Mid-June Review (6/13/11)
Monday, June 13 at 8:24 PM ET
It's been a little over a week since I last checked in. I apologize for the delay as I have been waiting for the NBA Finals to wrap up to review the NBA Playoffs. We've also been working diligently on baseball run-line plays and I have been waiting until I had a more definitive update for those before getting into great detail on the performance and future of run-line picks.
Off the top though, I thought that I would address a very fair sentiment posed through a support contact after the last blog. The gist of the note was that the user had not been winning at the rates described in the last blog and was frustrated with the tone of the blog. This is a perfectly logical response in that situation. As we have talked about before, unless we win every play or lose every play for a particular day, there will likely be some (hopefully most) who win with our info and some (hopefully few) who do not (even if they have implemented smart strategies - especially in the short-term, there will be ups and downs). This is even more evident in sports with daily plays such as baseball. In general, the blog is intended to educate, inform/report and relate.
Here is my specific response to the contact: While I totally understand your sentiment, I'm genuinely trying to help. Of course it is not fun to see me talk about where we are winning (and not be), but I view it as my responsibility to report what has and has not been working. Collectively (obviously it's always our focus internally), finding legitimate trends is always a positive. If we are genuinely really good in some circumstances, we want our users to know that. If we are not doing well in some circumstances, we want to note those, our suspected reasons and what we are doing to address any issues.
There is reporting on what has happened and there is recommending what to do based on what we think will happen. With the game picks, we do the latter every day (the former will be available daily and in automated fashion with the Trendfinder starting for football). The reason that I brought the half-bet strategy to the blog last month was to note the level of confidence (relative to odds) that generates an appropriate number of plays a day (and to illustrate that, up to that point, those higher confidence plays had been successful).
It's very unfortunate that that strategy did not generate great results last month (trust me, I understand this more than anyone), but I don't think it's an invalid strategy. We're looking for optimal bankroll management and that's as close as we can get... I believe in the numbers we publish and I think the strategy you employ should be a profitable one. (Editor's note: Fortunately, this strategy - playing ML, O/U and RL games where the recommended wager is greater than half the normal play - has generated +$501 for a $50 player in June, most of which has come from a very strong month on the ML... we'll get into more on June performance below - after the NBA...)
NBA Playoffs Recap:
Congratulations to the Dallas Mavericks (not many NBA Finals stories are leading with this today). The Mavs did what we originally only gave them a 5.7% chance of doing when they made it through the NBA Playoffs as the league's champion, knocking off four teams in the top 11 of our final NBA Power Rankings. Even by the time Dallas got to the NBA Finals, we still did not love the Mavericks' chances, only giving the team a 33.8% chance of taking out the Heat (chances that looked even more grim after Games 1 and 3). But, ultimately, Dallas did exactly what we said it needed to en route to upsetting the team we had as preseason, pre-Playoffs and pre-Finals favorites: the Mavs pushed tempo - and won.
Before Game 5, radio host and star of the "G-Bag Nation" on 105.3 the Fan in Dallas, Gavin Dawson, asked me about the Mavericks impressive play in the fourth quarter. Here were my thoughts based on what we had seen in the numbers:
"I think there is some legitimacy to it (the tremendous fourth quarter play by the Mavericks in the Playoffs). We're never going to expect a comeback like we saw in Game 2 of the Finals, but there are logical reasons Dallas succeeds in the fourth quarter (regular season data notes that the Mavs, despite being essentially league average in the other three quarters, had the second best fourth quarter margin of any team in the league – behind Chicago which was good in just about every quarter).
First of all, the Mavs are the deepest team in the league with respect to quality players. Even with Butler out, the Mavs consist of Dirk, two pairs of two good players at PG and C (when Haywood is healthy), two great shooters (Terry and Peja) and two good defenders (Marion and Stevenson). In the fourth quarter, when many teams are slowing down, resting stars early and dealing with foul concerns, Dallas is usually at full strength. It's not a knock on guys to say that Kidd and Barea and Chandler and Haywood are interchangeable. Dallas gets 48 minutes of strong play from those positions.
