MLB and NBA Update (5/9/11)
Update: Monday, May 9 at 4:35 PM ET
In this week's blog, we'll cover MLB and NBA performance, continue to hone in on some successful MLB strategies and get a little personal. First, I'd like to direct your attention to the NFL Draft content I just put together Ranking the Rookies and Ranking the Draft Classes. To me this provides my favorite blend of the deep, analytical, cross-methodology statistical approach that helps us provide the best inputs for our simulations and really sets us apart from most in this field as well as the optimistic nature of the NFL Draft - particularly in a year without much optimism otherwise. Check it out, let me know what you think and if you have any questions... And no, the Bengals' rankings amongst the top picks has little to do with me living in the area (I'm a fan of whomever we are picking - and the Green Bay Packers). In fact, it's ultimately surprising to see Cincinnati doing well in our numbers considering the team's unfortunate history in the draft, lack of GM (especially one with any football acumen) and paltry scouting department.
For baseball, it was another week with much of the same. On the whole, looking at every playable pick (all 200+), it was a down week to varying degrees for the money-line, over/under and run-lines. That's obviously not the intent, however, there are some very positive signs among the top plays. Logical strategies for our information are rounding into form. We'll cover each type of play individually below... Not too surprisingly, what worked in football - before we got caught up in daily products and daily debates - seems to work best for baseball. Simpler is better.
In fact, the first 35 games or so of the MLB season, which would translate into the about the first three weeks of the NFL, have performed eerily similar to what we saw last September in football season. Through three weeks in the NFL, we were 17-17 ATS and 15-15 O/U with playable picks. At mostly -110 odds, that's not actually profitable. It was not as much of a concern at the time because: 1) our top 3 picks each week - the "Paul's Picks" - were 6-3 ATS; 2) "Locks of the Week" were 3-0 and 3) it was obvious that it was early, meaning that the general assumption was that, as the season progressed, the in-season data would add value to the projections and we would improve on the whole. The NFL numbers (through the playoffs, but ignoring Week 17 - which I would correlate to being similar to the NBA, but we'll get to that later) finished 117-90 ATS (57%) and 103-83 O/U (55%) on all playable picks, topped off by an 11-0 ATS NFL Playoffs. Better yet, 60%+ confidence plays in the NFL, which would translate into about an $80 play or higher for a normal $50 player, went 62-38 against the number (62% - I literally broke out the calculator before I realized that's not too tough to do in my head).
Are we going to finish the season on an undefeated streak like that? It's doubtful. We are just over 20% of the way through the season and everything should improve going forward. Over/under picks got off to a fast start and have stayed pretty strong. Money-line picks are hitting over 50% (which actually doesn't mean that much), but haven't yet caught fire (save for the first/trial week) and the run-line has shown signs of great success, yet is coming off a couple rough days. We only now at a point when what a player (who has been healthy all year) has done this season means as much, if not a little more, to his statistical inputs than what we thought he was going to do based on his profile heading into the season (that's not quite linear; some players who have greatly exceeded or fallen below expectations are not yet 50/50 real numbers vs. statistical profile). There is no perfect answer to weighting those elements of one's numbers (that anyone has uncovered to our knowledge), so there will inevitably be some lag every year where we do not trust that a player has changed as drastically for the season as he may actually have. Weather has also been exceptionally poor to start this year and likely will not be so volatile in the summer months.
Anyway, the point is, that, while we're not quite to "get rich quick" levels by playing all playable picks, top picks are doing well, over/unders have been very good and we should only get better. For those reasons and others, the strategies discussed here get back closer to what we focused on in football each week - top plays win. All plays should also win, but are riskier and can cost one value against top plays. I think we have proven that there are enough strong plays over the course of the season (or even within a month subscription) that this approach can be both profitable and consistent, without being too conservative or monotonous.
However, there may be something to it, since we are generally on away teams in those circumstances. It's not a huge sample size, but 16 of our 22 normal+ money-line plays - including two tonight - have been on road teams. I'm definitely not saying that I think this is wrong. We have diligently researched true homefield advantage, ballpark factors and natural advantages of batting last. The closer the expected outcome, the more advantage goes to the home team when the game is played once as opposed to 50,000 times and that has dampened performance to-date (only a month of the season - and the losses definitely seem more frequent and painful in those situations than actually have been so far). We will continue to dig and monitor to see if any modifications are necessary (my best guess is that no further changes will be necessary and that we'll see those outcomes and our picks even out).
