[idea] Can we make money by exploiting the gap between the bottom of the large caps and the top of the small caps?

https://quantpedia.com/when-big-gets-small-trading-the-lower-tier-of-large-caps-and-upper-mid-caps/

When Big Gets Small: Trading the Lower Tier of Large Caps and Upper Mid Caps

One-line summary

“Can you make money shorting the smallest names in the S&P 500 — the ones that look about to drop out?” → No. In fact, doing the exact opposite produced a small alpha.

The starting hypothesis

These days, passive investing (index funds and ETFs) makes up a huge share of the market
Indexes mechanically push money around based on whether a stock is “in or out,” with no regard for its fundamentals
So the bottom-tier names in the S&P 500 (the smallest by market cap) sit at risk of being kicked out soon → once they’re dropped, passive funds are forced to dump them → the price falls
Hypothesis: short these bottom-tier names

The experiments (each step failed in turn)

Attempt What was tried Result

Round 1 Short the bottom 20 (or 5, or 50) of the S&P 500 ❌ No signal

Round 2 Short the bottom of the S&P + long the top mid-caps outside the S&P (long/short) ❌ Still no return

Round 3 (the twist) Flip it: long the bottom of the S&P + short the top mid-caps outside the S&P ✅ A small but steady effect

The key finding (a flipped conclusion)

Simply “being in the index” is itself a premium — a “membership premium”
A small name inside the S&P 500 > a bigger name outside the S&P 500 — index membership beats fundamentals
Why: index funds keep pumping money into their constituents (a persistent stream of demand), so prices get supported beyond what fundamentals alone would justify
In other words, the real alpha isn’t in the “deletion/addition events” — it’s in the ongoing demand imbalance that membership itself creates

The performance (looking at it soberly)

Best configuration (long 20 / short 20): 11.23% annual return

But the Sharpe is only 0.3–0.4 → not impressive on a risk-adjusted basis

The author himself candidly admits, “this involved parameter tuning, so I’m suspicious of overfitting.”

From this project’s perspective

The ratio is just too weak to use as a standalone strategy (CAGR 11% / Sharpe 0.4). It’s “an interesting phenomenon I noticed,” not “a deployable strategy.” Good material for the newsletter, but not a candidate to turn into a bot.