Secondly, and most importantly, the team that is losing in the fourth quarter often has to force tempo. Dallas may not force many turnovers or run the break much, but the Mavs are one of the most efficient teams in the league early in possessions. No team is more efficient 0-10 seconds into the shot clock (non-fast break plays). I've been saying that the way to beat Miami is to speed them up – not run, but increase possessions. The Heat have a great halfcourt defense and a very short bench. The more aggressively and quickly the Mavs play, the harder it will be for Miami to set its D and the more likely the Heat are to get in foul trouble. I don't get why Dallas has not forced tempo. I understand that Miami is trying to do the opposite, but the Mavs can't win consistently at Miami's game. The Mavs need to set the tempo early to cause the Heat problems.
And lastly, we discussed the need for Dallas to shoot well from outside to win. The Mavs have the talent to do so. In desperate situations, those shots have to be taken. In the cases where the team has made those comebacks, those shots are falling (that's a combination of luck and skill)..."
It may not have been what we thought was going to happen, but it happened the way we thought it could. In Games 5 and 6, Dallas played its game, put up over 100 points and won. Congratulations Mavericks.
In review of the entire NBA Playoffs, our performance was a little below average on the whole, yet strong for those who remained patient and only played our stronger plays:
NBA Playoff Performance:
- Normal+ Plays: 7-3
- Top Plays of the Day (ATS and O/U): 36-29
- ATS Playable Plays: 26-35
- O/U Playable Plays: 23-22
- Last Two Rounds (ATS and O/U): 15-10
As has shown to be the case, even with the significant amount of improvements designed to handle such circumstances, playoff series (as opposed to single elimination) do not seem to bring out the best in the Predictalator. Performance trended in a similar way to last year, though, with respect to the performance of stronger opinions, there was improvement. Normal or better (or even 56%+) plays in these situations just aren't easy to come by.
As we prepare for next season, our first full season of basketball, we look forward to building the best possible regular and post-season basketball engine, data and info that we can. As can be noted from our 2010-11 Preseason predictions, on the whole, the Predictalator looks really good.
After a rough last two weeks of May, the performance of the money-line picks has been strong enough and the over/under picks steady enough to put us in the black on the whole for ML and O/U plays for the season (we'll get into run-line plays in a second). As we have discussed in previous blogs, while bad luck, weather and the high propensity for extra inning games played a role, we also got a little ahead of ourselves with the manipulation of in-season baseball data (something we had never had to do professionally). In an attempt to stay out ahead of the competition and adapt quickly, the mix of current versus historical data was likely too aggressive. We researched the topic and worked very hard to find the right mix, ultimately settling on the current approach that, in its present state, was finalized and implemented two weeks ago (as soon as we found pieces to the puzzle that we knew were improvements, we put them in). With Interleague play fixes and some updated homefield advantage information, the engine was also updated around that time.
Since then, all playable money-line plays are +$500. Most importantly to me, is that, in those 132 games of data, there do not appear to be significant trends among the picks that are of concern (please contact us if you see some). The 28 plays between +125 and +200 (no bigger dogs have been chosen) are +$118 on an average recommended wager for $50 player of $25. The 31 plays from -145 to -200 (typically an area where we have struggled) are +$91 on an average recommended wager for $50 player of $19. Our 70 plays on favorites are +$181 with an average of $25. Our 53 plays on underdogs are +$237 with an average of $20 (especially since we are suggesting the "to risk" amount, we're naturally going to recommend greater average wagers on favorites we are confident will win strongly than underdogs we think are undervalued). And in the remaining nine pick'em games (-110 or -105 plays), we are +$82 on a $25 average recommendation. Normal+ ML plays in June are 5-0 and +$203, bringing the season total for normal+ ML to 29-15 and +$397. The ML plays without big profits is with the "Upset Watch" plays. There have only been five this month and the return is profitable, yet at just +$16.