As noted above, money-line performance has not yet matched the success of the over/unders. Overall, we have hit 51.2% of our playable ML picks. That can be good, bad or otherwise due to a wide range of money-line odds. In this case, it has not been profitable. But that's playing 432 games thus far (over ten a day). Normal or better picks - any time in which the suggested play is greater than or equal to the normal play - are 18-10 on the money-line, which is +$287 for a $50 player. That's not quite one play a day, but it is about six per week. Until the ML picks prove more valuable on the whole - which should happen as time goes by - patience is your friend. Ignoring all non-normal+ plays is not required, but focusing attention on normal+ plays should prove fruitful.
A few weeks ago, we discussed the success of stronger O/U, suggesting that around 55% or 56% confidence in picks would provide a successful threshold that also generated a healthy number of plays a day. Then, we added O/U odds. This is still a smart strategy, but 55% and 56% mean something a little different now. A 55% confidence against -110 odds almost perfectly correlates with a "half-bet" or $25 for a $50 player. On the whole, O/U plays are hitting 55.2% of the time and have been lucrative (+$832 on 299 plays). Normal or better O/U plays are 20-14 O/U (58.8%) and +$254. The "half-bet" threshold seems to provide the best bang-for-buck balance. Picks where the wager amount has been $25+ are 93-59 O/U (57.7%) and +$724. That's just about $100 less in profit on half the plays (approx. four per night). We'll effort to make the half-bet threshold more obvious to aid in identifying.
Going into the weekend, "half-bet" or greater run-line plays were 73% ATS and making a healthy profit (since we started tracking 4/26). Then something very bizarre happened on Saturday and Sunday. With a significant number of home underdogs and our (rightful) propensity to favor underdogs on the money-line, we ended up with several strong (not normal+, but still strong) run-line plays where we liked the home team +1.5, yet the road team on the money-line. Decent money-line days without many close games hurt performance. I don't like that. I'm guessing/hoping it's an aberration. Up until that point, we had had great success with +1.5 ROAD teams and were rarely in a situation where our money-line and run-line picks were from opposing sides. It's one thing to play value for value, but it became an issue when almost all of our plays were that way yesterday. Of course, there was only one game where the road favorite won by just one run, which in itself is an aberration. For now, you all should feel comfortable chalking this weekend's odd plays and our performance in those games to a bit of a fluke. I'll keep digging and thinking and tinkering though... because that's what I do.
In total, run-line plays over two weeks are 53.8% accurate, but not profitable due to the odds typically laid. Normal+ run-line plays are 5-3 ATS and profitable (+14 for $50 player). "Half-bet" or better run-line plays are still pretty strong at 37-22 ATS (63%), which is profitable and good to note, but not nearly as good as they looked three days ago. Personally, I think the half-bet+ approach is the way to go for O/U and RL (it generally provides outcomes that are 50%+ likely to happen, has performed well yields about four plays a day), but I know we're still young into RL tracking and, theoretically, our RL performance should be more closely correlated to ML performance than anything else.
In one of the most frustratingly ridiculous (many of you think I'm talking about our NBA picks for the last week right now...) chain of events that I have ever endured (more below), an untimely system restore (to default settings) robbed me of the database I was using to track NBA Playoff performance. Fortunately, not much else was lost. Someone may have been trying to tell me something. Of course, I can go (and will go/have gone) back through to review our numbers, but I know they are not good. In five of the last seven days (since the last blog), our information has failed to generate a winning ATS play.
That's typically not how I prefer to look at the numbers. I usually think of our prediction as as close to true/right as possible; where, if we get picks wrong, the opposing team just happened to hit its X% chance of pulling off the feat (i.e. luck). If that is as true as I prefer to believe, for the second year in a row, we've been really, really unlucky in the NBA Playoffs.
