Season over...

...something which wasn't overly surprising really, given Blackpool's form and Oldham's form going into the play-off matches played last week. Blackpool scored five goals (and could have had more), compared to Oldham's meagre two. Although Oldham did have two disallowed, and could ask about a number of handball appeals, the truth is Blackpool were the better team. It's incredible was confidence and form does for a team.

If Oldham fans feel bad, one wonders how Nottingham Forest fans feel. It's been a pretty humbling season for them. Thrashed 4-0 by Scunthorpe at home, 5-0 by the Latics, they lost 5-2 at home to Yeovil to tumble out of the play-offs on Friday.

The usual things are coming out of the City Ground, or at least out of the body of supporters of Forest. The chairman and manager are to blame, they must go or be sacked (who sacks the chairman and board, out of interest?!). Does that solve problems? In the article I read (on the BBC, here), it seemed no contradiction was noted between the fact Forest have had seven managers in five years and dropped a division, and not escaped League One in two attempts, yet still the author calls for another manager to be sacked! How does that work?

Blackpool fans were calling for Simon Grayson to be sacked after 11 games of this season. Now they're 90 minutes, and odds on favourites to be promoted to the Championship! There's absolutely no difference between that situation and the situation for Forest. Man U fans called for Fergie to go, over a year into his tenure, and probably still weren't entirely convinced by that first FA Cup win. No difference again.

I wonder if football supporters will ever learn that success requires patience.

Play-offs

It's the business end of the season, which is a big distraction from my Ph.D thesis write up, but thankfully my thesis is almost there.

So today sees the Premiership conclude, but more importantly for me, it sees Oldham play Blackpool in the first leg of the play-off semi-finals in League One. Blackpool fans are confident, and rightly so: Blackpool won their last seven league matches, including a 6-3 demolition of Swansea on the final day. But Swansea have lost against play-off sides all season, and the high-scoring match was because Swansea had to go for the jugular to try and displace Oldham in the play-off positions.

Nonetheless, Blackpool are clearly a very good side. They also won 1-0 at Boundary Park in the league earlier in the season, and Oldham have won just once in the twelve most recent matches between these sides - I don't like the sight of Blackpool much! But Oldham raise their game against the better sides, so let's see!

The now fail-safe model (haha) suggests that Oldham are actually favourites for the first leg, perhaps surprisingly, with a probability of 37.8%, which translates to decimal odds at 2.64. Not far from where Betfair is currently at, with Oldham favourites at 2.57. Goes somewhat against a strong statement here, that Blackpool are "firm favourites". Perhaps the Blackpool fan here ought to check out the bookies before making such assertions.

Nonetheless, any betting man fancies Blackpool over the two legs for sure. Hopefully the underdog tag will help the Latics. For the first leg, the draw has probability 31.2%, and Blackpool a surprisingly small 30.9% chance. Let's hope these numbers reflect reality - they reflect the model estimated over the season, but that's not today!

As for scorelines, a narrow 1-0 Oldham win is the most probable scoreline, with a probability of 14.8%, closely followed by a 0-0 draw, at 14.2%. So suggestions are for a cagey and tight match this evening. The 1-1 draw has a probability of 13.5%, and a 1-0 Blackpool win, a repeat of the league match, is at 12.9%.

In the absence of Finktank bothering to predict this one, I imagine these odds are similar to what the post, because generally they are.

Easter Saturday's matches...

Who's going to do what on this important footballing weekend? Today, 11 matches take place in League One, after Scunthorpe effectively sealed their promotion last night winning 1-0 over Yeovil Town.

Below are the predictions for this afternoon's matches:

Home Draw Away
Blackpool v Tranmere 46.6% 29.4% 24.0%
Bournemouth v Bradford 44.6% 29.2% 26.1%
Brentford v Cheltenham 33.8% 30.1% 36.1%
Brighton v Doncaster 36.7% 31.6% 31.8%
Bristol C v Swansea 12.3% 21.0% 5.8%
Chesterfield v Crewe 40.7% 26.3% 33.0%
Leyton Orient v Northampton 41.8% 26.6% 31.7%
Millwall v Nottm Forest 28.2% 26.6% 45.1%
Oldham v Huddersfield 66.5% 19.3% 14.2%
Port Vale v Carlisle 46.5% 26.3% 27.2%
Rotherham v Gillingham 52.0% 22.1% 25.9%

Disturbingly, for the biggest match of the day, the model appears to have been hit by a glitch; the probabilities don't sum to unity. This will be looked into.

For the remaining matches, the one I'm interested in is Oldham v Huddersfield. Oldham have never beaten Huddersfield twice in one season before, but today they appear to have a 66.5% chance of doing just that, having won 3-0 at Huddersfield earlier in the campaign. Oldham are indeed strong favourites for this match, which reflects that fact Oldham are considerably higher in the league table. Huddersfield are not in relegation trouble, but are firmly in mid-table. Fink Tank put Oldham at 59.7%, also rating Oldham as overwhelming favourites.

