Rolling Averages

[IainB]

Is there a case for changing the way in which rolling averages work?

Currently at Ipswich Tom Brennan is still heading the rolling averages with 7.24 but his actual 2026 average is 6.49 his 7.24 means he has to line up at number 1 but due to his very bad start to the season, possibly due to his leg break in the winter, Musielak, King & Lawson are all above him on 2026 figures.

Should team assembly be done to rolling averages and riding position based on actual in year averages? Tom being forced to ride at number 1, even now in July, is not doing him or Ipswich any favours at all.

[TheTopDrawer]

That shows why rolling averages are better.   It shows the true form over a period of 20 matches.

As any statistician will tell you, a larger sample gives a truer answer.

Taking a sample of the last 20 matches smooths out any anomolies like a temporary run of bad form (or an attempt to manipulate averages by purposeful low scoring).

The system works well.  (Or at least as well as it can).

[therefused]

with the lack of fixtures nowadays id say 20 matches is too many. 

[paulnolan]

Rolling averages are something that has come in since the last time I went to speedway and they do and don't make sense. They can reflect current form but not in their present sense. Unless I'm mistaken, a rolling average over 20 meetings - with the current lack of teams at the top level - could involve THREE seasons of racing, and cover scores from - for example - October 2024 to April 2026. Any rolling average which even counts meetings 6 months apart - as all do until the very end of the season, does not reflect CURRENT form. Six meetings is a far better barometer - too many to deliberately drop your average to gain an advantage but enough to gauge current form.

[IainB]

1 hour ago, TheTopDrawer said:

That shows why rolling averages are better.   It shows the true form over a period of 20 matches.

As any statistician will tell you, a larger sample gives a truer answer.

Taking a sample of the last 20 matches smooths out any anomolies like a temporary run of bad form (or an attempt to manipulate averages by purposeful low scoring).

The system works well.  (Or at least as well as it can).

I think rolling averages for team building purposes is fine as it takes into account a riders full capabilities over a season but as Tom Brennan has shown this season if a rider has a massive slump in form it just takes too long to escape from the number 1 berth.

[mikebv]

They were brought in to stop riders underperforming, and then suddenly improving again once their reduced average had allowed a significant strengthening up for the team...

So, a laudable idea...

With so few meetings though, 20 is probably too many now for the top tier, as it is almost a full season for a team knocked out of the KO Cup in the first round, and zero chance of a PO place..

(Does the KO Cup even count to the averages?).

Maybe 10 matches, (half a season), would be better and, as an add on, the team with the highest collective average becomes the new target?

Struggling teams, (who we will all accurately have agreed who they are going to be before a wheel is turned) always stay struggling when they can only replace a rider with an equal or lower average, or still have to run with the team average decided for the season..

Eg. If 42 is the season average and the top team is running at 45, then the struggling team will still be well behind the one at the top if all they can do is reset to 42, meaning their season will continue to be one of struggle, (and hardly worth the changes)..

Letting all teams match the No1 team, at say the half way stage, would mean a very competitive second half of the season, rather than the usual "game over" we regularly have as the PO teams stroll into them.. 

And a half a season contract may attract more riders who see an extra 10 matches as something they can work around..

Edited 9 minutes ago by mikebv