I cut my proverbial teeth forecasting the NBA. I wrote my Master's thesis on Measuring Individual and Team Effectiveness in the NBA Through Multivariate Regression. The first simulation engine I ever wrote was for the NBA. I'm proudest of my work modeling the NBA and would put our simulation engine up against any in terms of comprehensiveness, logic, accuracy and technological capabilities. I've consulted with NBA teams and written many research-based NBA essays for various sites, journals and books. I've got every NBA Draft and NBA All-Star Saturday since the early 1990s taped or DVR'd. David Robinson was my favorite athlete growing up... This is a different league (particularly in the playoffs). That's why it pains me so greatly to see our information and the game itself struggle.
We have to hold steadfast our assumptions that players will try their best, especially in the playoffs. We have to assume that officials will call the game as it should be called and not with any biases (for whatever reasons those may occur). And, if anything, as we have seen before from teams like last year's Celtics, we have to believe that the more talented and experienced teams are more prone to taking their games to a new, better level in the playoffs if/when they have to. After last season's NBA Playoff performance, I spent countless hours and nights research historical in series, game-by-game variation, tempo changes in the playoffs, officiating, etc. in hopes of performing much better this post-season.
It's a small sample size, yes, but this playoff season has seemed to belie those assumptions and that effort. The Lakers completely fell apart. The Mavs can't miss. The Hawks (maybe the most surprising team of them all) found new life. The Bulls, who finished second in the league in scoring margin and got the easiest draw of any team in the playoffs, can't blow anyone out. And the officials seem respite to apologize for missed calls after the games rather than get them right during them. Add in the unlikelihood of a 3.5-point underdog turning a double-digit fourth quarter lead into an ATS loss (which has actually happened multiple times this playoffs - most recently with Oklahoma City on Saturday - one of our biggest plays of the playoffs thus far and it looked like close to a sure thing until about three minutes to go) and the last week of the NBA playoffs has been an unfortunate combination of surprising and embarrassing (for us and the league).
We'll keep grinding to figure this out. I just wish the players, coaches, officials and league did their parts. And rest assured, we will do what we can to prove our mettle and improve the user experience for those who have subscribed to our basketball information.
I don't usually like to add in personal notes, but I'll make these quick in hopes that I can get some help (those who follow me on Twitter may have already seen this). I have an issue on my personal laptop in which my internet browser connection (DNS Server not found is the issue) gets interrupted for 3-5 minutes just about every time I go into (and attempt to go elsewhere from) Yahoo!. Unfortunately, I run a (money) fantasy baseball league through Yahoo! so it's hard to avoid. Every time I want to go check my teams or the league, I have to live with being kicked offline for a while after (not cool).
It happens in IE, Firefox, Chrome and even the PS3 browser (when that worked). I've run every malware/anti-virus software that I - or anyone that I have talked to has ever heard of. My ISP and the router company placed blame elsewhere and would not help. Yahoo! only responded with canned answers that were completely irrelevant. Dell forced me to do a system restore back to defaults, lost about 1/4 of my personal files and then, when the problem still existed, said that they would try to fix it if I paid them $240. My sanity may be worth that at this point, but I could not stomach the idea of paying anything more for the company that sent me a computer that has never worked to try to fix the computer with no guarantee of success. I've talked with all of my friends and colleagues on the tech side and posted on Twitter and my personal Facebook. And finally, I worked with someone who seemed really knowledgeable at GeekSquad. The guy spent about five hours working on my computer, but couldn't fix it. He said he would call back the next day after doing some research and never did (that cost me $100). A new router, which I suspect may do the trick (but which one?), would have been cheaper than paying GeekSquad not to fix my computer. Unless someone can help, I'll go in that direction, but there are no guarantees there either. Anyway, I have spent an inordinate amount of time just trying to figure out how to access my fantasy baseball league.
And lastly, I really enjoy music, particularly live music and with a eclectic taste. On Saturday, after the Memphis-OKC debacle, I caught Incline District live and loved them - so much passion in what they do and they're unique... That's it. Saw a good band; wanted to post a link. Thanks for indulging me.
As usual, if you have any of your own suggestions about how to improve the site, please do not hesitate to contact us at any time. We respond to every support contact as quickly as we can (usually within a few hours) and are very amenable to suggestions. I firmly believe that open communication with our customers and user feedback is the best way for us to grow and provide the types of products that will maximize the experience for all. Thank you in advance for your suggestions, comments and questions.