But will they prevail in this local derby? With it being a local derby, a large attendance can be expected. Attendances have been increasing gradually over the season, and a gate of around 10,000 ought to be expected this afternoon. Huddersfield may not bring as many fans over the Pennines as they would if they were challenging nearer the top of the division, but nonetheless, my forecasting model suggests an attendance of 9,479. Let's see if the Latics can send this bumper crowd home happy as they further cement their play-off place.

Another interesting challenge...

...Oldham Athletic will be charging just £2 for adults, and £1 for kids in their next two home matches. These are matches against Brighton and Hove Albion, and Northampton Town, matches which otherwise would not attract a very high attendance at all. It's a great idea, I'm totally for it - get a full house in, hopefully get some kids hooked on Latics, especially if the team can carry on playing as they have recently (unbeaten in nine league home matches, seven wins).

But it creates havoc for modelling and forecasting. On the other hand, it's a known break. One of the biggest problems with forecasting is structural breaks, if the break affects the mean of the time series being forecast. Before the break, it's generally quite hard to predict one will happen, and then if it can be predicted, the size is another matter entirely. If a break happens, next period when forecasting, one needs to know whether this is simply an outlier, measurement error, or an actual break.

So here, I have quite a lot of information: I know something will happen that can only be described as a break - the two previous big discount matches (Grimsby (free) and Torquay (fiver)) have attracted substantially larger attendances. I further know it will only last for two matches, it is not permanent, but it's not a one-off either. And I'll know the reason why the break has occured.

All in all, it makes for an interesting experiment. Will the attendances be as large as the Grimsby or Torquay matches? A method for capturing a break is to add a dummy variable estimated over some observations in the sample, and extend this dummy into the forecast period. Usually in time-series, the dummy is for the last few observations. However, here as we have some idea from these matches how big the break might be, it makes sense to use this information.

Extending the Grimsby dummy gives a forecast of 11,670, and extending the Torquay dummy just 6,856. I think somewhere inbetween these two examples is more likely. A combination, weighting Grimsby at 0.6 since £2 is closer to £0 than £5 gives a forecast of 9,442, but I think a higher weight should be used, since zero prices have wierd effects. A weight of 0.75 on Grimsby provides a forecast of 10,230, which pleases me.

Come on Oldham!

Attendance vs Crewe

It's been a couple of weeks since the last home match, and with a couple more wins under the belt, one might expect a bumper gate in the home match tonight against Crewe, which kicked off about seven minutes ago. Particularly given one of the two wins was a crushing 3-0 win at local rivals Huddersfield. However, my model of attendances predicts a gate as low as 4,289.

A number of factors cause this: it's a Tuesday night; it's December, Christmas is coming and all that. However, it's the first home game for a while.

I can't believe the gate will be that low. I'm looking into forecasting using what is apparently a very good forecasting device, the exponentially weighted moving average. By altering the parameter on the weight from 0.1 to 0.9 I get a range of forecasts between 5,122 and 5,903.

I'm inclined to go with the naive forecast with parameter 0.1 to be honest, I reckon the gate will be just around 5,000. I could be proved wrong, and my model right however. I just hope Latics win!

A challenge of another kind entirely...

Forecasting league attendances at Oldham has become reasonably simple - just below 5,000 is a good benchmark. However, the Association Members Trophy, now known as the Johnstone's Paint Trophy, poses another challenge altogether. This is because no-one really cares about the competition, and it doesn't provide interesting opposition as a rule, because the teams that participate are all from League One or below. Hence the best that could really be hoped for is Nottingham Forest. Oldham have drawn Chesterfield. Hardly glamorous, no offence to Chesterfield. I'm sure had they drawn Oldham at home they'd feel similarly indifferent. The dummy variable on these kind of cup matches suggests they knock something like 60% off the normal gate at a home game.

So from 5,000, that suggests there should be around 2,000, which sounds about right. However, estimating on data up to Saturday's pleasing 3-0 win over Brentford gives a forecast of 3,107, which is way too high. When Oldham made the semi-finals of this competition two seasons ago, the home game in the semi attracted only 2,881 people! However, that was probably because the club decided against offering substantial discounts for that game, which they will undoubtedly offer on Wednesday, and did in previous matches that season - the quarter final, against Tranmere, attracted over 4,000 people, while the second round match against non-league Accrington Stanley drew a crowd of 2,812.

Yet I still think 3,107 is far too high. Inserting dummy variables for a couple of outlying matches (Chasetown in the FA Cup last season, and Carlisle earlier this season), and inserting an intercept correction for the current season, provides a lower forecast of 2,675, which is better. Further, adding these dummies renders some other curious coefficients insignificant, and given their "wrong" sign, and plausible explanation (see paper), it seems sensible to get rid of them. These variables are a dummy for when the game falls on a Wednesday night, and the Days Since Last Home Game variable, which is the number of days since the last home game.

When got rid of, the forecast falls yet further into acceptably low territory: 2,504, and so this is the forecast I'm going with for the Chesterfield match.