F*ck historical comps

When a PE, just for fun, says "give me the sales, EBITDA, EBIT, P/E multiples for 50 comparable companies over the past 10 years on a monthly basis". Of course they don't want multiples based on actual financials but broker forecasts which were published in the past. Objective being of course to spot the trends beyond 1 industry cycle.

We have databases for these figures but it's not just about downloading them. Ideally you want to readjust the crap output you get based on: - How long the results were published after they actually occured (ie sales for year ending 31 Dec will be published 3/4 months later, and the brokers publish their revised forecasts a few days after that) - Random significant mistakes the database makes (eg. multiplying buy 10 some EBITDA figures for no reason or forgetting data all together) - Exceptionals (eg a random one-off charge to a company) - The accounting standards (ie in Europe may not be able to compare numbers published in 2000 with those from 2005 if the company switched from UK GAAP to IFRS accounting standards) - When the company either made a large disposed or acquisition since brokers don't all reflect these changes at the same time in their publications

This probably doesn't mean anything if you haven't done it before.

Basically you end up with sometimes well over 50Mb of data (in several spreadsheets because excel can't cope with all of that at once) which you've readjusted in all sorts of ways based on some kind of dodgy logic. Fascinating stuff.

The output is a few graphs which don't mean much because the source data is incomplete (hey, not all comps were quoted and well covered by brokers over the last 10yrs) and inaccurate, and may not be used anyway.

I think this confused and longish response gives you a vague idea of how frustrating this shit is, and that I happen to have been through the process...

Bottom line answers a question that has been floating around on this forum: don't be the bitch (IB), be the master (PE)