This time around they have taken to looking at personnel early and often for the Broncos, hoping to avoid the end of season workload that they had to suffer through last year. Curious about their early start this year, I asked them about what went down last year in the situation room (the room where they do their super computer analysis of the Broncos), and what was to be expected this year. Head Analyst was kind enough to take some time and answer my questions.
styg50: You've said things need to change for the NPLB analysts this year. Could you explain that?
Head Analyst: (sighing heavily) You should have seen it in there last year Styg. It was a mess, with half cocked theories, endless calculations and feathers everywhere. We made the mistake of getting started on the analysis much too late... with only 3 or 4 weeks left in the season if I recall correctly. Our first report was after the Chicago game I think. Part of that was, of course, funding. It wasn't until you got that NPLB Bill passed in the MHR Congress that we were really able to sink our teeth into the numbers, but we should have been organized way before that. But I don't want to make excuses, I just want to get the analysis in. To that end, I'm getting the troops armed early this year.
styg50: I'll say. Don't you think pre-season is a little TOO early for this kind of thing? I mean, we don't even know who the team is going to consist of yet.
HA: Now is the perfect time. It is a chance to test the theory. If it meshes with the results, the players who make the roster, then we have got something we can work with. Better to know now than when the Broncs are on their way to New York to play the Bretts... Keep in mind this is going to be a much different analysis from last year. We are striving for a good comparative structure for the reports on the players. That means trial and error. Theory just won't cut it.
styg50: You mention comparative structure. Care to elaborate? I know last year the focus was on drive comparison, so what exactly is going to change?
HA: This is really the heart of the whole system. Last year we worked from the drive charts. We were trying to establish trends and find the signature inconsistencies that would help point us in the right direction for our analysis of the film. We got lucky early when the analysts, on a hunch, focused on Matt Lepsis. At the time no one was questioning the ProBowler, but I think a lot of the work the NPLB did was able to shed some light on that issue. That saved us a lot of work, which was essential to turning out reliable reports at the time, as late in the season as it was. Can you imagine if the hunch had been wrong? How much time we would have spent breaking down non-critical film? Total we broke down 10 games for 2007, 4 in situ, and 6 from earlier in the season, 3 of the worst and 3 of the best.
styg50: And this year?
HA: We want to remove the need for a lucky hunch from the equation. That means tracking EVERY player, every down, every minute of the every game. More work from week to week, but less risk of overworking at the end of the season when the answers are most needed. The early numbers say that this is a borderline playoff team. Our goal is to present a record of this team that is unassailable. Win-Loss records can be questioned. Stats can be questioned. We want a document in hand that serves as a definitive answer.
styg50: Lofty goal.
HA: Perhaps. All I know is we have about fifteen coffee and coke-zero loving analysts who think its possible.
styg50: So where does the comparative structure come into the picture?
HA: The work the analysts are doing right now consists of essentially scouting every play, and every player for the Broncos. DTs are scouted as DTS, QBs are scouted as QBs. They are graded individually for technique, and the game itself, the context if you will, is graded ond overlaid on their evaluations. The second part is a work in progress and will be ready for the home-opener in oakland, but the personnel evaluations are going in full force. After we have individual grades for each player we establish an average, or baseline, for that player, that we subsequently compare to their evaluation, which we score on a rising scale (i.e 1-10 or whatever, it is different for different positions). The result is a percentage, a number universally consistent from player to player.
styg50: I see. With numbers for different positions that mean the same thing, you are able to compare them. But why would you want that?
HA: That is the comparative structure. When a DT can be compared to a QB, or when a MLB can be compared to a tailback, you gain perspective on who is contibuting and who isn't. Instead of relying on a hunch you have objective data to direct your focus. The hope is that the comparisons will allow you to cut through the mess of the game logs we are recording, and find the exact moments and players that require deeper analysis. The logs are huge. Sifting through them at the end of the season is too late to be truly objective and useful, but with a tool like the comparative structure, finding what you are looking for will be fast and easy. And accurate. Above all, acurate.
styg50: Sounds like a real asset. Speaking of game logs, huge is a good word. The one you turned in for the Dallas game was 36 pages long! Impressive figuring that you don't have any thumbs... I'm looking forward to breaking it down using the NPLB Comparative Structure (CS). I figured I would start with OLine and QBs.
HA: Starting at the begining is always a good idea. Good luck and GO BRONCOS!!!
Coming Tomorrow: NPLB Offensive Personnel Report--OLine and QBs!