I thought it would be interesting to look at the Rockets schedule in a graphical way. I came up with this: The blue line shows the "strength of schedule" (SOS) for the last 5 games. It is an average of the effective winning% of the last 5 opponents, adjusting for whether we're playing them at home or on the road. By "effective winning%", I'm referring not to actual wins/losses but rather a win% formula based on point differential for each team. The red line is our "effective winning%" over the prior 5 games -- again, based on point differential. Anyway, hope that's clear. You can see from the blue line that we've obviously hit a dip in the schedule, and we're taking advantage of it. But, its encouraging to see that we actually started playing better in late November, when the schedule was actually getting tougher. To me, this is a positive sign that we might be able to sustain our success even into following month when the opponents start to get very tough again. Hopefully we'll have hit our stride by the time of that road trip at end of Jan, early Feb. The schedule lightens up post all-star break, and if we pull of a trade at that time hopefully we can go on a nice run and put ourselves in good position to make the playoffs (for those of you that actually want us to make the playoffs, that is).
Great graph -- thanks for this. Interesting to note that we were unable to capitalize on the initial dip in the SOS.. it's like we were playing down to our competition.
can you clarify this for me? i'll admit, i'm not understanding if this is for the last five games or an overlay displaying strenth of schedule based on opponents current win % vs eff win% to date.
How are you calculating SOS? Is it a changing breathing input value or stagnated based upon today's rankings? The reason I ask, is the initial dip seem to correspond with the NOH game, and they got off to a very hot start, and were like 10-1...which meant our SOS at that time should have gone UP instead of down. If you are doing it based upon today's records only, then to me it would seem to be too stagnant. Does that makes sense? DD
Your point makes sense. Another way to put it: is a team's strength early in the season better quantified by their record up to that point, or by their record over a larger stretch of games (including future games)? I went with the latter approach, in part because it was easier to compute, but also because early schedules are out of balance plus the sample size is small so that could distort how good teams really are.
Sure. The SOS line describes how strong last 5 opponents are based on their current record up to his point in the season. It is a simplification, someone may have a better idea of how to depict it, but I think it serves the purpose here.
Good post. While the schedule was definitely easy recently, it seems the Rockets are playing better on an objective level (regardless of schedule) since they were losing to some of the same crappier level of teams earlier in the season. Interesting to see if they survive early to mid January and come out with at least a competitive (.500-ish, or maybe even .400-ish) record when playing against the better teams while continuing to take advantage of the expected lotto teams.
Yeah, but this is one of the reasons that I think stats sometimes give you an incomplete picture. Injuries skew the data, trades, hot streaks, opponents SOS, etc..etc..etc... Either way, I admire the work you put in. Nice job. DD
The "incomplete picture" line is a favorite of yours, even though I've never maintained that any stat (or any thing for that matter) gives you a complete picture. Not sure why you're so hung up on that. Back in your thread on Budinger's stats in the last 5 games, there was nothing complete about that picture either. You could say "incomplete picture" in response to every thread that's ever been created.
Sorry if it came off as an attack, that is not what I meant....and I agree with you about Budinger being a small sample size, thus me putting the qualifier of "Last 5 games" I just believe that data generated like this is not really valuable as it doesn't account for so much, it is a very small sample size of what the inputs should be to get a more complete picture. Again, not an attack, just the way I look at these things and trying to explain it. As I said...good work...and I appreciate the effort. DD
Sorry if I am not doing a very good job of explaining myself, I just think you would need a lot more variables to input in order to get a more complete picture. Anyway...I don't feel like arguing this with you Durvasa, this is your excercise, if you or others value the data generated, then great. DD
In theory a rolling average removes noise but still gives you a more up-to-date value. OTOH, one team's win % is entirely unrelated to another, so why are you averaging them? As an example, the MIN game in FEB looks much harder than it really is because you are averaging in good teams with that data point.
My purpose in using a rolling average wasn't simply to remove noise. The data points are meant to represent performance over the prior 5 games (red line) and how tough the competition was in those prior 5 games (blue line). Taking the data points at the MIN game as an example, those don't represent just the MIN game but rather that game and the prior 4 games.
Good work Durvasa. I like seeing what our strength of schedule is like for January and beyond. It doesn't take into other factors such as the back to backs but if we can play .500 or close to it through early Feb we should hit a roll afterwards based on our recent performances against an easier strength of schedule should all variables remain the same. I think with Brooks coming back, less Budinger in the line up (hopefully TWill gets time, Jordan Hill playing better, Lowry healthy, and just overall team chemistry this team is still capable of 45 to 47 wins or so and the 6th/7th seed.