Over/under picks in June have essentially been break-even. They're 54-53 overall, 27-23 on the "half-bet" (any recommendation greater than $25 for $50 player) and 2-1 on normal+ plays. That's nothing to get too excited nor discouraged about. In this case, there are a few trends worth noting that speak to many of the topics we have discussed in previous blogs. Scoring is up relative to public perception. In the 31 games where we liked the Over on lines of 7.5 or less, our plays are +$112 (this includes both normal+ wins). In the 25 plays where we liked the Under in games with lines of nine or greater, our plays are -$54 (just 11-14 overall). Our Overs overall have generated positive value and our Unders negative value. While I'd prefer to generate positive value for everything, I'm okay with this result given recent picks. Instead of picking the Under in the most recent (yesterday and today) Boston-Toronto and New York-Cleveland games (with lines in the 9 to 10 range where we had typically loved the Under), we have generated one Over (which covered) and two "no picks." More specifically, games in the last two days with lines of 8.5 or greater have generated seven Over picks. Any lag related to scoring - no matter what the reason - does not seem apparent now relative to where it may have been a week ago. Based on the positive movement we have seen from our other predictions and what I believe to be the natural negation of a previous trend (which had to have also been somewhat based in luck), I'm confident we'll take "break-even" for the month back solidly into the black.
It's an oddly different story on the run-line where picks this month are down $119 despite being 54-49 on the whole. Most notably among those picks, our normal+ run-line plays are down $90 even though they are 4-3. Normal+ RL picks on the year are just 12-12. The concern has not necessarily been with the performance of the plays, but more with the fact that we are consistently choosing home favorites (RL favorites, which is the ML underdog) with our strong RL opinions. Of our 103 playable RL plays in June, 81 have been home teams and 61 have been on ML underdogs (getting +1.5). The trend was similar for the last couple of weeks in May. I'd love to say that we were on to something that exploitable, but it was probably the other way around. With the added juice and extreme odds associated with these plays, profitability requires a very high rate of success (our "half-bet" RL plays are hitting 58% and have not turned a profit).
Run-lines were the focus of my work over the last ten days or so. When we introduced the run-line picks in late April, it was because we had seen verifiable proof that they were succeeding - and at a much more profitable rate than our ML plays. As time has gone and the market and our engine have evolved, RL plays have not been as successful as O/U or ML. While ML and O/U performance can so easily be impacted by the data, RL are far more connected to the engine because RL plays rely so much on how and when teams score. After much research, testing and deliberation, we've come up with some engine tweaks, particularly late in the game, that should improve the performance of our RL plays without impacting the integrity of the ML and O/U plays. Yesterday marked the first time this month and just the fifth time since we launched the RL play table in late April that we chose a road ML favorite (giving 1.5). And fortunately, the Atlanta Braves won 4-1 at Houston. We aren't on any such plays tonight, yet there is only one opinion that warrants a play greater than $18 for a $50 player.
The diversity in plays and generally weaker opinions we have seen in the last two days is encouraging. We believe the updated engine, as it relates to RL, should provide less formulaic and more appropriate results. That being said, since we originally launched daily RL plays on what we deemed a trial basis to determine if we could significantly add value without adding confusion to the product, after an as-yet-undetermined amount of time following this engine update, if we do not feel we are providing added value with the run-line plays, we may take them down and focus our bankroll management recommendations on the ML and O/U (where most people prefer to play anyway). Run-line plays can always be evaluated in the Customizable Predictalator. Look for an update to this blog or in an additional blog soon.
As we continue to monitor baseball, I will now shift the majority of my attention to one of my favorite times of the year. For the next few weeks, I will be going through all 120 FBS college football teams to set expected rosters, depth charts, coaching strategies and other data inputs for each team in preparation for our college football preview that will launch in July and the season, which starts September 1. This is a little earlier than usual for me to be focusing on college football instead of the NFL, but NFL rosters are too undetermined right now (not to mention the final practice, preseason and/or regular season schedules). No matter what happens with the NFL, we are hard at work adding many new features for the football season including: the TrendFinder database, parlay and teaser evaluations in the Customizable Predictalator, schedules/results, in-depth statistical rankings, playoff/final regular season projections, a play analyzer that will allow all lines to be updated and plays to be re-ordered like the current articles and more.
