Budgeting from historical data is one of the best ways to price future work. There are many of caveats of course, something might be more because there is no scrap, or a job may come in under because it used scrap, but theoretically it should even out for the company - but this may not track out in jobs. Also, how much use you can generate depends on your ability to track. Where I work now can tell you how a project did - but it is much harder to track how one element within that job did. At that point it is tracking done by hand based on time sheets and the way you broke out the budget. Budget breakouts are important in terms of reconciliation, and tracking history so that apples can be compared to apples.
At any rate, I have have been a PM for 8.5 years with my company and during that time I now have back 184 completed / reconciled files. Approximately 22 projects a year is an interesting statistic too many - on the higher side for theatre (but certainly not off the chart). But the flip side is that completed projects only, I think probably accounts for about 50% of my time - it doesn't account for any of the smaller jobs that go through as primarily rental jobs, and even more time consuming - it doesn't count the time I spend bidding projects for which we are not awarded the work in the end.
I used to be better about tracking jobs I sent in a proposal on in which we did not get the work - thus creating a "hit rate" of jobs won versus jobs lost. But I have not really kept up with that - and it is a little loose as some are very low probably jobs in the first place, and we provide budgets that are more generalized. If we chose to be more targeted on what we provided proposals for we would have a better "hit rate", but we could also lose potential customers.
I also track the cost difference between what we sold the project for, versus our actual cost and review those over time. Tracking it over times helps me to identify trends to see what the causes may be if the proposed totals make more or less money than expected over a specific period of time. Interestingly enough, one variable to consider, in my case, is that when the shop is busier jobs take less time than when the shop is slow. And it doesn't mean that when we are busy quality suffers (though it can happen). Rather time and projects have a way of expanding and contracting to fill the available time. If you have nothing else to do, a task may take a long time - where as the same task n a busy day could be done twice as fast because you have a list of other things that need to get done. (this is referred to as Parkinson's Law; specifically that 'work expands so as to fill the time available for its completion). (this concept is actually one interesting for further date as it actually works within the triangle of fast, good, cheap).