So this is the chart that prompted the question:
I have grouped the ages here using a simple criteria. The players age was determined by their birth date on January 1st. So Edin Dzeko, for example, has a birthday on March 17th. Dzeko turned 27 in March 2013. But this was Dzeko's 26yo season due to his age on January 1st. Get it?
There is probably a better way to decide the age cut-off, as always I am open to suggestions?
Strangely, a lot of 31 year old players. Just 10 strikers under the age of 25.
Goals Per 90
I took all the all the non-penalty goals scored and all the minutes played by each age bracket and ran my usual per 90 calculation. This is what Goals Per 90 by Age looked like for the 2012/13 Premier League season.
Now, let's not be hasty, this is only a 45 (46, I forget about Darren Bent. It happens, just ask Lambert) player sample and over only a single season so we don't want to leap to any premature conclusions.
Lukaku, the only player with a 19yo season, is a beast. We then see steady improvement from 22yo to 26yo. Then comes the fall at age 27, it continues at age 28 then we see a rise at age 29.
The information on players in their 29yo season is polluted by van Persie, an outlier in terms of performance at a late age. If we remove van Persie, which isn't entirely fair, then the 29yo goals per 90 drops to 0.27.
Their are just two players in the 30yo bracket, Defoe and Lambert.
This chart tells us things we probably already guessed at, chiefly that strikers performance drops off by the late 20's and early 30's. A chart like this needs far more data points before we can draw and further conclusions.
Shots On Target Per 90
Again, sample size restricts what we can take away from this chart. Ages 29 and 30 are hampered by van Persie (29) and tiny samples (30yo).
If I was to take a guess, and it is only a guess, I would think that more data would show a 5 year peak for strikers between the ages of 24-28. Players are still learning before that age group and any striker who plays after the age of 28 will probably see his numbers start to slow down and then drop off by the time he gets to 31 or 32.
*I'm going to expand this study back to 2007-8 which would give me a 5 year sample. A lot of work but could be worth it.