The Golden Nugget recently released lines on some of the more anticipated college football games of the year. While we are not quite ready to make predictions on any of these games, I do have some thoughts based on the line movement and my initial review of the college football landscape:
- With Florida in a down/transition year and with several marquee players on the team, I assumed we would be high on South Carolina this year. However, I don't assume we'll be as high on the Gamecocks as most of the sharps who bet the team heavily last week. The team's defense will be one of the best in the country and Stephen Garcia (if he can stay on the team) and Alshon Jeffery are adequate playmakers for Steve Spurrier's team, but the offensive line replaces three starters and Marcus Lattimore's hype seems unfounded (rushing a lot does not mean rushing well).
- Boise State should be really, really good this year, but I have a hard time believing they should be more than 14 point favorites against TCU. TCU in general is not getting the love that it should.
- Most publications and pundits have Oregon ranked third or worse in preseason rankings, but the books and sharps seem to agree with my feeling that the Ducks are as much a championship contender as any team (I was actually a little surprised to see Oregon open and remain a favorite against LSU in Week 1).
- If Wisconsin is still favored by three (or more) when I head to Camp Randall for the Big Ten opener against Nebraska on October 1, I will be shocked.
- Brady Hoke inherits 17 returning starters and has a clue about what he is doing. The line movement away from the Wolverines seems to be much more extreme than warranted (interestingly enough, except for the game against Hoke's former team and maybe the most underrated team by these lines and movements, San Diego State, where I like Ryan Lindley and the Aztecs to keep it interesting).
- Texas is better than Missouri (in Missouri), yet I don't think that Texas is better than Oklahoma State (in Texas).
- All of these early opinions may change when we actually run/simulate the games (both for our preseason preview and in those weeks).
And lastly, a bit of a public service announcement: don't watch the Top 100 NFL Players for 2011 (or at least until the top 20 or if you love silly reality game shows). The weekly unveiling of ten of the 100 players through well-produced breakdowns of the player's journey through the words of a fellow player or coach, followed by the reaction show represent a brilliant television move by a league looking to save some face and provide cheap programming during uncertain labor times. But the whole methodology is ridiculous and illogical. Even if we ignore the obvious - there may not be an NFL in 2011 and if there is we don't know who is going to be playing on what teams or at all - the premise of the show does not represent the nature of its approach.
To create this list of 100 players for 2011, NFL players were asked to rank their top 20 players "for 2011" (whatever that is supposed to mean). The NFL compiled those votes and ranked the top 100 based on votes stating the player is viewed by someone as a top 20 player. Yes, this means someone put Donovan McNabb (#100) in his top 20. That's not the only issue. This also means that if every player in the league agreed on the 21st best player for 2011 (remember, there are 32 teams in the league, 11 starters on offense, 11 starters on defense and special teams specialists meaning that 21st best is still in the top 3% of all starters and the top 1.2% of all active players in a given week), that player would not make the top 100. In other words, as a ranking, it's a completely useless list until the top 20 (the intent of its approach).
As alluded to, I equate this to silliness of reality television game shows in the mold of American Idol - and not just because of the NFL Network's abuse of social media, desperate hype and incorporation of "fan voting." In those shows, people vote for their favorite and the contestant with the least votes is kicked off. In that case, if everyone in the country had a different favorite, but the same second favorite, the second favorite would be eliminated even though he is everyone's second favorite contestant. Meanwhile, the terrible contestants continue on (likely due to friends, family, hometown, etc.).
For the NFL players, it was likely hard enough to get them to take this seriously for a top 20, let alone for 100. So laziness has led to nonsense. Watch the show if want to see highlights of professional football, but not to get worked up about who is and/or should be ranked where. That's not the point.
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.