What are the chances eh?

We have reached halfway through an exciting 2013 regular season, so it is time to add to the mid-season reviews that will proliferate the interweb football pages, both here and Across the Pond. Here I am going to look at what the standings and teams record may tell us, rather than how exciting they have been to watch. Actual statistics may be involved, just to warn the reader. Let us begin…

I’m going to defend using statistics, both descriptive and analytical, to assess teams and players. They really can give an accurate descriptor of how a team does. As an example, the final score of the Packers @ Vikings suggests a closer game than it actually was. A breakdown of the scoring shows the Vikings scored late on and the Packers had the game sown up by the 3rd quarter. As a lovely descriptor, the Packers had a 100% score per drive record (TD or FG resulted on each drive). As such the descriptive stats support what anyone who watched the game knows: the Vikings were swept aside.

So let us start by taking in the bigger picture, the current standings with win – loss and points for and against difference:

AFC East            AFC North      AFC South     AFC West

NE 6 – 2, +35;     Cin 6 – 2, +53;    Ind 5 – 2, +56;  KC 8 – 0, +94;

NYJ 4 – 4, -68;    Bal 3 -4, +2;       Ten 3 -4, -1;      Den 7 -1, +125;

Mia 3 – 4, -15;      Cle 3 – 5, -31;     Hou 2 – 5, -72;  SD 4 – 3, +24;

Buf 3 – 5, -37;      Pit 2 – 5, -28;      Jax 0 – 8, -188; Oak 3 – 4, -24.

NFC East          NFC North       NFC South     NFC West

Dal 4 – 4, +44;   GB 5 – 2, +54;    NO 6 – 1, +76;    Sea 7 – 1, +80;

Phi 3 – 5, -35;    Det 5 – 3, +20;   Car 4 – 3, +74;    SF 6 – 2, +73;

Was 2 – 5, -56;  Chi 4 – 3, +7;      Atl 2 – 5, -18;      Ari 4 – 4, -14

NYG 2 – 6, -82; Min 1 – 6, -62;   TB 0 – 7, -63;      StL 3 – 5, -33.

It is hard to really glean much from this and so rankings are used. I love power rankings but worry they are just a bit too subjective, so what can the above tell us? Let’s set two ranks, one based on w-l result (a bit rough and ready I know) and points difference (pd). Here is what we get

Rank by w – l Rank by PD
1.KC 1.Den 17.Ari
2.Den, Sea 2.KC 18.Mia
4.NO 3.Sea 19.Atl
5.NE, Cin, SF 4.NO 20.Oak
8.Ind, GB 5.Car 21.Pit
10.Det 6.SF 22.Cle
11.Chi, SD,Car 7.Ind 23.StL
14.Dal,NYJ,Ari 8.GB 24.Phi
17.Mia,Bal,Oak,Ten 9.Cin 25.Buf
21.Buf,Cle,Phi,StL 10.Dal 26.Was
25.Hou,Pit,Was,Atl 11.NE 27.Min
29.NYG 12.SD 28.TB
30.Min 13.Det 29.NYJ
31.TB 14.Chi 30.Hou
32.Jax 15.Bal 31.NYG
16.Ten 32.Jax

The correlation between the two is pretty good and so I am going to say the joint no.1 teams are the Chiefs and Broncos and should be ranked together. Overall tells us what we suspected, a group of teams are vying to be the worst in the league, though the Jaguars probably have that one Have to feel sorry for the Raiders, in any other division they would be knocking on the door for a wild card spot. Can we use the past 8 weeks to predict the final win – lose record of teams? Quite probably, if we consider the unlikely event of a draw like a loss (mainly because it is not a win, which are the most important), then we can do the same prediction as if tossing a coin. In this instance the current win (or success) record will be used as the probability of future success, using the pct value (no. wins/total no. games) obtained from the current standings available on www.nfl.com.  I will use this information to calculate the lowest and highest possible win – loss (w – l). This will be the 95% confidence interval around a middle predicted value. So I would be 95% confident that a team will finish with a 2-l record between the lowest and highest values.  The main assumption here then is that teams continue to play as they have done with no significant change in performance.  To be honest without needing to do any analysis, the AFC and NFC West are sown up, with the two current top teams taking a play-off and wild card spot. Oh ok so let us see if there is any proof of this and sort out the Chiefs and Broncos. The Chiefs are 95% likely to finish with (95% lowest probable<mid<highest probable result distribution if there was no significant change in performance) 13-3<15-1<16-0 record whereas the Broncos are 95% likely to finish with 12-4<14-2<15-1. This does assume that the current performances from the whole team continue and does not take into account how injuries could affect the dynamic of the team. What it does suggest though is that what we see now is likely to stay the same, based on the season so far, with 80% probability that the Chiefs will finish 15-1 or better and only 30% chance the Broncos will finish 15-1 or better. At the point of overlap the Broncos would take the division (due to points scored) so this actually equates to 55% chance that the division is decided i.e. it’s all to play for. Onto NFC West, well the Seahawks are the same as the Broncos (12-4<14-2<15-1) but the 49ers could finish with a score as low as 9-7 and a max of 14-2 (with a mid score of 12-4) so again they are up against it to win the division. So what about other contenders for play off and wild card spots?  Luckily we can sort of do a broad sweep using current w-l scores. We know the 49ers are 6-2 at the moment so any other team with 6-2 will be the same 9-7<12-4<14-2 with 95% confidence. This therefore goes for the Pat and the Bengals as well. The table below has every teams predictions along with their current win-loss (w-l) record

