FWIW, I've looked at this.

One thing to watch out for: often some bullish observers comment that the margin debt as a percentage of market cap isn't so bad, so buy buy buy! But that suffers from the problem that market cap itself is elevated during bubbles. Divide one cyclically inflated figure by another, and everything seems fine.

So I looked at the ratio of margin debt to S&P 500 revenues instead, since revenues are much more stable as a yardstick of the size of those companies.

Using margin debt to GDP as your post mentions is probably pretty good too. It doesn't have the market cap problem either, though GDP can be a bit cyclical too.

No great conclusions for market timing, though...it shows elevated risk periods for sure, but those can last a very long time. Calling tops is much harder than calling bottoms, for me at least.

Handy link if you want to play with it. finra.org - Margin statistics

I have no idea if this is correct, but Gemini reports these margin to GDP ratios:
Dot-com bubble peak (March 2000): ~2.6% to 3.0% of GDP
Pre-GFC peak (2007): ~2.3% to 2.5% of GDP
2021 market top: ~3.8% of GDP
Recent figures: ~4.5% to 4.6% of GDP, all time high

Rather than market timing, the best use might be this: if things turn downwards, and those people with too much margin are selling because they have to, what will they sell first and most? That may or may not be what's most overvalued, but might be the thing that is best to own put options against : )
Once the selling is well and truly underway, another problem arises: people in trouble often sell what they can, not necessarily what they want to. Liquidity matters.

Jim