Well that was just as enjoyable as the first game and it was nice to have Brendon along for a small part of it. Perhaps a bit less statistics than I’d intended, and certainly less on bowling after I concluded that bowling average is pretty good just the way it is. I’m convinced, though, by the need for a much better measure of immediate batting performance for a team than run rate. Expect further advocation of this idea in the future.
22:45 And a six to finish! Nan is impressed: “Excellent finish, with those fours!” is shouted up the stairway. Well done to the Australian team for a more balanced performance than the dominant display from David Warner in the first match. They’ve both been very entertaining matches to watch and I’ll be looking forward to the new IPL season, which will still apparently feature Matthew Hayden.
22:42 It’s a race to the finish here—two overs left and my laptop is down to reserve power. No, eight balls left; three runs to get. Surely it’s Australia’s game from here.
22:36 Good over for Cameron White there: four consecutive fours. I think that he’s not been dismissed in his three outings (in four matches) for Australia, so his orthodox average is impressive. I’m inclined to insist that runs per innings be used, but White is an interesting case because he so often bats down the order. I’m forced to further qualify the definition to be runs per innings weighted (somehow) by the portion of the innings available to the batsmen. That’s not quite so clean as runs per wicket.
22:24 A question from across the living room: does anyone have information on the Asus netbook that is available at Aldi? I’ve never heard of Asus or a netbook before, and I don’t think I’ve ever been in an Aldi. Are they like Lidl? This blog post was the only thing I could find. The hard drive space seems fine; what about RAM, and does it have an ethernet connection or just wireless?
22:16 Well that’s two quick wickets. And now Duminy’s going to bowl. Duminy : cricket commentators :: Radiohead : music critics. The game now “on Mike Husseys’ bat,” according to a commentator whose voice I don’t recognise. Frankly I don’t know what else to say. It’s extremely tense! Game tension: can that be quantified? Thinking caps on…
22:06 If there isn’t an Australian collapse coming in the next few overs, then it looks like it’s going to be a nail biter. But Australia ahead of where South Africa were at the same point.
21:59 Here’s a copy of the (low-resolution) data for the South Africa innings that I was using in case Brendon anyone wants to try their hand at a predictive probability distribution.
21:52 That was quite the six there from Ponting and well caught by a chap in a South African shirt in the crowd next to the group of lasses wearing singlets with “Hussey’s Husseys” emblazoned on the front. I don’t think I can really voice disapproval without sounding stodgy, so I will instead commend the singlet-maker on getting the apostrophe in the correct position.
21:47 Hashim Amla has thrown the boundary ropes into disarray and all to no avail. I just can’t get over how awful and blotchy the grass in the outfield looks. Foreigners make enough jokes about drought-stricken Australia as it is. Healy was saying earlier that it was due to some fungus or other. Australia positively limping to the half-century here.
21:42 Well I’m not happy at all with how things are going for Australia. Ponting looks deadly serious now with a strike rate of 25. There’s that Crown Lager ad again. I think I just like it because the people are all well-dressed. M. Morkel into the attack and a wicket first ball. Marsh didn’t really look in a dominant mood, but he was doing it less tough than Ponting. Who’ll be in…. it’s M. Hussey! Tony Greig: “He’s been having a bit of a tough time… of late.”
21:35 And Warner’s gone! One-hit wonder? Where are my odds for his getting a duck? Well, he was off the mark and was certainly having a thrash. Ponting in and not looking too worried about the relative paucity of runs. There was an LBW appeal there, but it’s turned down—attaboy Reiffel.
21:27 South Africa changing bowlers every over too! I think there could be a push for converting to indoor cricket rules sooner rather than later. Also: once again I state my opposition to LBW as a mode of dismissal in 20/20.
21:22 Some mid-innings conversation with B. J. Brewer:
Brendon: You could make a predictive probability distribution for the amount of runs scored on the next ball
me: and then update it from ball to ball?
Brendon: Yes
Yes, so that would be a good strategy, and a good alternative to having the pick a weighting function to incorporate previous overs, which is, as Brendon say, ‘ad hoc’. Oh and clearly ‘overs’ are an arbtirary division, so doing things on a per ball basis rather than pre-binning as I’ve done here make even more sense.
