# Introducing “Running the Numbers” at DallasCowboys.com

Jonathan Bales

I told you yesterday I had some big news coming. Starting today, I will be doing a blog at DallasCowboys.com called “Running the Numbers.” My first post is an analysis of historic draft data, both from a Cowboys and league-wide perspective, on the five positions I deem most suitable for the ‘Boys tonight and tomorrow. **Click here to read the full analysis**.

The conclusion of that study is that that two specific positions are undervalued in the draft, and they happen to be two spots the Cowboys could go with either of their first two selections. I think the analysis provides some insights on which prospects the Cowboys might be targeting most heavily tonight.

As far as The DC Times goes, I will continue to post here regularly. I will post to my **“Running the Numbers”** blog about twice a week to start, so continue to check both there and here.

I am headed off to Radio City Music Hall for the big night. Don’t forget to check the **Twitter account** periodically for any updates.

No related posts.

Related posts brought to you by Yet Another Related Posts Plugin.

I’ve always liked your articles based on numbers. This one wasn’t an exception. Just one quick suggestion/comment: the numbers in the graphs can’t be read well due the blue and black contrast.

Congrats JB!

You are greatly needed over there. You will bring facts and data to support you stances rather than “feelings”, which is what that site is aching for..

Outstanding. I wholeheartedly agree with Craig. I’ll be reading your posts regularly. I won’t bash their other writers, but suffice it to say, they made a wise decision in bringing you aboard. And I’m sure that your posts will be very popular.

Cesar–Thanks a lot. Yeah, I should have altered the color…will do for next time.

Craig and Mark–Thanks a bunch. Appreciate all the support over the years.

I think it’s great. In fact, I would say it’s about time. You are needed there because so many people there don’t understand statistics and think they are just excuses to fit whatever their opinion is.

The staff writers there aren’t statisticians and when they do try to use stats to show their side of things, it comes off like they are sucking up to Jerry. Their attempts fail because they don’t put the time in film study and analysis.

I am always posting links to your articles to show others how stats work for you when the time, effort and knowledge of what you are looking for are put into the right perspective.

Keep up the great work and I will enjoy reading your posts here and at the official site.

Jonathan,

Interesting article. Here’s a question I hope you will consider.

What is the better option:

Option A: a 50% chance of getting a $100 payoff

Option B: a 100% chance of getting a $30 payoff

here’s a related question.

you have two lotto scratch off cards you can purchase. either card would cost $10.

Card A you win 50% of the time. Card B you win 10% of the time.

Multiple choice.

a) Card A is better, you win more often

b) not enough information, you need to know the payoff for each card

Michael–Thanks for the praise and continued support. I look forward to posting some material there you guys find interesting and unique.

Valmont–My choice to the second question is B. Certainly you won’t pass up a 10% chance of $10 million for 50% of $100. The first one is tricker, and I think it depends on your current bankroll. For example, assume the “tipping point” for your goals is $100, and extra cash becomes incrementally worthless as you pass that number. If you already have $70, a guaranteed $30 payoff might be right for you. On the other hand, if your current bankroll is nothing, you might need to gamble on the $100 payoff. From a pure statistical standpoint, all other things being equal, the 50% chance for $100 is the better option.

If you are going where I think you might be, I see similarities in your examples and drafting, especially in the later rounds. Would you rather have 100% chance of a good sixth-rounder who makes your roster and plays special teams, or a 25% chance of a future All-Pro? For me, it is the latter. The late rounds are the time to maximize upside, whereas early on I want to minimize downside to offset costs. Let me know why you posted these…I am interested.

Jonathan,

Let me put this all together for you.

In the first case, assume you have a sufficient bankroll. You’re right. The correct answer is winning $100 50% of the time. The expect payoff if $50 vs. an expected payoff of $30.

And you’re correct about the 2nd question as well. You have to know the payoffs. You can’t talk about probabilities without know the payoffs.

How is this relevant? This is exactly analogous to drafting.

Option A: 50% probability for a $100 payoff, or

Option B: 100% probability for a $30

Except for the draft, replace probability with (1 – bust rate), and replace the payoff with what???.

Now here’s what I hope you’ll consider. For your payoffs you’ve used AV. But a top 10 AV for a guard is not the same as a top 10 AV for a QB. They’re not comparable.

Here’s a quick thought experiment. Let’s say the Colts did absolutely nothing but (a) get Peyton Manning back at 100% or (b) sign Carl Nicks.

In the Peyton at 100% scenario I think the Colts are arguably an 8-8 team. In the Nicks scenario the Colts are probably still a 2-14 team. Much much different payoffs.

You haven’t discovered a glaring inefficiency in the NFL draft. Guard’s having a lower bust rate doesn’t prove they’re undervalued. To prove they’re undervalued you have to put the bust rate together with the payoff (to do that we would need a statistic that allowed comparison between positions … something games won over avg)

and to be clear, the point of the first question is that a higher probability isn’t better per se … a lower probability with a higher payoff can be a better option.

Valmont–I think one of your points is strong and a solid criticism of my study, and I think another I addressed at least partially. Let me start with the weakness: I agree you can’t use AV to compare players among different positions (which I noted) and the importance of different positions (guard versus QB, as you point out) needs to be factored in. In no way do I think top guards should be taken ahead of the top QBs.

However, I DO think the study can measure both bust rate and payoff. When you measure how many players at a position end up in the top quarter of all players at their position, you are effectively measuring bust rate. With a first or second-round pick, I think being in the top quarter of all players is reasonable. Anything worse is a bust, albeit of varying proportions. When a player is in the top 5% at his position, though, I think that’s a solid measure of payoff. The top 5% of players are all elite.

For example, since 2000, the top 5% of cornerbacks include just 14 players: Asante Samuel and Nnamdi Asomugha at the top, and Ike Taylor and Sheldon Brown at the bottom. By measuring different levels of success, I think I effectively tracked payoff and bust rate.

I hope that didn’t come across as critical. I thought that your article was good. Better than I could do.

I just think the last step is to be able to connect the probabilities and payoffs. Now that’s a difficult step because we don’t have the statistics that allow one to determine something like a ‘win share’ for football players .. which is what you really need.

But in the absence of being able to precisely quantify payoffs/win share, you might want to consider what conclusions are justified.

I think you’re on the right track … you are attempting to quantify the probabilities and payoffs. You might consider that until one can relate AV to wins you’re not quite there. If the payoff for a top 10 AV QB is +4 wins and the payoff for top 10 AV guard is +1 win … you’d want to capture that.

Not at all..this site is all about being critical and questioning popular notions. As far as win shares, I think Advanced NFL Stat’s WPA is the closest thing we have.