Current w-l 95%prediction (low<mid<high) Team (div.pos)
8-0 13-3<15-1<16-0 KC(1)
7-1 12-4<14-2<15-1 Sea(1); Den(2)
6-1 11-5<14-2<15-1 NO(1)
6-2 9-7<12-4<14-2 NE(1);Cin(1);SF(2)
5-2 9-7<12-4<14-2 Ind(1);GB(1)
5-3 7-9<10-6<12-4 Det(2)
4-4 5-11<8-8<11-5 Dal(1);NYJ(2);Ari(3)
4-3 6-10<9-7<12-4 SD(3);Chi(3);Car(2)
3-4 4-12<7-9<10-6 Bal(2);Oak(4);Ten(2);Mia(3)
3-5 4-12<6-10<9-7 Phi(2);Cle(3);Buf(4);Stl(4)
2-5 2-14<4-12<7-9 Hou(3);Was(3);Atl(3)
2-6 2-14<4-12<7-9 <1% chance to finish with 8-8 NYG and Min
0-7 & 0-8 0-16<1-15<3-13 <1% chance to finish 5-11 TB and Jax

So what does this all mean? Probably easier to understand if we plot the above using pct instead of w-l record.

pct prediction

What the graph highlights is how the teams with current records of 5-2 or better are likely to make the playoffs, mainly by winning their division. So it supports the idea that the Chiefs and the Broncos are in the playoffs along with the Seahawks (I am 95% certain of this) and Saints. All four appear with top four current pcts  The 49ers are pretty well there as according to performance to date, there is less than 2.5% chance they will end up 8-8 and less than 50% chance they will finish less than 12-4 which means for the Cardinals there is a 1.25% chance they will finish above the 49ers.  Along the same lines I am pretty confident that the Patriots, Bengals, Colts and Packers will win their respective divisions. Their closest rivals fall further to the left of the graph, below a pct of 0.5 and so are unlikely to get a run to top the group, at the same time as the leaders have a bad run of form. Apart from the Packers, as there is a 15% chance the Lions (4th point in from the right on the graph) with their current results will finish better than the Packers (still unlikely). The Cowboys win because nobody else is likely to. Really the interest lies in who is likely to take the two remaining wildcard positions after the Broncos/Chiefs and 49ers. This is anyone with a current pct of 0.5 or above (the upper 3 of the middle set of points) Will 10-6 actually be enough? These are the teams most likely to take the last two spots and I would say maybe in order of likelihood: NFC = Lions, Panthers, Bears,Cardinals; AFC = Chargers, Jets and on the outside chance the Ravens and  Titans (they appear to the middle of the graph). From the NFC, the Lions are in prime position but the Panthers and Bears can still make the playoffs, though mainly if the Lions have a dip in form and finish towards the lower end of the possible record spectrum.

So from this, in summary,  my initial play-off predictions are: AFC = Patriots, Bengals, Colts, Chiefs, Broncos then one from the AFC list above; NFC = Saints, Seahawks, Packers, Cowboys, 49ers then one from the NFC list above.

Just to re-iterate the proviso all this goes with. If a team makes any changes that means their final win – lose record is better or worse than the range given, this means that change had a significant impact on the performance. All of the predictions are based on the record obtained so far. But they do resemble the rankings based on the points for and against difference.

I think this is quite enough  statistics for today. Keep checking our Facebook page (www.facebook/gridirongentlemen) and  our twitter feed (@GridironGents) for further updates. Do check out Dan’s pieces for the Huffington Post and our latest Power Rankings as well as keep your eyes peeled for our International Series specials.

Toodle pip!