The biggest issues I had with the Heat were the same ones everyone else harped on: lack of depth and interior defense. I was concerned about their offense as well, since Lebron, Wade, and Bosh all seem to love shooting contested 20-foot jump shots instead of looking for easier shots.

Going forward, I will offer another lucky guess - the Heat will continue to struggle for the short term, until Riley is forced to fire Spoelstra and take over. Eventually the Heat will figure out that the best way to win is to give Lebron the ball and get out of the way - and his talent will lead the Heat to 50+ victories and into the second round of the playoffs.

Alas, their weaknesses will catch up with them and they will fall short of the Finals this season. But there's always next season!

Making good predictions was my job. If my predictions were bad, I lost money - simple as that. Fortunately my methodology - which combined gut instincts, Vegas lines, identification of market biases, and statistic work - was effective. It's remarkable how fast one can improve when forced to put your money where your mouth is.

FYI, I have come across a number of professional gamblers and learned how they made good predictions. In general, the ability to sense biases in the betting market was far, far more important than the ability to build statistical models. Even the so-called "old-school" handicappers and the number-crunchers were starting to develop this skill (at least those who were successful).

Didn't Vegas have the Heat being the odds-on favorite to win the title this season?

I appreciate your posts Dsong and it seems that you are open to stat analysis. The main issue I have with some of your approaches to the rankings is that it is based a little more on "gut" than reason. It's easy to say "Well this might happen even though I can't prove it", then have it actually happen and then say "See, I told you this would happen! You can't trust the numbers!" Unless you had a crystal ball hiding before your back I didn't know about, I'd have to say that it's not necessarily having the inside track about something that you can actually PROVE using analysis, but that you simply made a lucky guess. From there, reason often gets thrown out the window and then it becomes a contest of who can get luckier guessing outcomes than taking empirical approaches to the matter. No one is going to sit here and say the models are perfect, but isn't that better than saying "________ will win the tile because, you know, I just feel they have that chemistry that other teams don't?" It's subjective and can mean many different things depending on the person you talk to.

By the way I haven't seen any purely stat models either here or over at APBRmetrics that have the Heat winning 70+ games either. I've seen high 50s-mid 60s win totals but I don't recall 70+ games.

The thing that I'm annoyed by are the people who create sophisticated models then treat all results from them like the gospel. We all knew that Miami wasn't going to win 70 games - and that the Lakers was probably the correct favorite (or at worst, a co-favorite). I appreciate all the work you and Justin do for the site, and hope you continue to make improvements in your analysis and models. One thing you seem to do well is admit address the limitations of your models! That mindset will prove valuable going forward.

I want to stress that I think the best pure numbers system would do better than any "gut instinct" method (witness the various experiments where computers outpredict human experts). The problem is that the SRS prediction system on the front page is hardly that system, and I can't say whether it would be in our company's best interests to develop that system.

I would be interested to see how the SRS system does stack up, though, if only just for curiosity's sake.

http://www.percentagesports.com/

http://probabilityfootball.com/2006/function.html?

I think this will be a very interesting experiment.

However -

The so-called "advanced methodology" has that same problem. Every model comes with a set of assumptions that may or may not hold. There is always a need to made adjustments and for a healthy dose of skepticism when interpreting results. You, Neil, are one of the few who are willing to accept that there are limitations to your methods and that some of your results are poor. The majority accepts their numbers like religion, even when actual events prove otherwise.

As for my own model - I have no magic formula. My models are very simple and based almost strictly on win probabilities and power ratings. The numbers pulled from "thin air" were derived from a good understanding of betting lines, years of experience and high-level analysis.

If you wish to know more - I would recommend reading David Friedman's blog. My methodology is very similar to his and we both try to look beyond statistics and examine what wins basketball games and championships.

Models are very useful but they are there to aid your brain, not to replace it. There is no substitute for experience, and your success with models depend much more on knowing how to use the model rather than the quality of the model itself.

Make no mistake, I love using models and have used them extensively in my own analysis. But sometimes you need a reality check - and understand the concept of garbage in, garbage out. For example, if your model predicts 68-75 wins for the Miami Heat, you have to be willing to swallow your pride and admit that the prediction is absolute junk.