Writers of Pro Football Prospectus 2008

29 Aug 2011

How to interpret risk?

How am I supposed to interpret the risk ranking (Blue, Green, Yellow, Red)? Is it a measure of standard deviation? The instructions say that injuries, age, etc. are included in its calculation, but I assume that these are already built into the projections (i.e. expected value). If it is just standard deviation, than presumably I should be risk-averse for high draft picks and risk-seeking for lower picks, in which case values should not all be adjusted downward.

Posted by: mbroshi on 29 Aug 2011

4 replies , Last at 09 Sep 2011, 3:28pm by Arjen.Robben

Re: How to interpret risk?
by Tutenkharnage :: Tue, 09/06/2011 - 11:23am

That's my interpretation, too. If the projections are averages of 10,000 simulations, then it stands to reason that if two players have the same projection and both hit it, the player with higher risk will, on average, outperform the less-risky player. Why? Because the subset of simulations in which the riskier player meets or exceeds his projections is smaller, so it stands to reason that these are better projections. (If they weren't, the two players' projections would likely be different, not the same.)

Can someone at FO tell me if this assumption is correct?

Anyway, I agre with mbroshi's assumption--avoid risk for high draft picks if you can help it, but feel free to court it for low draft picks. In other words, make sure your starters are reliable, and take flyers on your back-ups.

Re: How to interpret risk?
by Aaron Schatz :: Tue, 09/06/2011 - 11:38am

Actually, that's not quite true. We built it to look at how often players in the past fell short of their projections, not how often they surpassed them. Remember, more players will fall short of projections than surpass them, especially at the top of the projections, because of injuries and changes in usage. (This is one of the reasons why we project more receiving totals for each team than we do passing totals for the quarterback -- somebody is going to get injured, lose a job, etc.)

So it's a combination of two things.

1) Injuries, how often the player has played 16 games in recent years, etc.
2) Does a projection based entirely on objective variables fall short of our projection that includes subjective variables such as "expected role."

Re: How to interpret risk?
by Tutenkharnage :: Fri, 09/09/2011 - 11:06am

Thanks, Aaron. Can you tell me, then, which of the following more accurately represents "expected value"?

a. The non-risk-adjusted projection
b. The risk-adjusted projection

I'm having a hard time understanding this, and I'm sure I'm not the only one.

Re: How to interpret risk?
by Arjen.Robben :: Fri, 09/09/2011 - 3:28pm

My best guess:

The risk-adjusted projection is the final "expected value" after subjective (judgment) adjustments downward.

Given the components listed by Aaron [injuries (downside) and diminished playing time (downside)], I would think of it as a measure of isolated downside volatility rather than traditional "risk" measured by standard deviation, which would include upside.

I never quite understood this either. Thanks for asking the question.

Login or register to post comments