If you can continually get the best of the closing line, you'll be profitable long term.
My main priorities will be a) keeping the stats updated to produce sharper lines, b) bug fixes, and c) any enhancements that make the site more user friendly. The first major enhancement will be finding some way to make sense of our "Tracking" page so we can see if we're accomplishing our goal. The "Tracking" page right now is there more as a placeholder, and as proof to people that we are tracking our numbers; we just don't have a good way to display them in a way that makes sense.
RE: The @MLBPriceMachine is ready. - lindetrain - 04-09-2012 02:57 PM
Serious props must go out to Mount on this. Can't tell you how many emails I've received at 1-2 AM ET from this guy, with a wife and kid at home; has literally busted his ass on this site and created a pretty great tool. Take advantage of it all you can, as it's not every day you get access to something that can consistently help you beat the closing line.
RE: The @MLBPriceMachine is ready. - mount187 - 04-09-2012 03:09 PM
One thing that we'll have to figure out how to handle is lineup changes. We have a very easy UI for changing lineups, that has an immediate impact on our projected line, but neither myself or lindetrain have time to stay on top of them on game day. I may be enlisting the help of an admin to help with that process at some point. In some instances it makes a huge difference in the line (Berkman and Posey are both out today for STL and SFG), so it's worth noting; in some cases it doesn't make much difference at all (MIN's lineup changes only moved LAA's line ~$.02).
RE: The @MLBPriceMachine is ready. - Beyond the Bets - 04-09-2012 03:26 PM
(04-09-2012 02:57 PM)lindetrain Wrote: Serious props must go out to Mount on this. Can't tell you how many emails I've received at 1-2 AM ET from this guy, with a wife and kid at home; has literally busted his ass on this site and created a pretty great tool. Take advantage of it all you can, as it's not every day you get access to something that can consistently help you beat the closing line.
congrats on the national title, friend.
RE: The @MLBPriceMachine is ready. - BetATL - 04-09-2012 03:39 PM
Looks good guys.
RE: The @MLBPriceMachine is ready. - VincentValour - 04-09-2012 05:33 PM
Great job guys, looks like it's crushing opening numbers.
RE: The @MLBPriceMachine is ready. - MrLomez - 04-09-2012 05:38 PM
I think this is great, and really similar to what I'm doing -- though I'm sure from cross-checking the lines, that my weights are a bit different (not better, just different).
One thing I noticed from the "About" section is that you're using defense independent pitching stats to come up with expected Pitcher Efficiency. However, I don't see any mention of defensive stats to counter that. Of course, each lineup will have a unique Defensive Efficiency that will either improve or detract from a Pitcher's Efficiency.
Jeremy Hellickson is a perfect example. His defense-independent pitching stats are middling. But given TB's defense, he's much more effective than those numbers will ever suggest.
Likewise, LAA's lineup with Trumbo at 3B will definitely have an effect on a groundball pitcher like CJ Wilson.
I've found that Defensive Efficiency (good or bad) can account for up to a run +/- in any given game. Usually it's a small fraction of a run, but it's something worth including in your model, IMO.
RE: The @MLBPriceMachine is ready. - mount187 - 04-09-2012 06:27 PM
Without giving away too much of our process, we do account for defensive effects in our pitcher efficiency values. It's probably not as in-depth as some people would want, but again, we're not trying to predict the actual score (or even winner). We're just concerned w/ valuing the teams relative to each other, which we think the process should do. You definitely make a valid point, and that's one of the reasons we don't use the formula to predict totals.
RE: The @MLBPriceMachine is ready. - ScottKimberly - 04-09-2012 06:30 PM
(04-09-2012 06:27 PM)mount187 Wrote: It's probably not as in-depth as some people would want, but again, we're not trying to predict the actual score (or even winner).
So I'm unfamiliar with the site. Can you give me a paragraph in layman's terms on what you're trying to do, if not predict scores or winners? I'm very interested and would like to know more.
