A Hill To Die On
There is a chart that I believe represents the ultimate form of chart crime, one that fund managers, allocators, advisers, and consultants all use to look clever.
Apollo just released their latest version. Apollo nicely insinuating that you should sell public equities for private markets. And boy do they have something to sell!
Source: Apollo Academy
So, by this chart, emphatically forward returns for the next decade will be 0%.
Maybe? Maybe not? I do not know and neither do I care (okay maybe I do), but what I don't like is using it to make overarching predictions.
When charting a relationship, you hope your data points are independent as precision and robustness relies on independent observations. If one observation shares 9 out of 10 values with the next observation, how independent is it really?
Serial correlation, autocorrelation, call it what you want! The R-squared is so high because almost every data point overlaps with its neighbours. The January 2010 data point shares 9 years of returns with the February 2010 data point, which shares 9 years with March 2010, and so on.
If you flip it the other way, what happens when you reduce the overlapping window?
Source: J.P. Morgan Guide to Markets
At 5 years, R-squared drops to 32%. At 1 year, R-squared collapses to 6%. Thank you, JPMorgan, for showing us what everyone else obscures. Forward looking 1yr +20% to -20%...
Okay some more charts from me, this is charting Shiller CAPE from 1980 onwards using monthly data:
Source: Ausyield
And what if we swap out CAPE for trailing PE ratios for the S&P500? However in this one I show you what non-overlapping, independent observations look like:
Source: Ausyield
How many non-overlapping periods since 1880? 14 (ish?) Fourteen! That's astrology! That's not even enough data to draw lines on a chart!
Let's go one step further. We know what returns projected by CAPE were, and know the actual realised returns, so why not compare the projected vs realised returns?
Aptus Capital Advisers have kindly done this for us and realised returns are usually 5-10% above the projected returns based on CAPE data.
PE ratios & CAPE are challenged when comparing data across long periods of time. There has been significant changes to how goodwill, D&A, capex, stock-based comp and a few other bits & bobs are calculated. The PE ratio in 1980 and today in my view are not directly comparable.
I'm not saying forward returns aren't going to be zero, but it's worth taking a beat and acknowledging the crimes of this chart and that it is being advocated for by a private markets manager.
One final observation, if Apollo genuinely believes public market returns will be zero for the next decade, shouldn't they be trying to monetise their portfolio today? If forward returns are truly zero, why aren't PE firms dumping everything onto public markets while multiples are still elevated? (1,2)
The chart says one thing. The behaviour says another.
Footnotes
For what its worth, I'm more comfortable saying there is a bubble in private markets, not public. OpenAI val is dumb. OpenAI is a private company and frankly if there is one place I can say the words a "bubble" exists in, it is private markets, not public.
Some time ago I was being pestered by "mid-market" PE managers to allocate. Fascinating stuff, why would you allocate to one when every PE fund was looking to buy up listed companies. If PE was acquiring listed companies and the IPO market was dead, it meant that PE thought listed markets were cheap. Look at the ASX, an inordinate number of companies have gone private since 2022. Lends into another idea that private equity and private credit are not "alternatives". They are equity and credits and they should be competing for capital from those sleeves. Where you can allocate to either private and public managers across all sectors.