Will Oxford stay up? Simulating the Relegation Battle

At what point does four games unbeaten under Manning become four games without a win?

The former MK head coach has certainly stabilised United, who look better on the ball, more prepared in transition, and have stopped conceding from every shot they face (although Simon Eastwood's discovery of his inner prime Yashin may play some role in that). 

The reality is, however, that United went from three points and two places above the drop zone before Manning arrived, to two points and one place since his four draws. 

This is not to say things have declined under Manning. The nature of the fixtures and league meant things were always likely to get worse before they got better. 

But whether in desperate hope or genuine belief, fans who thought the removal of Robinson would lead to an immediate escape from the dotted line were mistaken. 

Oxford are in a serious relegation battle. 

But if you'd rather skip the hassle of checking the scores each Saturday and desperately hope that United actually score an open play goal soon, then you can check back at the table in May with the knowledge that everything will be fine... according to the data anyway. 

That's right, it's simulation time!

How the Simulation Model Works?

For a detailed explanation of this prediction model, you can check out this article I wrote when United were chasing the playoffs two seasons ago (how I wish I was writing in that context again), but below is a brief summary of how it works.

It uses underlying data for each team to analyse the likley result for each fixture till the end of the season. This involves combining a team's Expected Goals for (xG) with their opponents Expected Goals Against (xGA), both adjusted to home and away standards, to determine the probable goals for each team in any given game. 

Whichever side comes out on top in the xG earns the three points, with any matches with an xG difference of 0.10 or less counted as a draw. 

The system sounds overcomplicated (and it probably is), but provides a good account of how each game should finish. 

Results

Above is how the bottom of League One looks heading into the last six or so games, and it is undoubtably tight. 

Right now it appears to be a six horse race, although MK Don's up turn in form is seeing them pull away, whilst the likes of Port Vale could be sleep walking into a relegation scrap. 

But what does the simulation say? Can Oxford pull themselves together, will Cambridge's revival be enough, could Forest Green pull off the incredible? 

The answer to two of those questions are still a mystery, but this is how the simulation expects the table to develop, starting with what it will look like after the next three game weeks... 

With most teams having two games to play, the simulation suggests those four relegation spots will be all but decided. 

MK Dons slip down into 20th, whilst a couple of wins for United sees them climb away. 

Burton, meanwhile, are only expected to pick up two points from their four games, but do have that crucial extra match. 

As do Cambridge, but they look too far adrift by this point. 

This is what the table looks like after the final games are simulated...

For the four relegation places, it's as you were. 

In the past ten seasons, the average points needed to survive has been 49.2, but if all results go as expected (which it goes without saying it is very unlikely they will), MK Dons would survive with just 46. 

Fleetwood stayed up with 40 last season, but prior to this you'd have to go back to the Division Two days to see a side finish 20th on lower than 46 points, as none other than Oxford themselves accumulated 45 but stayed out of the bottom four. 

According to the simulation, then, it will be a scrappy, low points affair this time round, so let's delve deeper into why each team is expected to finish where they do.

Port Vale

Vale perhaps benefit more than anyone else from form not being factored into these calculations. 

They had lost ten out of sixteen prior to their draw with Oxford on Saturday, seeing them falling dangerously towards the trap door. 

However, the simulation suggests they have nothing to worry about, picking up ten points from a possible fifteen in their final five. 

Games against mid-table Lincoln and Bristol Rovers both provide wins in this model, and they are the key matches coming up for vale. 

The data also suggests they may be able to impact affairs at the top of the table, too, as they are predicted to beat Plymouth at home on the final day, possibly a big match for both teams. 

Vale's form means they may not be quite as comfortable as the model suggests, but their end of season wobble should be just a wobble rather than anything more serious. 

Oxford United

Oxford's good underlying numbers makes them well suited to pick up results in this model, but the rate they fly into mid-table mediocracy should be some comfort for fans.

Oxford's Expected Goals numbers, helped by good set play data, suggests that they will storm to five wins from six, including victories over Bolton and Portsmouth at home. 

They are also predicted to register fairly comfortable wins against Forest Green, Cheltenham and Accrington, although nerves are not part of the calculations here. 

Watching Oxford games might beg the question of where the goals are going to come from, but United's underlying data suggests that they will start finding the target. 

Manning's defensive improvements have led to two clean sheets in four, as many as they'd managed in the 22 matches prior to that in all competitions, so there is certainly a solid base to work off. 

If United can start finding the net, then, the model suggests they will have nothing to worry about.

Even the most optimistic fans, though, would admit they five wins from six is probably quite ambitious. 

Burton Albion

The always mysterious League One fixture machine has meant Burton have played fewer games than any of the other relegation candidates, and their fans would probably feel safe having clawed themselves away after a poor start to the season. 

They are right to... but only just. 

There next five games include four of the top six, and the simulation only has them picking up two points from these matches. 

This would leave them on 47 points heading into back-to-back home games against MK Dons and Cambridge to finish the season. 

They are predicted to win both of these matches, but the pressure on these matches, with both MK and Cambridge needing points themselves means it is unlikely to be simple wins. 

If Burton are to stay up comfortably, then, they will need to make use of one of their games in hand, which means an unlikely result at home to Sheffield Wednesday or Bolton is needed as they also face daunting trips to Plymouth and Derby. 