21:16 Right, Australia’s innings now! I’ve updated that run rate plot to include the whole of South Africa’s innings. Ignoring the spurious jump in the green line at the final over, who could possibly claim that it isn’t a more accurate representation of the way Duminy exploded during the last few overs. Good start from the South African bowling attack.
20:50 Another maybe not-so-obvious point to make is that the window of the smoothing function should be uni-directional, i.e. it should take in runs from previous overs, but not look forward in time. At least, that seems right. Almost the end of the innings now and bat and ball are flying everywhere.
20:48 Eighteen overs gone now. Just some notes on that last plot. One big problem with boxcar smoothing (that is, at each point x_i, the boxcar smoothing is the average of the points x_(i-n) to x_(i+n) for a fixed n; here n = 1 has been used) is that the endpoints are always the runs scored in that over exactly, which is not very good. On the other hand, the curve is much more responsive that the conventional definition of run rate.
20:45 Well Australia well on top here again, I think. Every time things start to blow South Africa’s way, there’s one or two quick wickets. Have been attempting to stitch together a quick plot of how this regression might work:
Two measures of how a batting team is performing: given the runs scored in each over (points), both the run rate (total runs divided by overs; blue line) and local measure of momentum from boxcar smoothing (green line) can be defined. Which is more descriptive?
20:22 What about Ponting mixing the bowlers up here? It’s so obvious even the commentators have noticed. It’s like he’s keen for indoor cricket rules.
20:18 Apparently it’s LOESS, for ‘locally weighted scatterplot smoothing’. I imagine that quantifying momentum at a point in a match would require taking wickets into account as well. Maybe runs divided by expended Duckworth-Lewis resources, with some normalisation. Do they have a D-L table for 20/20? Surely.
19:59 Momentum building for South Africa? Now I want to work out how to quantify that. I’m thinking these Loess regression curve things that Nate Silver uses for his plot of electoral predictions. Run rate as total runs divided by total overs is clearly not good enough to capture quick changes. And here’s a wicket taken! Gibbs though, not Amla.
19:50 Goodness. I do like this way they chat to players in real time. Even so, talking to Bracken as he’s starting his run-up seems a bit silly! Doesn’t he need to focus? I think they need to get some celebrity interviewers as well… now hang on, the commentator’s giving him feedback on his deliveries. Surely that’s against the spirit of the game.
19:48 Tait being a inconsistent with his accuracy. Or is it with his precision? I think the latter—he’s definitely aiming for the right spot, he’s just not getting it. That bouncer was insane… and over Haddin’s head for four. He can have a think about that through Bracken’s next over.
19:44 C. H. Evans: “South Africa are playing 50 over cricket.” Not test cricket, I wonder? “No.”
19:41 Yes, thanks Tony. Twenty overs for each side… that’s why it’s called 20/20. “It’s all action.” So I’m thinking this evening about how to quantify the quality of bowling in 20/20 cricket. Runs per wicket actually seems like a pretty good measure here, unlike the case of batting. Could there be room for indoor cricket rules in a short form cricket game: -5 runs for a ‘wicket’, but the batsmen stay out for the entirety of fixed period (say five pairs of batsmen facing four over per pair).
19:36 What’s with all the patchiness on the oval? I retract my comment about the pitch looking good. Shaun Tait looks sharp, though.
19:32 ‘Gabba looking good, but why isn’t new ex-cricketer Matthew Hayden out commentating at this home ground. They’re talking about Dave Warner, of course. Odds he gets a duck this time? Here come the Australians…
19:31 Okay here we go!
Letters to Nature 
I’m here. 🙂
One thing that disturbs me is the amount of KFC advertising…
It’s such a lame idea too. And yet, there’s so much stock footage to use. Imagine all those Question Time sessions in the ABC archives. There’ll be clips of J. W. Howard introducing legislation to gaol illegal immigrants, while the voiceover intones some nonsense about coupons and cheap greasy chicken. Mmmmm.