Sorry if this question seems to elementary.
RE: The @MLBPriceMachine is ready. - mount187 - 04-09-2012 06:55 PM
(04-09-2012 06:30 PM)ScottKimberly Wrote:(04-09-2012 06:27 PM)mount187 Wrote: It's probably not as in-depth as some people would want, but again, we're not trying to predict the actual score (or even winner).
So I'm unfamiliar with the site. Can you give me a paragraph in layman's terms on what you're trying to do, if not predict scores or winners? I'm very interested and would like to know more.
Sorry if this question seems to elementary.
The goal of the site is to come up with a valuation for each team's odds of beating each other, expressed in percentages, and then converted into American odds lines. The theory being, it's not in our best interest to try to predict how athletes are going to perform on a given night (picking a winner). Our goal is not to say, "The Orioles are going to win tonight," but rather to say, "If the Orioles played the A's 100 times with these same lineups and pitchers, we think they'd win 45% of the time." If we convert that 45% into a m/l, and it offers more value than what we have to pay at the book, then we think that's a good investment. Even by our own valuation, we think the Orioles are probably going to lose, but we also think over the course of 2,400 games in a season that we'll hit enough of the 45%'s to make money.
We also don't think that you can use stats to predict the score of the game. There are WAY too many variables to do that. You can come up with a similar model to say, "We think these two teams will score more than 7.5 runs 58% of the time," but predicting Team A to score 4 and Team B to score 6 seemed out of our reach, and seems kind of unrealistic for anyone to do (in my opinion). At some point we may add to our model to try to gauge the probability for over/unders, but we're not doing that right now. Anyways, we're just trying to come up with a number for Team A and a number for Team B that provides us a way to judge them against each other. Those numbers may come out to 12.50 and 15.25. Those would be unrealistic scores, but could be very realistic measures of the teams against each other. If we're using the same model to judge both teams, and the model is good at creating measurable values as it relates to winning percentages, then we think that's all that is important. In the end, you could come up with a model that says, "If you take fielding efficiency and multiply that by batting average, and divide the result by stolen bases, you can use that to create a model that provides you with win percentages that will get you on the right side of the line 60% of the time." That number may not come out anywhere near what a score would be in the game, but if it gives you accurate probabilities, that's all that matters, right? That's what we're trying to do.
I hope that made sense.
RE: The @MLBPriceMachine is ready. - ScottKimberly - 04-09-2012 07:03 PM
I think I understand, and I love it. As I understand it, it seems that you have the percentage they should win, which is put into a moneyline (X), and the moneyline in the sportsbooks (Y). If X indicates a higher probability of winning than Y, then it is good value bet. Is that right?
I really like the idea, especially if, as I think, it is a way to identify great value plays in the sportsbook (in terms of risk/reward).
Thanks for the clarification, at least as far as I understand it, and please let me know if my interpretation is a complete misunderstanding.
RE: The @MLBPriceMachine is ready. - mount187 - 04-09-2012 07:46 PM
(04-09-2012 07:03 PM)ScottKimberly Wrote: I think I understand, and I love it. As I understand it, it seems that you have the percentage they should win, which is put into a moneyline (X), and the moneyline in the sportsbooks (Y). If X indicates a higher probability of winning than Y, then it is good value bet. Is that right?
I really like the idea, especially if, as I think, it is a way to identify great value plays in the sportsbook (in terms of risk/reward).
Thanks for the clarification, at least as far as I understand it, and please let me know if my interpretation is a complete misunderstanding.
You nailed it.
RE: The @MLBPriceMachine is ready. - itg_sports - 04-09-2012 09:07 PM
The site looks really great. The idea even better. Good job on this guys.
RE: The @MLBPriceMachine is ready. - Beyond the Bets - 04-10-2012 12:28 AM
I'm a few posts behind, but it's late, and so I don't feel like reading the ones I've missed. But, have you addressed the weight you assign to home field advantage?