Burton are definitely not out of the woods yet, even if their games in hand might lead people to thinking so. 

Cheltenham Town

Cheltenham might also feel relatively free from relegation conversations, but the prediction model says that they could yet be dragged into an end of season scrap. 

The model has them dropping three places from 16th to 19th, the highest decline of any of the nine teams. 

This is partly down to having three away games left, and in MK Dons, Oxford and Wycombe, it is likely all three sides will have something to play for. 

This makes their home match against Forest Green a must win, as the model predicts this to be one of only two victories in their run in. 

The other comes at home to Charlton on the final day, but if this is to be believed then they won't have reached the hallowed 50 point mark heading into that all important May weekend. That game, then, could still have significant implications on the relegation picture. 

Cheltenham fans may rightly think they have enough to pull themselves away, but they could use picking up a point or two from those tough away matches or they could find themselves in trouble. 

MK Dons

Manning's old club MK Dons seem to have found the key to results recently, and looked the most likely side to propel themselves out of the relegation conversation. 

However, three wins in March has been followed with three draws in April, and it looks like they will need a few more results to get themselves clear. 

The model suggests these will not be easy to get. 

A win against Cheltenham at home in their next game is not assured, but is their only three points according to the prediction.

A trip to mid-table Fleetwood in two weeks is a big opportunity for MK to register some much needed points as a home game to Barnsley follows and then a potentially important trip to Burton on the final day. 

The next few games will be crucial for the Dons, and whilst 46 is predicted to be enough to keep them up, they will not want to rely on such a small total keeping them safe. 

Accrington Stanley

Accrington are predicted to be the closest of the bottom four to escape, but they fall narrowly short. 

The model predicts that they will collect seven points from their remaining six matches, including a draw at home to Peterborough. 

Their big game will be the final one, though, away at the Kassam. 

Whilst the model suggests Manning and co will be on the beach by then, it is very possible that game will be a relegation decider. 

Considering Accrington also face trips to Bolton and Portsmouth, picking up a result from this game feels pivotal to their chances of beating the drop. 

They are expected to beat Cambridge at home on the penultimate game, but even this will be tough with Cambridge needing points wherever they can get them. 

Cambridge United

Mark Bonner's men are expected to fall five points short of safety, and this is largely because of their next three opponents. 

They face Wycombe and Peterborough at home before travelling to table topping Plymouth. Any results from this run would be a bonus. 

The model suggests that their only point from that run will come at home to Wycombe, which would leave them on 38 points heading into their final three, and trips to fellow strugglers Accrington and Burton are huge games. 

Cambridge are in good form so might back themselves to pick up a result from their next three, but it might be asking a little too much for them to get enough points. 

Morecambe and Forest Green Rovers

Morecambe's survival hopes already looked slim and Forest Green's essentially non-existent, and unfortunately the simulation suggests there is no evidence they'll turn it around. 

Both teams are expected to pick up only a single point from their remaining matches, and are likely to be playing League Two football next year. 

Forest Green do have several games against other relegation candidates, so they will worth watching to see how they impact the wider picture, but the model suggests a draw at home to Fleetwood is the best it will get for the Gloucestershire outfit. 

For Morecambe, they have just run out of games, probably needing to win all of their final games to have a chance at staying up. 

What does another Model Say?

Whilst reading this article you may have noticed that the simulation does neglect some fairly key factors (form being a key one), and whilst it gives a good idea of the potential results, it predicts based off a team's underlying numbers rather than their actual data. 

This means that it thinks Oxford might actually score more than one a game (something which watching a match at the moment would tell you is unlikely), hence their remarkably high survival chances. 

That's why it is worth looking at some other models, particularly those generated by a computer which has far more data than my spreadsheets. 

The best site for this is called FiveThirtyEight, which uses form, underlying metrics and historic club performances to simulate each match. This is then done hundreds of time to find an average and likely final position for each team. 

This highly complex and expensive algorithm, though, provides relatively similar results to this simulation, with their predicted final table below... 

The bottom four are predicted to stay the same, but their simulation suggests they will be some way adrift by the end of the season. 

This would be a welcome thought for Oxford, who would not have to worry about that final match against Accrington, but are the side expected to come 20th with a relegation chance of 14%. 

Burton and Cheltenham both finish on similar points in both models, but FiveThirtyEight does not fancy the chances of Port Vale or Oxford picking up so many points, so they escape relegation comfortably. 

Burton's 4% relegation chance does speak of how important their games against Cambridge and MK are, though, as if these go against them, FiveThirtyEight predicts they could find themselves closer to the bottom.

In both cases MK Don's form is expected to run out, so they are definitely not clear yet. 

Clearly these are just predictions, and at this time of year results don't tend to follow expected patterns, but some conclusions can be drawn from this. 

The bottom four have all the work to do, and if results go as they should, it will be the current crop that go down in May. 

MK Dons have to be careful not to slip up, whilst Burton have a relatively high chance of falling into an unexpected final day scrap in both models, so should definitely be looking nervously over their shoulder. 

Most importantly though, they both suggest Oxford will be a League One team next season. One is drastically more optimistic, but if Oxford stay up, who cares if they finish 17th or 20th.