The season has finished, and we now know who has finished first, right down to 24th.

We’ve all made predictions at the beginning of the season, some with more success than others, and can now either brag about our achievements or (like me) cower in a corner and wish you’d never stuck your head on the line. Mind you, we’re only human; we think not only with our heads but also our hearts and our guts. If you remove emotion and base predictions on nothing but algorithms and facts, surely you’d do better.

Enter the supercomputer. I don’t know what constitutes a supercomputer, but all season you will have been reading articles about what the supercomputer predicts. I’ve written them; almost everyone in content creation has. We all turn to the supercomputer for comfort, hoping it predicts that our team will win the league, or at least stay out of the bottom four.

How accurate is the supercomputer? I thought, as a nice Sunday afternoon pursuit, I’d look at the predictions made by the fivethirtyeight computer over the season and see how accurate it really was. Of course, judging a computer has to be done properly, so I came up with a formula to test how good it really was. I’ve taken the top six, bottom four, and Lincoln’s position. For each of the top six and bottom four, the computer gets three points for a spot-on prediction (Forest Green 24th, for instance) and one point for the correct outcome (Forest Green relegated, but in 21st, 22nd or 23rd). For Lincoln, I’ve given a point for 9th or 13th, and three points for 11th.

So, how did the computer do? I took the prediction at the end of each month through the season, to see just how super it actually is!

July

3 Points

1	Derby
2	Sheff Weds
3	Ipswich
4	Wycombe
5	Barnsley
6	Portsmouth
14	Imps
21	Exeter
22	Accrington
23	Bristol Rovers
24	Morecambe

The supercomputer started out about as super as me. It did get two outcomes right – Barnsley in the top six and Accrington and Morecambe to be relegated. It had City within three places of the final finish, which was a good effort considering few of us did.

August

8 Points

1	Sheff Weds
2	Ipswich
3	Derby
4	Portsmouth
5	Peterborough
6	Barnsley
13	Imps
21	Cheltenham
22	Forest Green
23	Burton
24	Morecambe

Into August and the computer has a bit more to work with. From this point on, it always has one of the top two in the right place – in this instance, it has its first direct hit, Ipswich finishing second, and it’s the first time we see Posh in the top six (spoiler, it’s the only time). City are closer to their final place, whilst both Morecambe and Forest Green are relegated as well. Perhaps it is super, after all.

September

9 Points

1	Sheff Weds
2	Ipswich
3	Portsmouth
4	Derby
5	Barnsley
6	Plymouth
11	Imps
21	Bristol Rovers
22	Forest Green
23	Burton
24	Morecambe

We’re up to ten points, from a possible 33, so by the end of September, the supercomputer had just under 30% of the league (or the bits that we’re measuring) sorted out. Ipswich and City were spot on in 2nd and 11th, Barnsley were in the play-off, and Morecambe and Forest Green were going down.

October

8 Points

1	Ipswich
2	Plymouth
3	Sheff Weds
4	Derby
5	Portsmouth
6	Bolton
10	Imps
21	Accrington
22	Burton
23	Forest Green
24	Morecambe

October comes, and the supercomputer is getting better. It hasn’t sorted out exact placings, but it already has Plymouth and Ipswich going up automatically – this is the first month it gets one of the sections (autos) accurate to all teams. It’s got two play-off teams and three of the four relegated sides. It even has City within a single place as well. If it’s data-driven and football is all about the numbers, then. it’s only going to get better from here.

November

9 Points

1	Sheff Weds
2	Ipswich
3	Plymouth
4	Derby
5	Barnsley
6	Portsmouth
9	Imps
21	Cambridge
22	Burton
23	Morecambe
24	Forest Green

Football is a funny old game, and by the end of November, with more data available, we start to see some waning of the accuracy. Mind you, it has the top three correct for the second month in a row, just in the wrong order. It’s the first month Forest Green are bottom, and the highest prediction all season for the Imps. In terms of being super, the computer is beginning to look like it’s emerging from a phonebox wearing a cape.

December

7 Points

1	Ipswich
2	Sheff Weds
3	Plymouth
4	Barnsley
5	Derby
6	Bolton
14	Imps
21	Cambridge
22	Burton
23	Forest Green
24	Morecambe

The cold month of December, with a World Cup distracting us and a run of winless games. City slide to 14th, but many were predicting a relegation battle at this stage, so the computer did have some faith. If they can show faith. Which they can’t.

The top three are right, but in the wrong order; it’s the first month the computer brings Bolton into the equation, Barnsley are right in fourth, and we’ve only got two of the relegated teams now. For the third month in a row, the computer is less accurate than the month before.

January

5 Points

1	Sheff Weds
2	Plymouth
3	Ipswich
4	Derby
5	Barnsley
6	Bolton
16	Imps
21	Morecambe
22	Cambridge
23	Burton
24	Forest Green

More info, more games to simulate and a less accurate outcome than the month before. Of course, Wednesday’s form will contribute to this, City hadn’t won in a month or so, and Derby still look like promotion contenders. Rams fans will be gutted to see how the computer felt they should have been top six, all season. However, knowing it only got five points out of a possible 33 with three months and a week or so remaining in the division does somewhat sell the ‘super’ bit of its title short.

February

12 Points

1	Sheff Weds
2	Plymouth
3	Barnsley
4	Ipswich
5	Derby
6	Bolton
13	Imps
21	MK Dons
22	Morecambe
23	Cambridge
24	Forest Green

Remember, this is taken at the end of the month, and so it means with two months and a game or so left, the computer is getting there. We have Plymouth promoted, Barnsley and Bolton in the top six, and three of the four relegation spots accurate. It’s the first time the computer thought MK Dons were in trouble as well.

March

14 Points

1	Sheff Weds
2	Plymouth
3	Ipswich
4	Barnsley
5	Bolton
6	Derby
14	Imps
21	Accrington
22	Morecambe
23	Cambridge
24	Forest Green

Now we’re talking. The computer gets five of the top six teams right, with two (Barnsley and Bolton) in the right place. It’s still underselling the Imps, but with 14 out of 33, it is now 42% accurate. Mind you, that is with April, and May left, and I guess quite a few of us might have been able to have a stab in the dark at this point.

April

25 Points

1	Plymouth
2	Ipswich
3	Sheff Weds
4	Barnsley
5	Bolton
6	Derby
10	Imps
21	Cambridge
22	Morecambe
23	Accrington
24	Forest Green

By the end of April, meaning there was a single game left to play, the computer was fairly accurate – the top five were spot on, City were a single place out, and three of the bottom four were right. However, are you really a supercomputer if, with a single game left to play, you’re still not getting everything right? Or, are you just an algorithm model that doesn’t understand the truly unpredictable nature of football?

Conclusion

Supercomputer? No, not one bit. There will have been supporters who will be more super than the supercomputer. There will have been better-educated guesses and punditry predicting Mk Dons to go down and Posh to go up, both of whom were only factored into the supercomputer rankings once all season.

What it is good for is giving a general idea of how the league will look. Lincoln didn’t vary too much, seven places (9th to 16th) which was a good shout given our lack of wins between November 19th and February 4th. It also had the top three nailed on from the third month, and at no point had fewer than two of the relegated sides in place.

It’s fun, but if you think it’s going to turn up at a burning building, get changed in a phonebox and rescue the heroine of the hour, you’re mistaken.